Summary: The large increase in inflation in 2021 and 2022 in the United States exposed just how little deep thinking had been done about the issue of inflation-control by macroeconomists and policy makers in preceding decades. The inflation of that time has often been attributed entirely to an excess of aggregate demand over potential output. But these years saw historically large shocks to the real economy stemming from COVID-19 and the Russian invasion of Ukraine. These shocks imposed extreme distortions on sectoral demand and supply, distortions which seem to have generated inflation globally, not just in the U.S. Further, temporary policies and circumstances (particularly pandemic fiscal relief and the whipsaw of massive layoffs and rapid rehiring efforts in labor-intensive service sectors) gave U.S. workers a pronounced but temporary boost in wage-bargaining with employers. Accordingly, a “shocks and ripples” analysis of inflation explains the data better than analyses based on movements in aggregate demand and supply.1

Starting in mid-2021, inflation in the United States rose to levels not seen since the early 1980s. This inflation followed on the heels of the economic shock imposed by the global COVID-19 pandemic and the significant fiscal policy interventions meant to smooth the fallout of this shock. As of October 2022, inflation—both headline and core measures—remained at historically high levels, though there are significant signs of softening in the near future (evidenced in part by the bending down of the quarterly data series shown in Figure A).

Figure A

This episode has sparked furious debate over the proper policy response, and it has exposed how little innovative thinking has been done on inflation by either macroeconomists or policy analysts since the 1980s price acceleration was ended by the Volcker shock. This report identifies a number of key questions raised by the inflationary outbreak of the past 18 months and offers some answers. A brief summary of these questions and answers is provided below. The remainder of the report then expands on these points.

Macroeconomic overheating is not necessarily the culprit for the inflationary surge of 2021 and 2022

In early 2021, debate raged about the potential economic effects of the American Rescue Plan (ARP). ARP, passed in early 2021, was explicitly designed as fiscal stimulus, with large and front-loaded transfers to households as its centerpiece, along with substantial aid to state and local governments.

Some critics of ARP worried about its potential effect on inflation. The most famous of these worriers was Larry Summers. Summers explicitly framed his concerns as centered around estimates of potential output. He posited that excess fiscal stimulus would push gross domestic product (GDP) well over the economy’s long-run potential to deliver, hence causing inflation. As he put it:

I agree with the general consensus of progressive economists that it would have been much better if the Obama administration had been able to legislate a much larger fiscal stimulus in early 2009, in response to the Great Recession. Yet a comparison of the 2009 stimulus and what is now being proposed is instructive. In 2009, the gap between actual and estimated potential output was about $80 billion a month and increasing. The 2009 stimulus measures provided an incremental $30 billion to $40 billion a month during 2009—an amount equal to about half the output shortfall.

In contrast, recent Congressional Budget Office estimates suggest that with the already enacted $900 billion package—but without any new stimulus—the gap between actual and potential output will decline from about $50 billion a month at the beginning of the year to $20 billion a month at its end. The proposed stimulus will total in the neighborhood of $150 billion a month, even before consideration of any follow-on measures. That is at least three times the size of the output shortfall. (Summers 2021)

This argument might benefit from an illustrative figure. The green line in Figure B shows the estimates of potential output referenced by Summers (“GDP in overheating scenario”). The blue line shows the Congressional Budget Office’s (CBO’s) predictions of what GDP growth would have been without ARP through the end of 2020, and actual GDP growth since that date. We then add in a line showing the path GDP would have taken had ARP pushed up actual GDP 1-for-1 with spending, leading real GDP to exceed potential in the manner described by Summers. In this figure, one can see the still considerable negative output gap (shortfall of actual GDP relative to potential) that persisted at the end of 2020, as well as the very large positive output gap that was projected by reasoning like Summers’s after ARP’s passage by the end of 2022.

Figure B

The emergence of higher levels of inflation by mid-2021 led many to assume this output gap-based reasoning had turned out to be true. They thought that the inflation was clearly the result of macroeconomic overheating (with the level of actual GDP far exceeding the level of potential GDP). But it is far from obvious that this is the correct interpretation. For one (as we show later), even with the American Rescue Plan, real GDP growth (the red line) has barely beaten pre-pandemic projections of what it would be by mid-2022.

Below we highlight evidence that further complicates the narrative that inflation is the result of simple macroeconomic imbalances driven by a too generous ARP.

International evidence complicates the domestic overheating story

The most straightforward reason to doubt this narrative comes from a look at the international experience of inflation.

A look across member countries of the Organisation for Economic Co-operation and Development (OECD) shows that rising inflation was not unique to the U.S. and was in fact a global phenomenon throughout 2021 and 2022. Figure C shows the acceleration in core inflation from May 2021 through September 2022, compared with two years of pre-pandemic “normal” inflation (2018–2019), for 35 OECD countries. We use core inflation, which strips out food and energy prices, to better represent broad inflationary pressures in each economy. Using core inflation also allows for a better comparison between the U.S. and Europe given the volatility in food and energy prices affecting Europe due to the war in Ukraine. 

Figure C

As Figure C shows, all 35 OECD nations we examined experienced an acceleration in core inflation throughout 2021 and 2022 compared with the pre-pandemic period. While above the median, and on the higher side of inflation experiences worldwide, the U.S. is by no means an outlier and is just below the average for all other OECD countries. This global phenomenon of rising inflation casts doubt on the claim that U.S. inflation was caused purely by domestic policy decisions leading to macroeconomic overheating.

One might argue that the global acceleration in inflation simply meant that many countries overheated their economies and generated excess demand through too much fiscal spending. However, the data do not support this argument. For one, Figure C shows OECD nations with a wide range of fiscal responses, from aggressive relief spending to little intervention. Despite the varying responses, all countries experienced some level of inflation acceleration.

Figure D examines more closely the argument that global inflation is simply a reflection of global excess demand. To do this, we examine core inflation acceleration on the vertical axis (the same numbers shown in Figure C). On the horizontal axis, we show change in unemployment between September 2022 and the pre-pandemic 2018–2019 unemployment. This measure indicates how much unemployment has improved recently compared with the pre-pandemic period (for example, a fall in the unemployment rate of 2 percentage points would be shown on the graph as a positive 2%). If inflation was caused by excess demand growth (proxied by lower unemployment rates today), one would expect to see a positive relationship between unemployment improvement and acceleration of inflation. The data do not show this.

Figure D

As Figure D depicts, there is no significant positive relationship between unemployment improvement and inflation acceleration. If anything, there appears to be a slightly weak relationship in the opposite direction whereby countries with higher unemployment (or lower improvement) relative to pre-pandemic times experienced higher inflation levels. The fact that countries with larger decreases in unemployment (perhaps brought about by more expansive fiscal policy and economic stimulus) do not show larger spikes in inflation strongly complicates the claim that macroeconomic overheating applies globally.

Overall, the shared 2021–2022 international experience of high core inflation strongly counters the argument that fiscal relief in the U.S.—such as the American Rescue Plan—either drove up inflation or contributed significantly to its unusual persistence.

Domestic evidence is also underwhelming for simple overheating explanations

Turning to the domestic U.S. evidence, the case for recent inflation being sparked by a simple macroeconomic imbalance of aggregate demand and potential output is also weak. Many have presented the steepening trend in nominal spending over the past year and a half as evidence for the overheating view. This is tautological. Faster nominal spending growth could simply be a reflection of faster inflation; it is not evidence of its cause.

Take a totally trivial example: Imagine there was a rapid consolidation of market concentration across the economy. Firms with greater market power would likely raise prices. If the price elasticity of demand was relatively low in the short run (which seems like a safe bet), this would in turn make nominal spending rise more rapidly (even while real spending would actually fall). This could happen with no implication at all for the state of macroeconomic balance.

More realistically, one could imagine a scenario—like what happened following the pandemic shock—wherein the allocation of demand across spending categories rotated sharply into sectors with either impaired supply or a higher elasticity of prices with respect to demand. As this happened, there would be an increase in prices even without the level of aggregate demand being particularly high relative to the economy’s potential output. In the long run, the inflationary effect of very large relative price changes set off by such a process could be muffled by macroeconomic policy, but claims that over a 1–2 year period such relative price changes cannot be major drivers of inflation seem obviously wrong.

Decomposition of inflation into ‘demand’ and ‘supply’ factors

One method some have used to assess the role of ARP and excess stimulus in generating inflation is to decompose the recent acceleration of inflation into “demand” versus “supply” factors. Probably the most well-done and transparent version of this exercise is by Shapiro (2022). The categorization of price changes in a given economic sector as being driven by demand or supply is done by estimating the price and quantity levels of an industry in each month. Then, the “unexpected” components of monthly changes (basically those that exceed or lag a running trend) in both prices and quantities are extracted. If a sector sees both price and quantity growth above trend, price increases in that sector are categorized as demand-driven. If price growth is above trend but quantity growth is below trend, then price increases are characterized as supply-driven. If either price or quantity growth is near trend, then the industry’s price growth is labeled ambiguous.

The Shapiro (2022) decomposition is certainly clever. Based on these results, the rise of core inflation over the past year can essentially be attributed equally to demand- and supply-side measures. This decomposition for recent years is reproduced in Figure E. However, this technique and how its results are interpreted have a couple of potential shortcomings.

Figure E

Shapiro’s technique for decomposing demand versus supply drivers of inflation might stumble on one potentially important issue—changes in the elasticity of price changes with respect to demand shocks. Take the example of an industry that has seen a very large price increase relative to trend but has seen steady growth in output. Under the Shapiro (2022) decomposition, this would qualify as the source of inflation in the sector being “ambiguous.” But this could easily be a supply issue. If during normal times a mild uptick in demand (a percentage point or two above trend) led to tame price growth, but since the pandemic this mild uptick was associated with very large price increases, this could well actually be a signal that it is supply-side factors that are binding. Further, even for sectors that are characterized as demand- or supply- driven, if the price change associated with any demand or supply mismatch (regardless of which side initially caused it) is greater than it was in the past, this could signal that sectoral frictions—not just macroeconomic factors—are causing the rise of inflation.

On the issue of the interpretation of the results, identifying a given inflationary episode as being driven by “demand” or “supply” can sometimes be akin to asking which blade of the scissors cuts the paper. As Larry Summers put it (fairly enough):

I think it restates what I think is a bit of a popular confusion in the following sense—supply is what it is. Monetary policy can’t change it. Fiscal policy can’t change it, except in the long-run. And so given what supply is, it’s the task of demand to balance supply. And if demand is greater than supply, then you’re going to have excess inflation and you’re going to have the problems of financial excess.

So the job of the demand managers, principally the Fed, is to judge what supply is and calibrate appropriately. It’s not an excuse for inflation to blame it on supply. It’s a reality in the environment that you have to deal with. And so the job is to look for measures of overheating, and when you see measures of overheating, to apply restraint. (Klein 2022)

Real-time estimates of actual and potential GDP don’t look particularly inflationary

Summers’s point that attributing the recent rise in inflation to “demand” or “supply” does not end the debate about the role of excess macroeconomic stimulus in driving today’s inflation is well taken. However, his claim that “supply is what it is” simplifies far too much. The most obvious disruption to potential output (or aggregate supply) in the wake of the COVID-19 shock was the 2.5% decline in labor force participation between February 2020 and the end of 2020. But should policymakers really have looked at this decline and just thought “it is what it is” and pulled back demand growth to match this? Or, instead, was the decline in labor force participation (which fell 3.5% in a single month in April 2020) better seen as a mostly temporary economic casualty of the pandemic that would eventually heal?

So, in some sense it is true that categorizing some inflationary shocks as “supply-driven” does not map perfectly onto a recommendation to keep demand policy stable. But the larger claim that inflation is ipso facto evidence of aggregate demand overshooting supply and hence requires contractionary macroeconomic policy does not follow.

We can get some sense of how much the aggregate levels of demand and supply have shifted relative to pre-pandemic trends using data on GDP and potential output. At the end of 2019, the Congressional Budget Office made projections of both of these variables for the coming years while forecasting little to no change in inflation (or interest rates). The Summers argument above is that either GDP began rising faster than forecast in 2019 (due to excessively expansionary fiscal policy) or that potential output shrank, with either influence (or both influences) leading to a positive “output gap” that drove up inflationary pressures.

Figure F shows real GDP and potential output, both as ratios to what CBO projected they would be before the pandemic. For the measure of potential output, we allow developments since the pandemic to affect the CBO projection. Specifically, we reduce the labor input into potential output by assuming that the decline in the labor force participation rate is driven solely by supply-side factors.2

We also account for changing capital services input and total factor productivity growth relative to CBO projections. For capital services, we construct a measure of growth of the aggregate capital stock that accounts for the nonresidential fixed investment (NRFI) that has occurred since the pandemic and we compare this against CBO projections of capital services input growth. For total factor productivity, we employ the utilization-adjusted measure of total factor productivity growth compiled by John Fernald (2023) and compare that with the CBO forecast.

As can be seen in Figure F, potential output fell sharply (not as sharply as GDP, but still noticeably) in the immediate post-pandemic shock period. As of the third quarter of 2022, it still remained a bit under 2% below what CBO forecast it would be in that quarter. GDP fell very sharply in the pandemic recession, but by the third quarter of 2022 sat roughly 1% beneath what CBO forecast it would be before the pandemic struck.

Figure F

There was a period of time during 2021 when GDP rose above our adjusted measures of potential output for a stretch. Over the five quarters from the end of 2020 to the end of 2021, the cumulative positive output gap (GDP exceeding potential output) was 5.8%, with an average gap of around 1.2% in each quarter.

That GDP exceeded potential output as inflation rose gives some plausibility to claims that macroeconomic overheating contributed to the recent inflationary spike, but the magnitude of the spike makes it highly unlikely that this overheating played a starring role. There is a well-established literature on how much each 1 percentage point positive output gap should be expected to drive up the inflation rate. These estimates do not exceed 0.5% and cluster more tightly around 0.3% or even lower. This implies that the 1.2% average output gap in that five-quarter stretch should be expected to raise subsequent inflation by roughly 0.4–0.6%, or by about a tenth of its actual acceleration over this period.

A historical example might help make this clearer. According to CBO estimates, the U.S. economy ran a cumulative positive output gap of over 17% of potential output, with an average gap of 1.2%, over the period from 1997 to 2000. This was the same average gap as that seen in 2021, but sustained for four times as long. Yet there was no inflationary increase at all during this period. In short, running the economy this “hot” for a year is not supposed to yield anywhere near the degree of inflation that we have witnessed since the middle of 2021.

Figure G shows the history of output gaps since 1995. For the last two years, we show the gap with an unadjusted measure of potential output from CBO’s last pre-pandemic projection, plus the gap with our adjusted measure of potential output. Even with our adjusted measure, which accounts for pandemic damage to the economy’s aggregate potential output, the positive output gaps of the past 18 months are utterly unremarkable relative to recent U.S. economic history—a history that saw no similar inflationary spike.

Figure G

Tight labor markets usually boost—not reduce—labor’s relative bargaining power

Finally, we highlight some evidence from the labor market to assess the claim that a straightforward story of macroeconomic overheating is at the core of recent inflation. Generally, claims that inflation accelerations are driven by an excess of aggregate demand over potential output rest on theories of labor market overheating. As aggregate demand exceeds potential output, unemployment falls. In turn, this boosts workers’ bargaining position with employers and accelerates wage growth. If nominal wage growth begins exceeding price inflation, this leads to a rise in labor’s share of income.

The general logic that lower rates of unemployment boost nominal wage growth more than price inflation is sound and supported by empirical evidence. As the recent inflationary episode began in 2021, it was often accompanied by stories of labor shortages in many sectors. This led far too many to assume that wage pressures were pushing up price growth, and the simple story of the labor market overheating due to a macroeconomic excess of aggregate demand over potential output gained credence.

The first bit of evidence against the claim that rolling labor shortages across sectors led to prices rising can be seen in Figure H. This graph shows the acceleration in price inflation and the acceleration in nominal wage growth across 61 industries. It measures acceleration of prices and wages as their annualized growth rate between the second quarter of 2020 and the third quarter of 2022 relative to the annualized growth rate that prevailed on average between 2018 and 2019. There is no discernible correlation at all between these measures.

Figure H

Moreover, while nominal wage growth did accelerate in 2021, it never exceeded price inflation. This means that real (inflation-adjusted) wages have been falling since early 2021. This also led to a pronounced fall in the labor share of income in the corporate sector, which has largely not recovered from its post-pandemic low. It seems odd that a labor shortage could somehow be the source of inflation given this data—it is rare for services in short supply to command less and less income growth on a per-unit basis.

This fall in real wages and the labor share of income is absolutely not the norm for the U.S. economy as it “heats up” in recoveries. This fact has been missed by far too many commenters. Many have made implicit claims that a sharp fall in the labor share of income and real wages is the norm for an economy with positive output gaps. Rampell (2022), for example, writes:

The greedflationists argue that something fishy is afoot because companies are not merely “passing along” their higher costs; their profit margins are expanding, too. But this is exactly what you’d expect when flush customers are buying more stuff and willing to pay whatever’s necessary to get what they want. Prices and profits rise.

Read “flush customers willing to pay whatever’s necessary to get what they want” as “high levels of aggregate demand relative to potential output.” Is it really true that historical experience would lead one to expect that high levels of aggregate demand lead to prices and profits rising?

Not really. Figure I shows the labor share of income in the corporate sector since 1949. The cyclical dynamics of the labor share are slightly complicated: The labor share is not “countercyclical” as it is sometimes described. It does rise sharply during outright recessions, as more volatile profits decline sharply during economic downturns. But in early recoveries with unemployment still high, the labor share universally falls sharply. Then, in mid-recovery as unemployment starts to approach (or fall beneath) pre-recession lows, the labor share begins to rise as unemployment falls—or, as the economy “heats up.”

Figure I also shows variability and potential decade-specific trends in labor’s share. This explains why a simple scatterplot of the relationship between the change in labor’s share of income and the unemployment gap is very noisy, with only a mild (if statistically significant) downward correlation, which indicates that low unemployment gaps (signifying tight labor markets) are weakly associated with an increased labor share.

After we control for decade-specific dummy variables and decade-specific trends, this relationship dramatically strengthens, as shown in Figure J. The figure shows the coefficient on the unemployment gap from a regression of the change in the labor share on the unemployment gap, plus decade-specific dummy variables, decade-specific trends, and productivity growth. It shows this regression for all periods in our data (quarterly data from 1949 to 2018), as well as periods when the unemployment gap is greater than 1, less than or equal to 1, greater than 0, and less than or equal to 0. An unemployment gap of 0 or below indicates a tight labor market with actual unemployment either equal to or less than estimates of the natural rate. An unemployment gap of 1 or below indicates an economy operating below full employment but within shouting distance of it. An unemployment gap of above 1 indicates an unhealthy labor market.

What does this tell us? That it is extremely unusual for labor’s share of income to fall (or even stagnate) even as unemployment falls beneath 5%: Higher profits are not the expected signature of an overheating economy. In this sense, the recent low levels of labor’s share and the poor performance of real wages are signs that the current economy does not look anything like a typically overheating economy.

Figure I

Figure J

If not macroeconomic imbalances, then what? Sectoral shocks and their ripples

If the driver of recent inflation was not large macroeconomic imbalances, then what was it? Put simply, extraordinarily sharp sectoral shocks and the large ripples these shocks generated drove recent inflation. Tobin (1972) provides probably the best description of how large sectoral shocks can cause persistent inflation. Key to his reasoning is the empirical finding that nominal wages are extremely rigid downward. Given this downward nominal wage rigidity, adjusting to sectoral shocks to demand and supply will always require inflation (rising nominal wages in expanding sectors) rather than deflation or neutral aggregate wage and price growth (i.e., rising or flat nominal wages in expanding sectors matched by falling nominal wages in contracting sectors). These insights are profound enough to quote at length:

The overlap of vacancies and unemployment—say, the sum of the two for any given difference between them—is a measure of the heterogeneity or dispersion of individual markets. The amount of dispersion depends directly on the size of those shocks of demand and technology that keep markets in perpetual disequilibrium, and inversely on the responsive mobility of labor. The one increases, the other diminishes the frictional component of unemployment, that is, the number of unfilled vacancies coexisting with any given unemployment rate. A central assumption of the theory is that the functions relating wage change to excess demand or supply are non-linear, specifically that unemployment retards money wages less than vacancies accelerate them. Non-linearity in the response of wages to excess demand has several important implications.

First, it helps to explain the characteristic observed curvature of the Phillips curve. Each successive increment of unemployment has less effect in reducing the rate of inflation. Linear wage response, on the other hand, would mean a linear Phillips relation. Second, given the overall state of aggregate demand, economy-wide vacancies less unemployment, wage inflation will be greater the larger the variance among markets in excess demand and supply. As a number of recent empirical studies have confirmed (see George Perry and Charles Schultze), dispersion is inflationary. Of course, the rate of wage inflation will depend not only on the overall dispersion of excess demands and supplies across markets but also on the particular markets where the excess supplies and demands happen to fall. An unlucky random drawing might put the excess demands in highly responsive markets and the excess supplies in especially unresponsive ones. Third, the nonlinearity is an explanation of inflationary bias, in the following sense. Even when aggregate vacancies are at most equal to unemployment, the average disequilibrium component will be positive. Full employment in the sense of equality of vacancies and unemployment is not compatible with price stability. Zero inflation requires unemployment in excess of vacancies. (p. 10)

If Tobin is right that “dispersion [of sectoral shocks] is inflationary,” then the mammoth response of inflation to the COVID-19 shock becomes very easy to understand—this pandemic effect was the mother of all shocks to sectoral dispersion. Further, specific features of the 2021 economy meant that any shock to sectoral imbalances would have led to large ripple effects, mostly through shocks’ effects on the labor market, which saw nominal wages respond to nonlabor cost shocks and support inflation to an unexpected degree.

These “ripple” effects stem in part from the distributional conflict resulting from inflationary shocks as various economic groups try to protect their real incomes. As Ros (1989) puts it: “A common form of [conflict inflation] arises when the real wage reflecting the balance of power in the labour market, and expressing the expectations created in wage bargains, is not validated by the real wage implied by price formation in other markets” (p. 8). So, if a shock to the cost of nonlabor inputs (say lumber used in home building and chips used in automobile production) pushes up prices, workers might respond by bargaining for higher nominal wages to protect their living standards. In turn, firms may accommodate their own workers’ nominal wage demands (or at least some of them) yet maintain or even expand profit margins to protect their own incomes.

This conflicting-claims view of U.S. inflation is not well known or often wrestled with in most macroeconomic commentary. There’s one pretty good reason for this—for decades, it has largely not been an issue, as a number of policy changes have so disempowered U.S. workers that their efforts to protect real incomes from any shocks have been limited enough to leave almost no mark on inflationary dynamics. Ratner and Sim (2022) provide compelling evidence that the extremely low inflation that characterized the 30 years before COVID-19 is likely largely explained by a pronounced shift in bargaining power from workers to firms. Yet in 2021, these conflicting claims on real output following large exogenous shocks led to the large and persistent ripple effects in inflation.

What are the analytical and policy stakes in distinguishing between inflation driven by macroeconomic overheating (imbalances in the level of aggregate demand and potential output) versus a “shocks and ripples” theory? Even if they are large, as long as the ripple effects following inflationary shocks dampen rather than amplify the initial inflationary shock, then macroeconomic policymakers should not have to pursue aggressively contractionary policies to rein inflation back in. This is not simply tautological—sometimes shocks really do set off ripple effects that amplify the initial impulse and need some external force (looser labor markets in the current context) to provide dampening. But so long as wage growth lags behind price inflation, the ripple effects—large as they might be—will steadily dampen the initial shocks and return inflation to more normal levels over time, even absent any effort to engineer looser labor markets.

Below we more sharply distinguish just what the economic shocks caused by COVID-19 and the Russian invasion of Ukraine were. We also outline how the ripple effects kept inflation more persistent than what many forecast going into this episode, though the effects still look set to fade as long as the shocks stop coming.

What were the shocks?

The main shocks to the U.S. economy from the pandemic and war were the economic distortions that they created in both demand and supply patterns. On demand, the composition of GDP shifted with a historically rapid reallocation in spending and demand away from services and government and into durable goods consumption and residential investment. On the supply side, the pandemic and war contributed to massive supply chain snarls, further heightened by port shutdowns and the global spike in raw material, energy, and commodities prices.

Demand shocks: consumption patterns and the underappreciated role of housing

The shift in demand patterns away from face-to-face, high-contact services (such as gyms, movie theaters, travel) and toward durable goods and residences (cars and houses) was clearly a consequence of the pandemic, and it has shown remarkable persistence. Figure K displays the shock to the composition of demand in historical context from 1980 through the present. We examine the share of GDP made up of durables and residential investment and demonstrate how it has changed relative to the average of the previous two years. Clearly, the onset of the pandemic led to a historically unprecedented jump in the share of durable goods consumption and residential investment (the last rise, though at a much slower rate, can be seen in the early 2000s). In recent years, the share of durables consumption and residential investment has moved a bit closer to normal, but it remains at a high level relative to historical averages. (In Figure K, one can see that the level of demand as of 2022 Q3 was roughly in line with what it has been for the past two years—i.e., the line hovers near zero—and these past two years have been dominated by the COVID-19 patterns of spending.) This historically sharp swing in demand across sectors is certainly a large enough shock to explain the beginning of the recent inflationary episode.

The swing toward durable goods consumption and away from face-to-face services is intuitive to understand (the classic example being the substitution of Peloton purchases for gym memberships). However, the boost to housing demand driven by the pandemic is even better documented by the data. Apparently, the prevalence of remote work led to a large positive shock in housing demand as more people worked from home, first out of necessity of social distancing for public health, but then (for many) out of choice. Working from home in turn inspired demand for more space and smaller households, leading to a large surge in new purchases and household formation that ran far ahead of population growth for 2021.

Figure K

This pandemic shock to housing demand had profound implications for subsequent inflation. Housing is a key component of inflation, making up 40% of core consumption spending in the CPI. Housing prices (including rents) have also increased dramatically since 2019. Figure L shows the tight relationship between remote work and the growth in home prices, as shown in Mondragon and Wieland (2022).

Figure L (taken directly from Mondragon and Wieland 2022) shows a strong positive relationship between home price growth and exposure to remote work, meaning that the areas most exposed to remote work had home price growth twice as high as the areas least exposed. Their model further estimates that remote work raised aggregate home prices by 15.1%, accounting for well over half of the rise in housing prices over that time. Clearly, the pandemic shock to housing demand and subsequent price growth is a crucial component of the 2021 inflation story. 

Though housing prices have been high through the pandemic, they seem to have been assigned less blame in the recent inflation episode compared with the overheating or fiscal over-stimulus arguments. Why has housing been such an underrated contributor to high inflation in economic policymaking discussions? Mostly because official measures of housing costs were one of the last components of inflation to noticeably accelerate. The measurement of housing prices is one of the most backward-looking price indicators, with increases in new rents and home prices in many industry data sources only visibly pushing up costs in the CPI 6–12 months later.

Figure L

Given this lag and the backward-looking nature of housing measurement, a shock to housing demand generally does not manifest in an increase in housing prices and rents until the following year. This means that many policymakers and economic commentators were unable to track the extent of price changes as they occurred. Figure M shows the correlation between annual growth in the Case-Shiller home price index, lagged one year, and annual growth in the shelter component of the CPI since 1989.

This lag between home price changes and when they are reflected in falling shelter in the CPI meant that the 2021 positive shock to housing demand stemming from the pandemic only pushed up official measures of inflation later in 2022. However, it is also important to note that the reverse dynamic is likely to characterize rental prices going forward—substantial weakness in early-warning measures of rental prices will show up only in a slower rate of CPI growth with a significant lag.

Figure M

Supply shocks: Supply chains and much spikier effects on labor supply than appreciated

While the pandemic shocks to the demand side are evident, the pandemic created important supply shocks as well.

The most well-known shocks were pandemic-driven snarls in global supply chains of durable goods and materials for construction. These supply-chain snarls were largely due to rolling port shutdowns throughout East Asia in key manufacturing hubs. The Federal Reserve Bank of New York maintains an index of global supply chain pressure. Figure N shows that this index hit its highest points on record in 2021, and only by late 2022 had the index begun showing real signs of normalizing. The pandemic supply-chain shock was quite persistent.

Figure N

Another underappreciated part of the pandemic’s effect on the U.S. economy’s supply side was its effect in temporarily sidelining millions of employed workers each month. This effect became historically pronounced during the omicron wave of January 2022. Figure O shows the number of people who were employed with a job but were not at work due to illness or medical problems in the reference week of the Current Population Survey (the survey used to calculate the unemployment rate and other key labor market indicators). While there are spikes in this series in 2020 and when the delta variant was spreading in summer 2021, the number skyrockets to over 3.5 million people in January 2022 during the omicron wave. This rolling shock in labor supply very likely disrupted the labor market and economic system as well but shows some hopeful signs of normalizing in recent months.

Figure O

Large—but settling—labor market ‘ripples’

Perhaps one of the most notable elements of the 2021 labor market was the growth in nominal wages. Nominal wage growth in 2021 was extraordinarily fast relative to recent history and is even turning out to have been fast relative to what has prevailed in 2022—even as other measures of the labor market have seemingly tightened. Many policymakers have claimed that increased nominal wage growth has been a key driver of inflation since early 2021. This claim is not totally implausible—historical episodes of price-wage spirals really have occurred and required some exogenous forcing mechanism to bring down wage growth as part of the anti-inflationary strategy. However, a close look at the evidence indicates that the focus on wage growth as a key driver of inflation in the past 18 months seems misplaced. Further, it seems quite likely that the abnormally fast wage growth of the past 18 months can be normalized without a significant forcing mechanism (like substantially higher unemployment rates engineered by Fed interest rate hikes). Indeed, wage growth already seems to be normalizing pretty quickly.

In short, the rapid nominal wage growth of 2021 should not be understood as a major cause of the inflation of 2021 and should not be expected to continue (even if the unemployment rate remains low) going forward. To support these claims, we highlight a number of features of the 2021 labor market that allowed for this nominal wage growth in the first place and argue that they are largely unique to that year.

Put simply, workers had high degrees of bargaining power in 2021 relative to what the overall unemployment rate might have indicated. Well before the unemployment rate approached its pre-pandemic levels, employers were pushed to raise wages in order to attract and retain workers. Most notably, this wage growth occurred in industries in which workers often have the least bargaining power and face the lowest pay: retail, services, food, and accommodations.

There were likely two major changes to labor markets in 2021 that provided temporary boosts to workers’ bargaining power. First, the massive level of layoffs and business closures that accompanied the pandemic meant that labor market frictions that gave employers a degree of monopsony power over their workforce were dissolved in one fell swoop. These frictions are highly powerful in preventing workers from even obtaining information about jobs with higher wages in their immediate area (Jager et al. 2022). By the end of 2020, tens of millions of employee-employer ties had been severed by the pandemic, but at the beginning of 2021, the extremely large fiscal relief convinced employers to staff up quickly. This rapid staffing-up happened in the context of workers facing far fewer frictions tying to their current employer (and muting upward wage pressure) than is the norm. As more and more new employee-employer matches were cemented as 2021 turned into 2022, the same forces that introduce frictions into workers’ job searches and competitive searching seem highly likely to reassert themselves.

The second major component of workers’ empowerment in 2021 was the role of pandemic aid in providing a wealth buffer (we present evidence on the size of this buffer in Figure V). This buffer bought workers time to find employment that suited them while still covering their costs, rather than being forced back into the first available job regardless of its fit for them. Chetty (2008), for example, has identified the powerful role that having some liquid wealth buffer has in allowing workers to be choosier in their job searches.

Economic impact payments (EIPs, often called stimulus checks), expanded unemployment insurance, and the monthly Child Tax Credit gave workers the ability to build up savings and accumulate a level of financial security that had been largely unavailable for tens of millions of workers any time before the pandemic. This translated into significant bargaining power in the labor market. However, while this support was unprecedented, it was also short lived, and both employers and workers knew with a high degree of certainty when this aid would turn off. The last stimulus check was mailed in January 2021. Enhanced unemployment insurance and the CTC phased out in fall and winter 2021, respectively. This wealth buffer for all made job searches and wage offers in 2021 far different than they were during normal times.

One could imagine how policy efforts to restrain employers’ monopsony power and to give workers a better fallback position in the face of job loss could have permanent effects. If, for example, a major change to labor law allowed workers to unionize even in the face of today’s hostile employer class, then this could easily provide a permanent source of countervailing power to monopsony (see, for example, Benmelech, Bergman, and Kim 2021). Aspects of the pandemic relief (particularly the enhanced child tax credit and an increase in the protectiveness of the UI system) could have also been made permanent. But the simple fact is that none of the underlying boosts to workers’ bargaining power that characterized the 2021 labor market continue to exist today. This fact strongly suggests that any unexpected labor market worker power experienced in 2021 is likely to be temporary rather than permanent.

Why were the large wage ripples such a surprise—and why are we sure they’ll settle?

We noted before that the primary policy-relevant distinction between a view that sees recent inflation as the result of macroeconomic overheating and a view that sees it as a series of shocks and ripples concerns the role of demand management. If inflation is the result of aggregate demand exceeding potential output (and if one imagines potential output is fixed—“it is what it is”), then the only remedy is to slow demand growth, even if that leads to higher unemployment. If, instead, inflation has been driven by shocks and ripples, and if the ripples eventually settle, then inflation can normalize without engineering higher unemployment.

We also noted that wage growth in 2021 and early 2022 was quite rapid in historical terms, and the ability of U.S. workers to shield their real incomes from inflationary shocks was unexpectedly robust. This raises a couple of questions: (1) If the ripple effects of higher wage growth following inflationary shocks were so large, why can we be sure that they will eventually dampen on their own?; and (2) Was the inflation of the past 18 months driven by wage growth or not?

On the first question, the simplest answer is that for decades American workers’ wages have responded only weakly to price shocks in the short run. Figures P1 and P2 highlight two separate time periods—1949–1988 and 1989–2019. In each period, the growth of wages and growth in prices lagged just two quarters is shown. In the earlier period, shown in Figure P1, wage growth was tightly linked to price inflation even in the short run. In the latter period, shown in Figure P2, there is essentially no durable relationship at all. In sum, recent decades seemed to break any quick link between price spikes and immediate changes in wages. It’s certainly possible that the pattern that held between 1989 and 2019 was somehow completely overturned in the post-pandemic period, and we are headed back to an era in which wages will respond quickly to price shocks. But there needs to be a long period of compelling evidence on this before we should assume this tight link has been reestablished. If instead the nonrelationship that has prevailed for the last 30 years is the better predictor of future wage-price dynamics—particularly once the temporary sources of bargaining power we highlighted previously are behind us—then it seems a safe bet that the wage ripples from recent price shocks will settle soon.

Figure P1

Figure P2

Further, as long as nominal wage growth adjusts only partially to price shocks and lags at all behind inflation, then wages are providing a dampening effect on inflation. This has clearly been the case in the recent period. Since May 2021, for example, CPI inflation has risen at an average annualized rate of 6.8%, while average hourly earnings have risen at a rate of 5.0%.

Even more compelling, the ripple effect of faster wage growth clearly seems to be abating now that large shocks have stopped coming (and temporary labor market supports have ended). This is true even as quantity side measures of the labor market (like the unemployment rate) remain quite strong. Figure Q shows the growth of average hourly earnings and unemployment over the past two years (note that we suppress the very large wage jump accompanying the pandemic-driven layoffs of mostly low-wage workers in mid-2020). Besides showing a pronounced nonrelationship between unemployment and wage growth in recent times (casting some doubt on a simple story of labor market overheating), this graph also shows a pretty clear recent deceleration of wage growth.

Figure Q

On the second question (“was the inflation of the past 18 months driven by wage growth or not”), the answer is nearly as simple: largely not. It is true that if nominal wage growth had not budged from the 3% pace that persisted pre-pandemic then inflation would have been slower over the past 18 months. But it still would have been a historically large inflationary spike.

Further, given that most of the price pressure started from outside labor markets and would have happened anyway, the ability of nominal wage growth to accelerate over this period really did protect workers’ real incomes. If all policymakers cared about was keeping inflation as close to the Federal Reserve’s 2% inflation target as possible, then the nominal wage growth acceleration of the past 18 months was a problem. If instead one also cared about protecting the living standards of U.S. workers in the context of nonexplosive inflation, this wage growth was clearly beneficial.

Figures R1 and R2 provide some rough simulations showing the inflationary effect of various paces of nominal wage growth. They use real data on wage growth and then infer what portion of overall inflation was driven by other factors over the past 18 months. They then subtract out the influence of the faster wage growth seen over the pandemic recovery while allowing these other factors’ contribution to inflation to persist. Figure R1 compares the resulting evolution of actual inflation against the counterfactual in which nominal wage growth does not accelerate past its pre-pandemic pace. Flat wage growth would have indeed lowered inflation, but a historically notable spike still would have occurred. Finally, Figure R2 highlights how much lower inflation-adjusted wages would be today had nominal wage growth not accelerated but other inflationary forces were felt over the past 18 months. Even with the higher inflation rates prevailing in the model in which nominal wages partially adjust to price shocks, real (inflation-adjusted) wages fall less in the scenario with partial wage adjustment relative to the one in which wage growth remains flat in the face of price shocks.

Figure R1

Figure R2

The role of mark-ups

The price of just about everything in the U.S. economy can be broken down into the three main components of cost. These are labor costs, nonlabor inputs, and the “mark-up” of profits over the first two components. Good data on these separate cost components exist for the nonfinancial corporate (NFC) sector, which accounts for roughly 60% of the entire private sector (and likely has strong effects on price setting even throughout the noncorporate sector).

Since the trough of the COVID-19 recession in the second quarter of 2020, overall prices in the NFC sector have risen at an annualized rate of 7.3%—a pronounced acceleration over the 1.8% price growth that characterized the pre-pandemic business cycle of 2007–2019. As Figure S shows, 40.8% of the increase in the former period (since the recession’s trough in 2020 Q2) can be attributed to fatter profit margins, with labor costs contributing less than a quarter of this increase. Some have argued that starting this measurement from 2020 Q2 could represent cherry-picking that overstates this effect. Measuring from the previous business cycle peak of 2019 Q4 still sees fatter profit margins accounting for a third of the rise in prices in the current business cycle. This is a very high share. From 1979 to 2019, profits contributed only about 11% to price growth (and—not shown in this figure—labor costs contributed over 60%). Through the end of 2021—the period of greatest price acceleration—profits contributed well over half of the entire increase in prices.

Figure S

Do fatter profit margins imply more corporate power—or just power channeled differently?

The rise in profit margins that accounts for a disproportionate share of price growth in the current recovery has led to speculation that increased corporate power has been a key driver of recent inflation. Corporate power is clearly playing a role, but an increase in corporate power likely has not happened recently enough to make it a root cause of the inflation of 2021–2022. In fact, the rapid rise in profit margins and the decline in labor’s share of income during the first six quarters of the current recovery is not that different from the rise in the first few years following the Great Recession and financial crisis of 2008. Figure T shows that starting from the trough of the recession (zero on the horizontal axis), the fall in the labor share of income was actually more pronounced during the early recovery from the Great Recession than it has been so far in the recovery from the COVID-19 recession.

Figure T

In the recovery from the Great Recession, increased corporate power did not manifest in faster price growth that made room for fatter profit margins—price growth was actually quite subdued (Bivens 2015). Instead, corporate power manifested itself in extreme wage suppression (aided by high and persistent levels of unemployment). Unit labor costs actually declined over a three-year stretch from the recession’s trough in the second quarter of 2009 to the middle of 2012. The general pattern of the labor share of income falling during the early phase of recoveries characterized most of the post-World War II recoveries, though it has become more extreme in recent business cycles (see Figures G and H in Bivens and Shierholz 2014).

Given that the rise in profit margins was similar in the 2008 recovery and the current one, it is hard to say that some recent rise in corporate power is the key driver of current inflation. Rather, a chronic excess of corporate power has built up over a long period of time, and it manifested in the current recovery as an inflationary surge in prices rather than successful wage suppression. What was different this time that channeled this power into higher prices rather than slower wage growth? Put simply, the main influence conditioning the recovery from the Great Recession was anemic growth in aggregate demand, whereas the main influences conditioning recovery post-2020 were the pandemic and the Russian invasion of Ukraine.

In previous recoveries—like the one following the Great Recession—domestic demand growth was slow and unemployment was high in the early phases of recovery. This led firms to become desperate for more customers but also gave them the upper hand in negotiations with potential employees, which led to subdued price growth and wage suppression.

This time around, the pandemic drove demand through the roof in durable sectors, and employment has rebounded rapidly, but the bottleneck in meeting this demand on the supply side was largely not labor (Bivens 2022). Instead, it was shipping capacity and other nonlabor shortages. Firms that did happen to have supply on hand as the pandemic-driven demand surge hit had enormous pricing power vis-à-vis their customers.

Policy in hindsight

Inflation has reached higher peaks and been more persistent than many would have predicted in early 2021. Given this, it is natural to ask what (if anything) should have been done differently by policymakers over this time. If one restricts this policy revisionism to, say, things that could have been done differently only since the end of 2020, obvious answers like “invest more in pandemic preparedness or more resilient supply chains” are off the table.

The most pressing policy debate concerns the actions of the Federal Reserve. Many inflation hawks would claim that the Fed has been far “behind the curve” on inflation. It’s not always entirely clear just what this means, however. Almost by definition, if the Fed had raised interest rates far enough and fast enough, inflation could have been significantly reduced. But the collateral damage of simply raising rates until inflation returns to 2% no matter the broader consequences could have been immense and made this approach easily not worth pursuing.

It is crucial to remember that inflation—particularly a short-run inflation that does not persist for years—generally has no aggregate cost. Instead, it is a purely distributional event. One actor’s cost is another actor’s income: As some group (workers, say) must pay more at stores for their consumption goods, the higher nominal prices feed directly into higher nominal incomes for somebody else. We may not like the pattern of redistribution caused by the current inflation (I certainly don’t), but it does not follow from this that it carries large aggregate costs.

Unemployment, conversely, really does carry high aggregate costs. By definition, an increase in unemployment caused by insufficient demand is economic waste—useful resources that could be deployed to produce more goods and services instead sit idle.

Costless rate hikes through expectations management?

A serious case that the Fed had gotten too far “behind the curve” on inflation would wrestle much harder with this potential trade-off. If the claim was that raising interest rates sooner would have squelched inflation while not requiring much increase in unemployment, this would be a compelling argument. This case is theoretically possible. If one believed that inflation expectations were the driver of nominal wage growth and subsequent price increases in 2021, these expectations (or at least expectations of inflation over the next year) really did move up sharply, and efforts—like starting rate increases sooner—that could have kept expectations in check might have helped.

But this assumes that expectations strongly condition subsequent inflation and that interest rate increases—even if they do not materially affect unemployment—have strong effects on these expectations. Neither of these propositions are well supported by the data.

The role of interest rates and housing

Outside of expectations, the one area where arguments about quicker rate hikes taking out some inflationary steam without harming the economic recovery more generally have some potential validity is housing. As we noted earlier, private industry data indicate a very sharp bounce-back of both rent inflation and housing prices by early 2021. Subsequent research by Mondragon and Wieland (2022) shows that the shift to remote work constituted a large positive shock to housing demand in 2020 and 2021.

Housing is by far the largest single component of price indices, and an acceleration of housing costs in mid-2022 was a key reason why core inflation remained substantially above the 2% target for most of this year. All of this provides some support for claims that the Fed should have raised rates more quickly on the heels of the passage of the American Rescue Plan.

In real time, however, it is not a complete certainty that this should have happened based on trends in the housing market. The Mondragon and Wieland (2022) results clearly imply that the housing price increases have a strong transitory element—unless a growing share of the population switches to remote work each and every year for the rest of the decade, there is little reason to think the upward price pressure imposed by this boost to housing demand will be sustained.

Further, if one thought that the shock to housing demand was transitory, then raising interest rates in response has potentially mixed effects. In the longer run, higher interest rates are clearly associated with reduced housing construction, limiting supply and exacerbating any excess demand. But Dias and Duarte (2016) have found evidence that, even in the short run, interest rate increases can actually increase rent inflation. The mechanism is through tenure choice—as interest rate increases boost the user cost of homeownership, prospective buyers switch into the rental market. In time, if the higher user cost pushes down purchase prices of homes enough, homeowners may choose to rent out rather than sell their homes when they wish to move, thus boosting rental supply. If in the short run the effect of interest rate increases on housing prices is ambiguous, and in the longer run it is potentially inflationary, it becomes less clear that the housing channel provides strong evidence that the Fed should have raised rates sooner in the current inflationary episode.

That said, the recent Fed rate hikes do seem to have relatively quickly released much inflationary pressure in housing markets, first in home prices and then (relatively quickly) in rental markets. As of October 2022, a few months of actual rent declines had occurred in many cities, and forecasters were predicting sharp rental price declines in 2023.

Was the American Rescue Plan the original sin of today’s inflation?

Previously, we highlighted evidence casting doubt on the claim that the American Rescue Plan was a primary contributor to the 2021–2022 inflation episode. Among other issues, the decomposition of inflation into “demand” and “supply” factors by Shapiro (2022) indicates that above-trend demand can account for just about 1 percentage point of core inflation acceleration by August 2022. One would have to attribute the entirety of this above-trend demand influence on inflation to ARP to use this evidence to indict ARP as more than a bit player in the inflation acceleration. But ARP’s spending impulse into the economy had largely petered out by the last quarter of 2021. Since the beginning of 2022, the federal fiscal impulse had actually turned historically contractionary. Figure U shows the change in federal net borrowing (-) or lending (+) over the previous year.

What it shows is that net borrowing by the federal government declined by an average of 10% of GDP over the first three quarters of 2022 (see the large upward spikes at the right edge of Figure U). This is roughly three times as much as the largest pre-pandemic reduction of borrowing (3.4%), which occurred in 2013 when fiscal austerity was widely acknowledged to be dragging heavily on growth from the Great Recession. Before 2007, the largest change in year-over-year borrowing was just 2.0%, a fiscal contraction less than a fifth as intense as the one in 2022. 

For further perspective, note that the swing in net borrowing by the household sector and financial crisis of 2008 was roughly 9% of GDP, but was spread over more than two years (for this calculation, see Bivens 2011). In that episode, the deflating housing bubble led families to reduce spending to make up for lost wealth driven by falling home prices. This bursting of the housing market bubble is why the Great Recession began and why it was so damaging. Further, this private-sector contraction in borrowing in 2006–2009 was even larger than the one that led to the Great Depression in the 1930s. In short, this evidence should make it hard to blame fiscal policy writ large for inflation that has persisted (and even accelerated) after fiscal policy swung hard from expansionary to historically contractionary.

One possibility for ARP’s effects to spill over well into 2022 is the ability of households to spend down the “excess savings” made possible by the fiscal aid in 2021. This is certainly plausible. The fiscal aid was almost surely largely saved (which is why actual GDP did not spiral rapidly above potential GDP in 2021 and early 2022). Figure V shows the increase in net worth of the bottom 50% of households and the size of pandemic fiscal relief. This relief can easily explain the rise in net worth, and this in turn can explain a potential “long fuse” of ARP as the aid initially boosted personal savings rates and then was spent out over time.

Figure U

Figure V

Some have pointed to the recent rapid falls in personal savings rates as evidence that this built-up excess savings from ARP was being rapidly spent down in 2022 and fueling too-fast demand growth (see Figure W for recent fall in savings rate).

Figure W

However, much of the rapid decline in personal savings might be mostly a statistical quirk unrelated to households spending down their pandemic assistance. This savings rate is measured as one minus the ratio of personal outlays divided by disposable personal income. As disposable personal income falls, the ratio of outlays to income rises and the measured savings rate falls. A very rapid increase in tax collections in 2022 led to a sharp fall in personal disposable income. Further, this increase can be almost fully explained by “nonwithheld” income taxes—which largely consist of capital gains taxes. Crucially, capital gains taxes push down measures of disposable personal income, but capital gains themselves are not included in measures of income. So as these collections rise, the savings rate is pushed down mechanically. Figure X shows the very sharp rise in federal income taxes as a share of personal income in recent quarters and shows that nearly all of this rise is accounted for by nonwithheld personal income taxes.

Could the American Rescue Plan have been structured differently to have caused less inflation?

With the benefit of hindsight, there are some changes to ARP one could have imagined. One reasonable-sounding change that was discussed in real-time—spreading the disbursement of funds over a longer time span—would likely not have made much of a difference. As Figure W shows, the large rise in pandemic aid was associated not with a huge wave of new spending, but instead with a large rise in savings (and net worth). Almost by definition, the large spike in savings kept much of the pandemic aid from translating quickly into new demand. Over time the excess savings have been rundown, but in a sense, households’ decisions smoothed out the spend-out of pandemic aid by themselves. A legislated “longer fuse” on this spending would not have slowed the spending much relative to what actually occurred.

The highest value of this pandemic aid—even when not spent—may have been the potential boost it gave to job seekers’ fallback positions when searching for jobs and the resulting acceleration of nominal wage growth. This wage growth is often seen solely as a contributor to inflation. But as we show in Figure R, most of the inflation seen over the past 18 months would have occurred even if nominal wage growth had not accelerated at all. Given this, the nominal wage acceleration was valuable in protecting workers’ real incomes against the inflationary spike.

The alternative changes to ARP that could have potentially blunted inflation in 2021 and early 2022 would have required simply a significantly lower level of spending or would have been seen as extremely heterodox. Simply reducing ARP’s spending levels would have led to marginally less inflation, but also would have led to significantly higher unemployment and even larger losses to real (inflation-adjusted) wages.

In terms of heterodox changes, one could imagine making some of the fiscal relief come in the form of vouchers that could be spent only on goods with a substantial lag. This would have given supply chains time to heal and provided an incentive to firms to invest heavily in repairing these distribution networks, knowing that customers would be waiting.

Another heterodox policy that could have blunted some of the major drivers of inflation was a temporary excess profits tax. We pointed out before the large role of widening profit margins in driving price increases. Imposing a large windfall tax on profits exceeding pre-pandemic margins could have blunted the incentive for firms to respond opportunistically to pandemic distortions (like impaired supply chains) that had temporarily reduced competitive pressures to keep prices low. There were some sectors in which the pros and cons of such a tax would have needed to be carefully weighed. Oil drilling and refining, for example, has been plagued for years with depressed investment, and this investment has responded sluggishly even to the extraordinary profits of recent years. This investment dearth has made the energy price spike in the U.S. historically large. An excess profits tax could have even further reduced this type of investment and made the energy price spike even worse. Then again, if investment in oil drilling and refining did not respond robustly to the highest profit margins in history for the sector, maybe relying on high after-tax profit margins to relieve price pressure in this sector was never going to work?

Figure X

What do macroeconomists and policy analysts need to know about inflation going forward?

There is a lot of truth to claims by macroeconomists that monetary policy can eventually neutralize the effect of relative price changes and restore inflation to a target level. It is also true that looking at the contributions to overall inflation in a given month made by specific sectors and then removing those sectors to find reassurance that what remains is not-that-fast inflation is a bad way to do policy analysis.

But throughout the current inflationary episode, a stronger claim has been often made: Relative price changes (and the sectoral shocks that caused them) are irrelevant to inflation even in the short run. Inflation is, in this view of the world, by definition evidence of a macroeconomic imbalance that needs to be rectified by changing macroeconomic aggregates.

This absolutely does not follow. The initial surge of inflation in 2021 occurred with unemployment still substantially higher than it was in the two years before the pandemic. As unemployment fell and other measures of macroeconomic tightness surged in late 2021 and early 2022, core inflation largely stabilized and key measures like nominal wage growth began falling.

Restoring intellectual respectability in policy debates to explanations that hinge on key sectoral imbalances is a key task for inflation analysis moving forward. It really should not be that hard. Analyses that highlight the crucial importance of particular sectors (and shocks to them) loom large in macroeconomic theories of long-run growth (see Blanchard and Kremer 1997 and Jones 2006). It hardly seems like a huge stretch to go from sectoral shocks causing long-run collapse in aggregate output to sectoral shocks causing an increase in medium-run (say 3–5 years) inflation dynamics.

Another crucial task for making inflation analyses smarter going forward is returning conflicting claims explanations of inflation’s persistence to prominence. Again, Tobin (1981), writing about the last large American inflation, expressed much wisdom that has seemingly been lost:

[I]nflation is the symptom of deep-rooted social and economic contradiction and conflict. There is no real equilibrium path. The major economic groups are claiming pieces of pie that together exceed the whole pie. Inflation is the way that their claims, so far as they are expressed in nominal terms, are temporarily reconciled. But it will continue and indeed accelerate so long as the basic conflicts of real claims and real power continue. (p. 28)

This will become especially important in any happy scenario in which the decades-long effort to shift bargaining power away from workers and toward employers is overturned. Distributional conflict—and nearly every other determinant of inflation’s persistence—has been easy to ignore for decades, simply because this conflict was well and truly settled in capital’s favor and inflation remained entirely quiescent. This settlement on capital’s terms was a disaster for the living standards of the vast majority, and it should be a progressive priority to overturn it and restore some bargaining power back to typical workers. But doing this—as 2021 demonstrated—will require keeping a close eye on inflationary dynamics.

Finally, today’s inflationary episode raises many questions about housing. The most obvious one is whether or not more timely measures of rent inflation can be used in analysis of macroeconomic stabilization policy. The backward-looking nature of housing prices in official indices really did leave many of us behind the curve on both the upslope and downslope of price changes. Adams et al. (2022) have done much of the work in demonstrating that more timely measures of building inflationary pressure in housing can be constructed. These more timely measures should be a bigger part of the monetary policy “dashboard.”

Another obvious issue in regard to housing is how it responds to interest rate hikes. There are potentially cross-cutting effects. Higher interest rates that slow growth of labor income will reduce demand for all types of housing. But if higher interest rates increase monthly costs of homeownership more rapidly than prices decline, there can be a period of time when these rate increases reduce the demand for homeownership, but this in turn increases the demand for rental housing. Because inflation as measured in the CPI or PCEPI is rent inflation, this means that interest rate hikes could actually raise housing inflation. Dias and Duarte (2016) provide evidence that this effect could be relevant empirically. All in all, the evidence of the current episode supports a view that interest rate increases reduce housing and rental prices, but the issues of tenure choice highlighted by analyses like Dias and Duarte (2016) should at least make policymakers think hard over the time horizon in which they are hoping prices will respond to rate increases.

Another issue, however, regards the treatment of housing in macroeconomic models. Rognlie (2015) has demonstrated that much of the rise in wealth documented by Piketty (2014) was driven by the rising price of housing. A number of analyses of the current inflationary period (and not just journalistic accounts) have argued that a very “hot” economy should naturally lead to rising profit shares and margins (i.e., debates over whether or not mark-ups were pro-cyclical). Earlier in this report, we show that this really did not seem to be the case for the corporate sector. But the corporate sector does not include housing. If housing is in quasi-fixed supply over the short run, then it really could be the case that hot economies start directing more and more income to landlords (and homeowners) than either workers or capital owners.

There is real reason to think this dynamic is getting more likely over time. Figure Y shows the share of personal consumption expenditures going to housing (either tenant-occupied rent or owners’-equivalent rent). It shows the actual share, as well as the share that would have prevailed had the price of housing risen at the same rate as nonhousing consumption expenditures. This counterfactual actually shows the share of housing rising more quickly than it actually did in the years leading up to 1979—meaning that housing prices rose consistently more slowly than nonhousing prices. Beginning in the early 1980s, there is a steady upward trend (punctuated by up and down spikes driven by the early 2000s housing bubble and the pandemic) in the actual housing share and a steady downward trend in the counterfactual, meaning that housing prices are rising substantially faster than nonhousing prices.

In short, if there were some wealth class in the economy that threatened to generate “forced savings” away from workers as the economy heats up over the course of a business cycle, it seems like housing might be it. The policy agenda to combat this is a whole new topic, but incorporating the dynamics of housing prices in a wider macroeconomic model could be a fruitful range of research spurred by the current inflationary episode.

Figuring out the impact of a global pandemic and war on inflation dynamics was always going to be challenging. Even worse, smart analyses of inflation, its causes, and proper remedies atrophied over recent decades as inflation seemed nearly permanently tamed. It is highly likely that in a few years, once the pandemic shock has passed, inflation will have returned to near irrelevance. But we should realize now that shocks happen: If smart analysis is not in economists’ mental toolkits, less smart reflexes will dominate policy discussion.

Figure Y

Notes

1. This report is a lightly-edited version of a working paper written in December 2022.

2. The decline in labor force participation likely slightly overstates the size of the supply shock hitting the labor market in recent years. Much of the decline in this measure is driven by older workers who did not work full time before the pandemic. Hence, the decline in potential output driven by a given percentage-point decline in labor force participation among this workforce is likely less than if it were driven by reduced participation among full-time and younger workers.

References

Adams, Brian, Lara Loewenstein, Hugh Montag, and Randal Verbrugge. 2022. “Disentangling Rent Index Differences: Data, Methods, and Scope.” Bureau of Labor Statistics (BLS) Working Paper no. 555, October 6, 2022.

Benmelech, Efraim, Nittai Bergman, and Hyunseob Kim. 2020. “Strong Employers and Weak Employees: How Does Employer Concentration Affect Wages?” Journal of Human Resources 58, no. 3.

Bivens, Josh. 2011. “The Stimulus: Two Years Later.” Testimony for a hearing before the Committee for Government Oversight, U.S. House of Representatives, Washington, D.C., February 16, 2011.

Bivens, Josh, and Heidi Shierholz. 2014. Lagging Demand, Not Unemployability, Is Why Long-Term Unemployment Remains So High. Economic Policy Institute, Briefing Paper #381, August 2014.

Bivens, Josh. 2015. A Vital Dashboard Indicator for Monetary Policy: Nominal Wage Targets. Center for Budget and Policy Priorities, June 2015.

Bivens, Josh. 2022. “U.S. Workers Have Already Been Disempowered in the Name of Fighting Inflation: Policymakers Should Not Make It Even Worse by Raising Interest Rates Too Aggressively.” Working Economics Blog (Economic Policy Institute), January 21, 2022.

Blanchard, Olivier, and Michael Kremer. 1997. “Disorganization.” The Quarterly Journal of Economics 112, no. 4: 1091–1126.

Bureau of Economic Analysis (BEA). 2022a. Gross Domestic Product (GDP) by Industry, various tables. Accessed November 2022.

Bureau of Economic Analysis (BEA). 2022b. National Income and Product Accounts (NIPAs), various tables. Accessed November 2022.

Bureau of Economic Analysis (BEA). 2022c. “Table 1.1.5. Gross Domestic Product,” National Income and Product Accounts (NIPAs). Accessed November 2022.

Bureau of Labor Statistics (BLS). 2022. Current Employment Statistics. Accessed November 2022.

Chetty, Raj. 2008. “Moral Hazard Versus Liquidity and Optimal Unemployment Insurance.” Journal of Political Economy 116, no. 2: 173–234.

Congressional Budget Office (CBO). 2021. Budget and Economic Outlook, 2021–2031. February 9, 2021.

Dias, Daniel A., and João Duarte. 2016. The Effect of Monetary Policy on Housing Tenure Choice as an Explanation for the Price Puzzle. International Finance Discussion Papers no. 1171, Federal Reserve Board, June 2016.

Economic Policy Institute (EPI). 2022. Current Population Survey Extracts, Version 1.0.36, https://microdata.epi.org.

Federal Reserve Bank of New York (Fed NY). 2022. Global Supply Chain Pressure Index. Accessed November, 2022.

Fernald, John. 2023. Total Factor Productivity, data page. Accessed November 2022.

Jones, Charles. 2006. “Intermediate Goods and Weak Links in the Theory of Economic Development.” American Economic Journal: Macroeconomics 3, no. 2: 1–28.

Klein, Ezra. 2022. “Transcript: Ezra Klein Interviews Larry Summers.” New York Times, March 29, 2022.

Mondragon, John, and Johannes Wieland. 2022. “Housing Demand and Remote Work.” Federal Reserve Bank of San Francisco Working Paper 2022-11, May 2022.

Organisation for Economic Co-operation and Development (OECD). 2022. OECD.Stat online database. Accessed November 2022.

Piketty, Thomas. 2014. Capital in the Twenty-First Century. Translated by Arthur Goldhammer. London, United Kingdom: Belknap Press of Harvard University Press.

Rampell, Catherine. 2022. “An Inflation Conspiracy Theory Is Infecting the Democratic Party.” Washington Post, May 12, 2022.

Rognlie, Matthew. 2015. “Deciphering the Fall and Rise in the Net Capital Share.” Brookings Papers on Economic Activity, spring 2015: 1–54.

Ros, Jaime. 1989. “On Inertia, Social Conflict, and the Structuralist Analysis of Inflation.” Working Paper no. 128, Kellogg Institute for International Studies, Notre Dame University, August 1989.

Shapiro, Adam. 2022. Supply- and Demand-Driven PCE Inflation. Federal Reserve Bank of San Francisco Economic Research.

Summers, Lawrence. 2021. “The Biden Stimulus Is Admirably Ambitious. But It Brings Some Big Risks, Too.” Washington Post, February 4, 2021.

Tobin, James. 1972. “Inflation and Unemployment.” American Economic Review 62, no. 1/2: 1–18.

Tobin, James. 1981. “Diagnosing Inflation: A Taxonomy.” In Development in an Inflationary World, edited by M. June Flanders and Assaf Razin. Cambridge, Mass.: Academic Press.

Zhou, Xiaoqing, and Jim Dolmas. 2022. “Rent Inflation Expected to Accelerate Then Moderate in Mid-2023.” Federal Reserve Bank of Dallas, August 16, 2022.

Differences in the supplemental and official poverty measures highlight the impact of pandemic support programs 

In 2011, the Census Bureau began annually releasing an additional poverty measure called the Supplemental Poverty Measure (SPM). Although imperfect, the SPM is a much better measure of poverty than the official poverty rate. SPM accounts for major government benefits like Social Security and child tax credits, and uses a more holistic measurement of modern costs of living and geographical differences in those costs. The latest data show that the 2021 SPM rates are the lowest on record for all years for which SPM estimates are available, starting from 2009. This is even more remarkable considering that the economic hardships and disruptions brought on by the COVID-19 pandemic were still very present during 2021.  

The impact of government safety net programs can be seen across the country, as the SPM was lower than the official poverty rate in 38 states when looking at a 3-year average of 2019–2021. (To ensure accurate comparison, the official estimates used for the comparison are based on an expanded definition of the official poverty rate that includes all the people within the universe of the SPM.) In fact, the SPM was higher (where the difference was statistically significant) than the official poverty measure in just three states (California, Maryland, and New Jersey), likely a reflection of these states’ higher costs of living.  

Comparing the 2019–2021 estimates to 2017–2019 rates, poverty as measured by the SPM decreased in all 50 states and the District of Columbia. The state SPM poverty rate fell the most in Louisiana, by 4.5 percentage points. This is followed by New Jersey and Maine, which saw 4.4 and 4.2 percentage point decreases, respectively. Unfortunately, the state-level SPM estimates are not disaggregated by race/ethnicity. See analysis by Valerie Wilson and Adewale A. Maye for a discussion of the racial disparities in poverty rates at the national level, as well as the impact of safety net programs as told by differences in the two poverty measures. 

Figure A

The official poverty rate changed little across the states in 2021 

The 1-year estimates from the American Community Survey (ACS) allow for year-to-year comparisons of economic conditions at the state level. However, given the disruptions to data collection and quality the pandemic caused, we compare 2021 data to estimates from 2019. Unfortunately, the ACS only reports the official poverty rates and thus SPM estimates are not available in published data at the state level for a one-year period. This is notable for 2021 because, as discussed above, the most striking changes in poverty occurred because of social insurance programs, many of which are not captured by the official poverty measure. 

As shown in Figure B, in 2021, the official poverty rate at the state level was highest in Louisiana (19.6%), Mississippi (19.4%), New Mexico (18.4%), and West Virginia (16.8%). The official poverty rate was lowest in New Hampshire (7.2%), Utah (8.6%), Minnesota (9.3%), and Colorado (9.7%). Comparing 2019 to 2021 estimates, official poverty rates did not dramatically change in most states—in 44 states, the increase or decrease was less than a percentage point. The biggest change was in the District of Columbia, where the rate increased by 3 percentage points, from 13.5% in 2019 to 16.5% in 2021. 

Figure B

The state-level official poverty rates do not capture the drastic impact of tax credit expansions on child poverty 

From the Census Bureau’s national poverty report based on 2021 ASEC data, we know that child poverty across the country fell to a historic low of 5.2%, as measured by the SPM. This is directly because of the expansion of anti-poverty programs at the height of the pandemic, including the Child Tax Credit (CTC). 

A remarkable 9.6 million people, including 4.9 million children, were kept out of poverty by refundable tax credits. The Child Tax Credit alone accounted for 5.3 million people kept out of poverty (out of the 9.6 million total).  

In contrast, according to the official poverty rate, child poverty at the national level sits at 15.3%. (To ensure accurate comparison, this estimate is based on an expanded definition of the official poverty rate that includes all the people within the universe of the SPM). As noted earlier, the official poverty rate does not account for the CTC and other government programs that saw great temporary expansions during the height of the pandemic. Unfortunately, the state-level child poverty rates published from the 2021 1-year ACS data are all based on the official poverty measure. The map in Figure B reports ACS child poverty rates in every state as measured by the official poverty measure. There are no published state-level SPM child poverty rates. 

State lawmakers should prioritize proven poverty-reduction programs 

Federal lawmakers made powerful improvements to the safety net in response to the pandemic. Bolstered unemployment insurance and expanded child tax credits, combined with existing federal supports such as Social Security and the Supplemental Nutrition Assistance Program (SNAP), kept millions of people out of poverty in 2021. With many of these programs expired and no indication that federal lawmakers will restore them soon, state policymakers should step in to build on the success of these policies that visibly lessened working people’s economic hardships and bolstered state economies. Many states still have federal funds available to them that they should use to set up programs that support working families and children. These programs should be made permanent by enacting progressive tax revenue collections that provide dedicated funding. Specific policies include enacting permanent UI reform, making the child tax credit permanently refundable, and increasing funding for and undoing the racist legacies of Temporary Assistance for Needy Families (TANF). 

Public investments in safety net programs continue to be extremely effective poverty reduction tools, as newly released Census income data show. Government social programs kept tens of millions of people out of poverty in 2021. Because of expansions to programs like unemployment insurance benefits and the Child Tax Credit, poverty rates were actually lower in 2021 than they were prior to the COVID-19 pandemic.  

The poverty reduction achieved through expanded social insurance programs highlights how much policymakers’ choices can impact poverty.   

Unfortunately, some of the program expansions enacted in the pandemic have already been reversed, and cuts to programs like unemployment benefits and the Child Tax Credit will increase household economic distress going forward. 

Last year, Social Security had the largest anti-poverty impact, reducing the number of people in poverty by 26 million. Recent policy expansions including refundable tax credits, like the earned income tax credit (EITC) and Child Tax Credit (CTC), and economic impact stimulus payments also reduced the number of people in poverty by roughly 10 and 9 million people respectively.  

Figure A

As Figure A shows, the third and final economic impact stimulus payment was measurably effective in alleviating economic distress, keeping close to 9 million people from poverty. The stimulus payment likely reached many who were ineligible for Social Security, the EITC, or CTC and who were still struggling from the effects of the pandemic shock to the economy. The sheer number of people this program supported underscores the success of direct income support programs.  

Another example of successful expansions to economic policy programs is the refundable tax credits, which include the EITC and CTC. These programs kept close to 10 million from poverty. The CTC benefit amounts were increased and made fully refundable in 2021 and have kept over 5 million people from poverty. Despite a remarkable reduction in child poverty in 2021, policymakers let the refundable CTC expire in December 2021.  

Census estimates show that unemployment insurance (UI) benefits kept 2.3 million people out of poverty in 2021. Prior to the pandemic, the UI system was much less effective at poverty reduction because of uneven state-based implementation, restrictive eligibility requirements, and relatively stingy benefits. But in 2020 and 2021, the UI system began to play an unprecedented role in reducing economic distress after Congress passed temporary extensions, including expanded eligibility for those with low incomes and independent contracts, longer benefit durations, and additional benefit amounts like the $300 supplement that expired mid-year. Ending these expanded programs in the middle of 2021 sharply reduced incomes and the consumer spending they bolstered. 

Notably, Census figures likely undercount the actual poverty reduction due to unemployment benefits. As in previous years, Census surveys severely underestimate unemployment benefit receipt. In 2020, the Current Population Survey that underlies these income and poverty measures captured less than half of the unemployment benefits reflected in administrative records. It is likely more complete data would show many millions fewer people in poverty as a result of receiving UI benefits. 

The Census poverty data offer strong evidence that millions were still experiencing economic hardship and financial insecurity throughout 2021. Despite the COVID-19 recession only officially lasting three months according to the National Bureau of Economic Research, the economic pain lingered. The sheer number of people who needed these programs offers a strong rebuke to those who claimed that economic recovery legislation like the 2021 American Rescue Plan was unnecessary spending.  

Looking forward, policymakers must continue to prioritize poverty reduction and make these high-impact social insurance programs permanent rather than temporary. The U.S. has long lagged behind its international peers in spending on anti-poverty, social welfare, and family benefits, choosing to accept high levels of poverty as the status quo. 

Racism has been a major barrier to expanding social insurance programs. Recent research shows that Americans consistently overestimate the share of Black people supported by government social programs, leading white Americans in particular to oppose many forms of social insurance. In addition, the temporary nature of the 2020 and 2021 expansions to government social programs suggests that, despite their huge success at reducing economic deprivation, many policymakers choose austerity over effective poverty reduction. 

Social insurance programs kept over 25 million people out of poverty in 2021. Millions of low-income Americans were supported by new or expanded programs, such as the economic impact stimulus payments, the refundable Child Tax Credit, and expanded unemployment insurance. Given the overwhelming effectiveness of these programs in keeping people out of poverty, it is unconscionable that policymakers have allowed them to expire and added to the stress of low-income families in the years to come.  

Education funding in the United States relies primarily on state and local resources, with just a tiny share of total revenues allotted by the federal government. Most analyses of the primary school finance metrics—equity, adequacy, effort, and sufficiency—raise serious questions about whether the existing system is living up to the ideal of providing a sound education equitably to all children at all times. Districts in high-poverty areas, which serve larger shares of students of color, get less funding per student than districts in low-poverty areas, which predominantly serve white students, highlighting the system’s inequity. School districts in general—but especially those in high-poverty areas—are not spending enough to achieve national average test scores, which is an established benchmark for assessing adequacy. Efforts states make to invest in education vary significantly. And the system is ill-prepared to adapt to unexpected emergencies.

These challenges are magnified during and after recessions. Following the Great Recession that began in December 2007, per-student education revenues plummeted and did not return to pre-recession levels for about eight years. The recovery in per-student revenues was even slower in high-poverty districts. This report combines new data on funding for states and for districts by school district poverty level, and over time, with evidence documenting the positive impacts of increasing investment in education to make a case for overhauling the school finance system. It calls for reforms that would ensure a larger role for the federal government to establish a robust, stable, and consistent school funding plan that channels sufficient additional resources to less affluent students in good times and bad. Furthermore, spending on public education should be retooled as an economic stabilizer, with increases automatically kicking in during recessions. Such a program would greatly mitigate cuts to public education as budgets are depleted, and also spur aggregate demand to give the economy a needed boost.

Following are key findings from the report:

Our current system for funding public schools shortchanges students, particularly low-income students. Education funding generally is inadequate and inequitable; It relies too heavily on state and local resources (particularly property tax revenues); the federal government plays a small and an insufficient role; funding levels vary widely across states; and high-poverty districts get less funding per student than low-poverty districts.

Those problems are magnified during and after recessions. Funding inadequacies and inequities tend to be aggravated when there is an economic downturn, which typically translates into problems that persist well after recovery is underway. After the 2007 onset of the Great Recession, for example, funding fell, and it took until 2015–2016, on average, to return to their pre-recession per-student revenue and spending levels. For high-poverty school districts, it took even longer—until 2016–2017—to rebound to their pre-recession revenue levels. And even after catching up with pre-recession levels, revenue levels in high-poverty districts lag behind the per-student funding in low-poverty districts. The general, long-standing funding inadequacies and inequities combined with the worsening of these problems during and in the aftermath of recessions have both short- and long-term repercussions that are costly for the students as well as for the country.

Increased federal spending on education after recessions helps mitigate funding shortfalls and inequities. Without increased federal education spending after recessions, school districts would suffer from an even greater decline in funding and even wider gaps between funding flowing to low-poverty and high-poverty districts.

Increased spending on education could help boost economic recovery. While Congress has enacted one-time education spending increases in difficult economic times, spending on public education should be considered one of the automatic stabilizers in our economic policy toolkit, designed to automatically increase and thus spur aggregate demand when private spending falls. Deployed this way, education spending becomes part of a set of large, broadly distributed programs that are countercyclical, i.e., designed to kick in when the economy overall is contracting and thus stave off or lessen the severity of a downturn. Along with other automatic stabilizers such as unemployment insurance, education spending thus would provide a stimulus to boost economic recovery.

We need an overhaul of the school finance system, with reforms ensuring a larger role for the federal government. In light of the concerns outlined in this report, policymakers must think differently both about school funding overall and about school funding during recessions. Public education is a public good, and as noted in this report, one that helps to stabilize the entire economy at critical points. Therefore, public spending on education should be treated as the public investment it is. While we leave it to policymakers to design specific reforms, we recommend an increased role for the federal government grounded in substantial, well targeted, consistent investment in the children who are our future, the professionals who help these children attain that future, and the environments in which they work. To establish a robust, stable, and consistent school funding plan that supports all children, investments need to be proportional to the size of the problems and to the societal and economic importance of the sector.

Introduction

The hope for the public education system in the United States is to provide a sound education equitably to all children regardless of where they live or into which families they are born. However, the COVID-19 pandemic exposed four interrelated, long-standing realities of U.S. public education funding that have long made that excellent, equitable education system impossible to achieve. First, inadequate levels of funding leave too many students unable to reach established performance benchmarks. Second, school funding is inequitable, with low-income students often and communities of color consistently lacking resources they need to meet their needs. Third, the level of funding reflects an overall underinvestment in education—that is, the U.S. is not spending as much as it could afford to spend in normal times. Fourth, given that educational investments are not sufficient across many districts even during normal times, schools are unable to make preparations to cope with emergencies or other unexpected circumstances. An added, less known feature is that economic downturns make all four of these problems worse. Downturns exacerbate funding inadequacies, inequities, underinvestment, and unpreparedness, causing cumulative harm to students, communities, and the public education system, and clawing back any prior progress. The severity of these problems varies widely across states and districts, as do the strength of states’ and localities’ economic and social protection systems, which may either compensate for or compound the problems.

The pandemic-led recession made these four major financial barriers to an excellent, equitable education system more visible, leading to serious questions about the U.S. education-funding model, which relies heavily on local and state revenues and draws only a small share of funding from the federal government. While public education is one of our greatest ideals and achievements—a free, quality education for every child regardless of means and background—the U.S. educational system is in need of significant improvements.

As the report will show, the core barriers to delivering universally excellent U.S. public education for all children—funding inadequacies and inequities that are exacerbated during tough economic times—were present in the system from the very start. They are the outcomes of a funding system that is shaped by many layers of policies and legal decisions at the local, state, and federal levels, creating widespread disparities in school finance realities across the thousands of districts across the country in all 50 states and the District of Columbia. This complex funding puzzle speaks to the need for a funding overhaul to attain meaningful and widely shared improvements.

In this report, we first provide an overview of the characteristics of the U.S. education funding system. We present data analyses on school finance indicators, such as equity, adequacy, and effort, that expose the shortcomings of funding policies and decisions across the country. We also discuss factors behind some of these shortcomings, such as the heavy reliance on local and state sources of funding.

Second, we illustrate that recessions exacerbate the funding challenges schools face. We parse a multitude of data to present trends in school finance indicators both during and after the Great Recession, demonstrating that the immediate effects of federally targeted funds helped schools navigate recession-induced budget cuts. We also look at the shortfalls and inequitable nature of those investments. We explore how increased federal investments—in good economic times and bad—could help address these long-standing problems. We argue that public education funding is not only an investment in our societal present and future, but also is a ready-made mechanism for countering economic downturns. Economic theory and evidence both demonstrate that large, broadly distributed programs providing public support serve as cushions during economic downturns: they spur overall spending and thus aggregate demand when private spending falls. As we note, there are strong arguments for placing public education spending within the broader category of effective fiscal responses to recessions that are countercyclical—designed to increase spending when spending in the economy overall is contracting and thus stave off or lessen the severity of a downturn. Increases in public education spending during downturns work as automatic stabilizers for schools and provide stimulus to boost economic recovery. We review existing research on the consequences of funding in general and of funding changes—evidence that supports a larger role for the federal government.

Third, we discuss the benefits of rethinking public education funding, along with the societal and economic advantages of a robust, stable, and consistent U.S. school funding plan, both generally and as a countercyclical policy. We show that federal investment that sustains school funding throughout recessions and recoveries would provide three major advantages: It would help boost educational instruction and standards, it would provide continued high-quality instruction for students and employment to the public education workforce, and it would stimulate economic recovery. Education funding, in particular, would blanket the country while also targeting areas with the most need, making the recovery more equitable.

We conclude the report with final thoughts and next steps.

A funding primer

The American education system relies heavily on state and local resources to fund public schools. In the U.S. education has long been a local- and state-level responsibility, with states typically concerned with administration and standards, and local districts charged with raising the bulk of the funds to carry those duties and standards out.

The Education Law Center notes that “states, under their respective constitutions, have the legal obligation to support and maintain systems of free public schools for all resident children. This means that the state is the unit of government in the U.S. legally responsible for operating our nation’s public school systems, which includes providing the funding to support and maintain those systems” (Farrie and Sciarra 2021). Bradbury (2021) explains that state constitutions assign responsibility for “adequate” (“sound,” “basic”) and/or “equitable” public education to the state government. Most state governments delegate responsibility for managing and (partially) funding public pre-K–12 education to local governments, but courts mandate that states remain responsible.

States meet this responsibility by funding their schools “through a statewide method or formula enacted by the state legislature. These school funding formulas or school finance systems determine the amount of revenue school districts are permitted to raise from local property and other taxes and the amount of funding or aid the state is expected to contribute from state taxes. In annual or biannual state budgets, legislatures also determine the actual amount of funding districts will receive to operate their schools” (Farrie and Sciarra 2020).

The federal government seeks to use its limited but targeted funding to promote student achievement, foster educational excellence, and ensure equal access. The major federal agency channeling funding to school districts (sometimes through the states) is the U.S. Department of Education.1

Figure A shows the percentage distribution of total revenue for U.S. public elementary and secondary schools for the 2017–2018 school year, on average. As illustrated, revenues collected from state and local sources are roughly equal (46.8% and 45.3%, respectively). Two other factors also stand out. First, revenue from property taxes accounts for more than one-third of total revenue (36.6 %). Second, federal funding plays a minimal role, providing less than 8% of total revenue (7.8%). As discussed later in the report, this heavy reliance on local funding is a major driver in the funding challenges districts face.

Figure A

Key metrics reveal the four major financial barriers to an excellent, equitable education system 

Fully comprehending how school funding works and how it contributes to systemic problems requires drawing on key metrics and characteristics that define the education investments or education funding. Understanding these metrics is the first step toward designing a comprehensive solution.

The adequacy metric tells us that funding is inadequate

Adequacy, one of the most widely used school finance indicators, measures whether the amount raised and spent per student is sufficient to achieve a certain level of output (typically a benchmark of student performance or an educational outcome).

We use the adequacy data provided by Baker, Di Carlo, and Weber (2020). These authors, who use the School Finance Indicators Database, compare current education spending by poverty quintile with spending levels required for students to achieve national average test scores—typically accepted as an educationally meaningful benchmark. The authors’ estimates account for factors that could affect the cost of providing education, including student characteristics, labor-market costs (differences in costs given the regional cost of living), and district characteristics (larger districts for example may enjoy economics of scale).

Figure B reveals that spending is not nearly enough, on average, to provide students with an adequate education. As this figure illustrates, relative to the wealthiest districts, the highest-poverty districts need more than twice as much spending per student to provide an adequate education. As the figure also shows, the gaps between what is spent on each student and what would be required for those students to achieve at the national level widen as the level of poverty increases. Medium- and high-poverty districts are spending, respectively, $700 and $3,078 per student less than what would be required. For the highest-poverty districts, that gap is $5,135, meaning districts there are spending about 30% less than what would be required to deliver an adequate level of education to their students. (Conversely, the two low-poverty quintiles are spending more than they need to reach that benchmark, another indication that funds are being poorly allocated.)

Figure B

The equity metric tells us that funding is inequitable 

An equitable funding system ensures that, all else being equal, schools serving students with greater needs—whether for extra academic, socioemotional, health, or other supports—receive more resources and spend more to meet those needs than schools with a lower concentration of disadvantaged students. Across districts, states, and the country as a whole, this means allocating relatively more funding to districts serving larger shares of high-poverty communities than to wealthier ones. While our funding system does allocate additional funds based on need (e.g., to students officially designated as eligible for “special education” services under the federal Individuals with Disabilities Education Act and to children from low-income families through the federal Title I program), in practice, more funding overall goes to lower-needs districts than to those with high levels of student needs.

Figure C compares districts’ per-student revenues and expenditures by poverty level, and shows gaps relative to low-poverty districts. The figure is based on data from what was, when this research was conducted, the most recent version of the Local Education Agency Finance Survey (known as the F-33) (NCES-LEAFS, various years). As shown in the figure, on average, per-student revenue and spending in school districts serving wealthier households exceed revenue and spending in all other districts. In low-poverty districts (i.e., districts with a poverty rate in the bottom fourth of the poverty distribution), per-student revenues averaged $19,280 in the 2017–2018 school year, and per-student expenditures averaged $15,910. In the high-poverty districts (i.e., in the top fourth of the poverty distribution), per-student revenues were just $16,570, and per-student expenditures were $14,030. High-poverty districts raise $2,710 less in per-student revenue than the lowest–poverty school districts, reflecting a 14.1% revenue gap—meaning high-poverty districts receive 14.1% less in revenue. Per-student spending in high-poverty districts is $1,880 less than in low-poverty districts, an 11.8% gap.2 In other words, rather than funding districts to address student needs, we are channeling fewer resources—about 14% less, per student—into districts with greater needs based on their student population.

Figure C

Adequacy and equity are closely intertwined

In recent decades, researchers have explored challenges to both adequacy and equity in U.S. public education. For example, Baker and Corcoran (2012) analyzed the various policies that drive inequitable funding. Likewise, lawsuits that have challenged state funding systems have tended to focus on either the inadequacy or inequity of those schemes.3

But in reality, especially given extensive variation across states and districts, the two are closely linked and interact with one another. At the state level, for example, apparently adequate levels of funding can mask disparities across districts that innately mean inadequate funding for many, or even most, districts within that state (Farrie and Schiarra 2021).4

In addition, disparate levels of public investments in education are often made in a context that correlates positively with disparate levels of parents’ private investments in their children’s education and related support (Caucutt et al. 2020; Duncan and Murnane 2016; Kornrich 2016; Schneider, Hastings, and LaBriola 2018). Substantial research on income-based gaps in achievement demonstrates that large and growing wealth inequality plays a role. Parents at the top of the income or wealth ladders, who can and do pour extensive resources into their children’s human capital, constantly set a baseline of performance that can be hard for children and schools without such investment to attain (Reardon 2011; García and Weiss 2017).5

The “effort” metric tells us that many states are underinvesting in education relative to their capacity

 “Effort” describes how generously each state funds its schools relative to its capacity to do so. Researchers measuring effort determine capacity to spend based on state gross domestic product (GDP), which can vary widely (just as wealthier neighborhoods can raise more revenues even with lower tax rates, states with higher GDP and thus greater revenue-raising capacity can attain higher revenue with a lower effort, i.e., generate more resources at a lower cost). The map (Figure D), reproduced from Farrie and Sciarra 2021, shows state funding effort from the 2017–2018 school year.

Figure D

As Farrie and Sciarra (2021) note, states fall naturally into four groups:

• High-effort, high-capacity: States such as Alaska, Connecticut, New York, and Wyoming are high- capacity states with high per-capita GDP, and they are also high-effort states: They use a larger-than-average share of their overall GDP to support pre-K–12 education, which generates high funding levels.
• High-effort, low-capacity: States such as Arkansas, South Carolina, and West Virginia have lower-than-average capacity, with low GDP per-capita, but they are high-effort states. Even with above- average efforts, they yield only average or below-average funding levels.
• Low-effort, high-capacity: States such as California, Delaware, and Washington are high-capacity states that exert low effort toward funding schools. If these states increased their effort even to the national average, they could significantly increase funding levels.
• Low-effort, low-capacity: States such as Arizona, Florida, and Idaho are low-capacity states that also make lower-than-average efforts to fund schools, generating very low funding levels.

Evidence shows that districts and schools lack the resources to cope with emergencies

As the COVID-19 pandemic has made clear, our subpar level of preparation to cope with emergencies or other unexpected needs reflects another aspect of underinvestment. As García and Weiss (2020) not about the COVID-19 pandemic, “Our public education system was not built, nor prepared, to cope with a situation like this—we lack the structures to sustain effective teaching and learning during the shutdown and to provide the safety net supports that many children receive in school.”

Whether due to lack of resources, planning, or other factors, districts, schools, and educators struggled to adapt to the pandemic’s requirements for teaching. Schools were unprepared not only to support learning but also to deliver the supports and services they were accustomed to providing, which go far beyond instruction (García and Weiss 2020). This lack of preparation was the result of both a lack of contingency planning as well as a failure to build up resources to be ready “to adequately address emergency needs and to compensate for the resources drained during the emergencies, as well as to afford the provision of flexible learning approaches to continue education” (García and Weiss 2021).

A lack of established contingency plans to ensure the provision of education in emergency and post-emergency situations, whether caused by pandemics, other natural disasters, or conflicts and wars (as examined by the education-in-emergencies research), prevents countries from being able to mitigate the negative consequences of these emergencies on children’s development and learning. The lack of contingency plans also leaves systems unprepared to help children handle the trauma and stress that come from the most serious events. This body of literature has also shown that access to education and services—and an equitable and compensatory allocation of them—helps reduce the damage that students experience during the crisis and beyond, since such emergencies carry long-term consequences (Anderson 2020; Özek 2020).

Public education’s over-reliance on local funding is a key factor behind the troubling funding metrics

The heavy reliance on local funding described above is at the core of the school finance problems. Extensive research has exposed the challenges associated with this unique American system for funding public schools.6 The myriad factors that drive school funding—politics and political affiliation, state legislative and judicial decisions, property values, tax rates, and effort, among others—vary substantially from one community to another. Thus, it is not surprising that this system has contributed to institutionalizing inequities, especially in the absence of a strong federal effort to counter them.

It is well understood that the local sources of revenues on which school districts heavily rely are often distributed in a highly inequitable way. Revenues from property taxes, which make up a hefty share of local education revenues, innately favor wealthier communities, as these areas have a much larger capacity to raise funds based on higher property values despite their lower tax rates.7 These higher property-tax revenues in wealthier areas lead to greater revenues for their districts’ schools, since property-tax revenues account for such a significant share of the total.

State and federal funding are insufficient to compensate for these locally driven inequities

State funding of public education is the largest budget line item for most states.8 Along with federal funding, state funding is expected to make up for local funding disparities and gaps.9 Federal funding, in particular through Title I of the Elementary and Secondary Education Act (ESEA), is specifically designed to compensate low-income schools and districts for their lack of sufficient revenues to meet their students’ needs.10 Similarly, state funding is intended to offset some of the disparities caused by the dependence on local revenues. However, in reality, state and federal sources do not provide enough to less-wealthy school districts to make up for the gap in funding at the local level, as shown in Figure E.

As the figure shows, the U.S. systematically funds schools in wealthier areas at higher levels than those with higher rates of poverty, even after accounting for funding meant to remedy these gaps. On average, local property-tax funding per student is $5,260 lower in the poorest districts than in the wealthiest districts.

Figure E

While state revenues are a significant portion of funding, they only modestly counter the large locally based inequities. And while federal funding, by far the smallest source of revenue, is being deployed as intended (to reduce inequities), it inevitably falls short of compensating for a system grounded in highly inequitable local revenues as its principal source of funding. As such, although states provide their highest-poverty districts with $1,550 more per student than to their lowest-poverty districts, and federal sources provide their highest-poverty districts with $2,080 more per student than to their lowest-poverty districts, states and the federal government jointly compensate for only about half of the revenue gap for high-poverty districts (which receive a per-student average of $6,330 less in property tax and other local revenues). That large gap in local funding leaves the highest-poverty districts still $2,710 short per student relative to the lowest-poverty districts, reflecting the 14.1% revenue gap shown in Figure C. Even though high-poverty districts get more in federal and state dollars, they get so much less in property taxes that it still puts them in the negative category overall.

Disparities shortchange states’ (and districts’) ability to access and allocate the resources needed for effective education

Given the heavy reliance on highly varied local funding, it is no surprise that there is similarly significant variation across states with respect to almost every aspect of funding discussed here. Table 1 reports federal, state, and local funding for each state and for the District of Columbia, with local funding broken down into three categories.

Table 1

Nationally, in 2017–2018, local and state sources accounted for 45.3% and 46.8% of total revenue, respectively; just 7.8% comes from the federal government. However, these averages mask substantial variation in the shares of revenue apportioned by each source across states. Local revenue, for example, ranges from just 3.7% of total public-school revenue in Vermont and 18.2% in New Mexico, on the lower end, to a high of 63.4% in New Hampshire. The same is true with respect to state revenue. The state that contributes the smallest share to its education budget is New Hampshire at 31.3%, with Vermont contributing the largest share (89.9%). There is also quite a bit of variation in the share represented by federal funds—from just 4.1% in New Jersey to 15.9% in Alaska. (The cited values are highlighted in the table. We omit the District of Columbia and Hawaii from these rankings because of the unusual composition of their funding streams, but we provide their values in the table.)

As shown earlier in the discussion of the map in Figure E, there are also large disparities in funding effort—how generously each state funds its schools relative to its capacity to do so, based on state GDP. High-effort, high-capacity states such as Alaska, Connecticut, New York, and Wyoming use a larger-than-average share of their overall GDP to support pre-K–12 education and they generate high funding levels.

As a result of funding and effort variability across states, the levels of inequity and inadequacy across states also vary substantially (Baker, Di Carlo, and Weber 2020; Farrie and Sciarra 2021). Notably, funding variability translates into significant disparities in overall per-student revenue and per-student spending levels, as shown in Figures F and G. In Wyoming, for example, where effort is relatively high (4.36%; see Figure E) and there is a higher-than-average contribution of state funds to total revenue and a lower-than-average contribution of local funds to total revenue (56.8% and 36.8%, respectively, versus 46.8% and 45.3% averages across the U.S.), per-student revenue is among the highest of any state, nearly $21,000. In contrast, Arizona and North Carolina—which are among the lowest in effort in the country (2.23% and 2.28%, respectively), but where state funds account for 47.1% and 62.1% of the state’s total public education revenues, respectively, and local funds account for 40.4% and only 27.0%, respectively—collect about half of what Wyoming collects per student. (Data accounts for differences in states’ cost of living; see the appendix for more details on our methodology.)

Figure F

Figure G

These substantial disparities in all the school finance indicators, and in per-pupil spending and revenue across states, are mirrored in capacity and investment patterns across districts and, within them, individual schools.

As such, these systemic and persistent inequities play a decisive role in shaping children’s real school experiences. As Raikes and Darling-Hammond (2019) note, “As a country, we inadvertently instituted a school finance system similar to red-lining in its negative impact. Grow up in a rich neighborhood with a large property tax base? You get well-funded public schools. Grow up in a poor neighborhood? The opposite is true. The highest-spending districts in the United States spend nearly 10 times as much as the lowest-spending, with large differentials both across and within states (Raikes and Darling-Hammond 2019). In most states, children who live in low-income neighborhoods attend the most under-resourced schools” (see also Turner et al. 2016 for the underlying data).11

These gaps in spending capacity touch every aspect of school functioning, including the capacity of teachers and staff to deliver effective instruction, and pose a huge barrier to the excellent school experience that each student should receive. In Pennsylvania, for example, where districts tend to rely heavily on local revenues to finance schools, per-pupil spending ranges dramatically. Indeed, in 2015, the U.S. Department of Education flagged the state as having the biggest school-spending gap of any state in the country (Behrman 2019). One illustrative example is in Allegheny County, on the western side of the state, where the suburban Wilkinsburg school district outside of Pittsburgh spent over $27,000 per student in the 2017–2018 school year, while the more rural South Allegheny school district spent just over $15,000, roughly 45% less.

With salaries being the largest line item in school budgets, these disparities substantially affect schools’ ability to hire the educators and other school personnel needed to provide effective instruction, the school leaders to guide instructional staff, and the staff needed to support administrative needs and to offer other services and extracurricular activities. As a result, these resources vary tremendously not only among states, but within them from one district, and even school, to another.12 Overwhelming research exposes large disparities in access to counselors, librarians, and nurses, and in access to up-to-date technology and facilities. Facilities are literally crumbling in lower-resourced states and districts, painting a clear picture of the dire straits many schools face. (See, for example, Filardo, Vincent, and Sullivan 2019 regarding added consequences for low-income students and their teachers in schools that are too cold, full of dust or lead paint, and have broken windows or crumbling ceilings.)

Baker, Farrie, and Sciarra (2016) note that “increasing investments in schools is associated with greater access to resources as measured by staffing ratios, class sizes, and the competitiveness of teacher wages.” The findings presented here are backed by the extensive body of literature on the positive relationship between substantive and sustained state school finance reforms and improved student outcomes. Together, they make a strong case that state and federal policymakers can help boost outcomes and close achievement gaps by improving state finance systems to ensure equitable funding and improved access to resources for children from low-income families.

Economic downturns exacerbate the problems with our school finance system and, over time, cause cumulative damage to students and to the system

Recessions lead to depleted state and local budgets and, in turn, to cuts in education funding. Trends since the Great Recession demonstrate that it can take a long time to restore education budgets and that our practice of balancing budgets on the backs of schoolchildren is an unwise and, ultimately, costly one in terms of educational and societal outcomes. As we show in Figure H, reductions in revenue for public education often outlast the official length of the recession, lasting much longer than the point when state and local budgets have returned to pre-recession trajectories in other areas of spending. In addition, a poor allocation of resources across high- and low-poverty districts disproportionately harms students in the highest-poverty districts relative to their peers in better-off districts, compounding the existing challenges described above and impeding their recovery.

It took the United States nearly a decade to restore the national per-student revenue to its pre-recession (2007–2008) school-year levels. Figure H shows national trends in revenue per student, by source (federal, state, and local), from the onset of the Great Recession through 2017–2018.13

Figure H

Per-student state revenue fell precipitously between 2007–2008 and 2012–2013—it was down nearly $900 at the low point. While revenue from property taxes did not decrease, on average, other local revenues fell by $160 by 2011–20121, only recovering to 2007–2008 levels in 2014–2015. Federal funding for schools, together with the additional recovery funds targeted to education through the 2009 American Recovery and Reinvestment Act (ARRA), provided an initial and critical counterbalance to these reductions; in 2009–2010 and 2010–2011, districts were receiving slightly over $600 more per student from the federal government than they were before the recession.

The peak in federal revenue is also visible in Figure I, which depicts the distribution of funding by sources by year. Total federal funds accounted for 12.7% of total revenue in 2009–2010, compared with just 8.2% in 2007–2008, an increase of over 50%. (Note that this increase was made larger by the reduced total amounts of revenues, i.e., it constituted a greater share of a smaller whole).

Figure I

While these federal investments provided a critical boost by temporarily upholding education funding, our analyses suggest an opportunity to shorten the slow recovery to pre-recession levels was lost. Just as they effectively operated during the recession, it is likely that larger and more sustained federal investments would have better assisted the students, schools, and communities that suffered major setbacks due to the Great Recession. We come back to this idea in sections below.

In keeping with the discussion on broad funding disparities by state, the road to recovery from the Great Recession also varies across states and districts, with some still lagging from the Great Recession as they struggled with the COVID-19 crisis.

Research demonstrates that well after the end of the Great Recession, a significant number of states were still funding their public schools at lower levels than before the recession. As late as 2016, for example, per-student funding in 24 states—including half of the states with over a million enrolled students—was still below pre-recession levels (Leachman and Figueroa 2019). For some of these states, the failure to return to prior funding levels was driven by the lack of recovery of the per-student state revenue (for example, Alabama, Alaska, Arizona, Florida, Mississippi, Montana, New Mexico, and Oklahoma). In some of the “deepest-cutting states — including Arizona, North Carolina, and Oklahoma,” note Leachman and Figueroa, the state governments made significant cuts to income tax rates, “making it much more difficult for their school funding to recover from cuts they imposed after the last recession hit.” In other states, lack of local revenue was the culprit (as in Hawaii, Indiana, Kansas, and Vermont, for example). Finally, in some of these states, this shortfall fell on top of a rapidly growing student population (i.e., even had their total revenues recovered to pre-recession levels, they would still fall far behind on a per-student basis). Exploring the various drivers of these trends and their variation across states is beyond the scope of this report but would undoubtedly be fruitful.14

Putting aside state trends and underlying causes, a focus on school districts reveals a strong correlation between poverty rates and education funding recovery. The following figures show the trends over time in total per-student revenue and spending by school district poverty levels. As we see, high-poverty districts and their students experienced both the biggest shortfalls and the most sluggish recoveries.

Figure J shows that, as discussed above, districts with relatively small shares of low-income students (low-poverty districts) never saw revenues per student fall below pre–Great Recession levels, adjusted for inflation and state cost of living. By contrast, the one-fourth of districts with the largest share of students from poor families (high-poverty districts) stayed below their pre–Great Recession level of per-student revenues long after recovery was in full swing, through 2015–2016. In keeping with that spectrum, the medium-high poverty districts did recover to their pre-recession per-student revenue levels, but not until 2014–2015.

Figure J

Figure K tells a similar story regarding trends in per-student expenditure across school districts. As such, it took until 2017–2018, a decade after the Great Recession had first hit, for high-poverty school districts to surpass their pre-recession levels, though they still lagged far behind their wealthier counterpart districts. Moreover, though not shown in this graph, for high-poverty districts, getting back to pre-recession status means catching up to revenue and spending levels that were lower than in the wealthier districts to begin with. (Figure C earlier in the report illustrates the gaps between high- and low-poverty districts in 2017–2018.)

Figure K

Balancing budgets on the backs of children during a recession has serious consequences

Inadequate, inequitable funding relegates poor children to attend under-resourced schools even in good economic times, and to suffer disproportionately during and in the aftermath of economic downturns. We have for far too long been balancing recession-depleted budgets on the backs of schoolchildren, in particular low-income children and children of color. This not only hurts these children immediately, but severely limits their prospects as adults. As such, this practice has broader implications for the future of the country, both economically and regarding the strength of our societal fabric, given that the students of today are the workers and the citizens of tomorrow.

Indeed, these negative patterns are just the first indications of a cascade of consequences that result from funding cuts. This section describes those consequences and their flip side, which is more frequently the focus of education researchers—the positive effects of increased investment. First, we review the literature demonstrating the impacts of various levels of funding on student outcomes. Next, we point to analyses that have shown some other associated school problems (education employment, class size, and student performance, among others) that were contemporaneous with the declines in spending and revenue. Thought it is difficult to quantify the exact and independent impact of the funding cuts on these factors, the strong correlations suggest that they are related.

Substantial evidence points to the positive effects of higher spending on both short- and long- term student outcomes, as well as on schools overall and on adult outcomes (Jackson and Mackevicius 2021; Jackson, Johnson, and Persico 2016; Gibbons, McNally, and Viarengo 2018; Hyman 2017; Lafortune, Rothstein, and Schanzenbach 2018; Jackson 2018; Jackson, Wigger, and Xiong 2020; Baker 2018). This body of research also provides evidence that the impact of school spending differs by students’ family income (Lafortune, Rothstein, and Schanzenbach 2018; Jackson, Johnson, and Persico 2016). And, though less has been studied in this specific area, the evidence also shows that a misallocation of resources and/or a decrease in spending has a negative influence on student outcomes, as well as on some teacher outcomes (Jackson, Wigger, and Xiong 2020; Greaves and Sibieta 2019).15

A recent summary of the literature provides compelling evidence of the effects of school spending on test scores and educational attainment. Based on 31 studies that provide reliable causal estimates, Jackson and Mackevicius (2021) find that, on average, a $1,000 increase in per-pupil public school spending for four years increases test scores by 0.044 percentage points, high school graduation by 2.1 percentage points, and college-going by 3.9 percentage points. Interestingly, the authors explain that “when benchmarked against other interventions, test score impacts are much smaller than those on educational attainment—suggesting that test-score impacts understate the value of school spending.” Consistent with a cumulative effect, the educational attainment impacts are larger after more years of exposure to the spending increase, and average impacts are similar across a wide range of baseline spending levels, indicating little evidence of diminishing marginal returns at current spending levels.

Other research suggests that the effect of spending is greater on disadvantaged students. Bradbury (2021) investigates “how specific state and local funding sources and allocation methods (redistributive extent, formula types) relate to students’ test scores and, especially, to test-score gaps across races and between students who are not economically disadvantaged and those who are.” Her findings suggest that statewide per-student school aid has no relationship with test-score gaps in school districts, but that the progressivity of the state’s school-aid distribution is associated with smaller test-score gaps in high-poverty districts.16

Other studies further affirm the implications of equity-specific funding decisions. Jackson, Johnson, and Persico’s (2016) study assesses the impacts on a range of student and adult outcomes of a series of court-mandated school finance reforms that took place in the 1970s and 1980s. Linking information on the reforms to administrative data about the children who attended the schools, the authors found that the increase in school funding was associated with slight increases in years of educational attainment, and with higher adult wages and reduced odds of adult poverty, as well as with improvements to schools themselves—increased teacher salaries, reduced student-to-teacher ratios, higher school quality, and even longer school years (Jackson, Johnson, and Persico 2016). Specifically, a 10% increase in per-pupil spending each year for all 12 years of public schooling leads to 0.27 more completed years of education, 7.25% higher wages, and a 3.67 percentage-point reduction in the annual incidence of adult poverty. As with the other studies, the benefits from increased funding are much greater for children from low-income families: 0.44 years of educational attainment and wages that are 9.5% higher.

In another study drawing on data from post-1990 school finance reforms that increased public-school funding in some states, Lafortune, Rothstein, and Schanzenbach (2018) estimate the impact of both absolute and relative spending on achievement in low-income school districts, as measured by National Assessment of Educational Progress (NAEP) data.17 They find that the reforms increase the achievement of students in these districts, phasing in gradually over the years following the increase in spending/adequacy. While the measures employed to estimate the impact tend to be technical, the authors emphasize that this “implied effect of school resources on educational achievement is large.”18 Similar adequacy-related reforms that resulted from court mandates, rather than state legislative decisions, prompted significant increases in graduation rates (Candelaria and Shores 2019).

Conversely, research shows that both the reallocation of resources and/or a decrease in spending have a negative influence on both teacher and student outcomes. Jackson, Wigger, and Xiong (2020) find that the cuts to per-pupil spending that occurred during the Great Recession reduced test scores and college enrollment, particularly for children in poor neighborhoods. Shores and Steinberg (2017) reaffirm these findings, noting that the Great Recession negatively affected math and English language arts (ELA) achievement of all students in grades 3–8, but that this “recessionary effect” was concentrated among school districts serving both more economically disadvantaged students and students of color. Greaves and Sibieta (2019) find that changes that required districts to pay teachers following higher salary scales, but that provided no additional funding to implement the requirements, did lead to increased pay for teachers as intended, but at the expense of cuts to other noninstructional spending of about 4%, with no net effects on student attainment. That is, reallocating resources across functions, without increasing the overall levels, did not improve outcomes.

Other studies explore disappointing trends across multiple education parameters during the decade preceding the COVID-19 pandemic, including teacher employment, class size, aggregate student performance, and performance gaps by socioeconomic status and/or racial/ethnic background. Several analyses show that recession-led school funding cuts were contemporaneous with significant reductions of teacher employment. The number of teachers in the United States public-school system reached its highest point in 2008, and then dropped significantly between 2008 and 2010 because of the recession (Gould 2017; Gould 2019; Berry and Shields 2017). Evans, Schwab, and Wagner (2019) estimated a decrease in total employment in public schools of 294,700 from the start of the recession until January 2013. Gould (2019) estimated that, in the fall of 2019, there were still 60,000 fewer public education jobs than there had been before the recession began in 2007 and that, if the number of teachers had kept up proportionately with growing student enrollment over that period, the shortfall in public education jobs would be greater than 300,000.

Related to these challenges, in the aftermath of the Great Recession through the 2015–2016 school year, schools’ struggles to staff themselves increased sharply. García and Weiss (2019) showed that the share of schools that were trying to fill a vacancy but could not do so tripled from the 2011–2012 to the 2015–2016 school year (increasing from 3.1% to 9.4% of schools in that situation), and the share of schools that reported finding it very difficult to fill a vacancy nearly doubled (from 19.7% to 36.2%).19

Although class size, and the closely related metric of student-to-teacher ratios, have declined over the long term, they are higher, on average, in 2020 than they were in 2005 (the closest data point prior to the Great Recession) in 29 out of the 50 states plus the District of Columbia (NCES 2020d; Hussar and Bailey 2020). (See Mishel and Rothstein 2003 and Schanzenbach 2020 for a recent review of the influence of class size on achievement.)

Understanding overall trends in student performance over this period helps to put the impacts of trends in these other metrics in context. We have cited research that links school finance trends and educational outcomes in the aftermath of the Great Recession, but it is worth describing what the trends in student performance looked like across the country. It should not be surprising that scores from the National Assessment of Educational Progress (NAEP), the most reliable indicator over time of how much students are learning, show stagnant performance in math and reading for both fourth- and eighth-graders between 2009 and 2019 (NAGB 2019). As Sandy Kress, who served as President George W. Bush’s education advisor, commented, “The nation has gone nowhere in the last ten years. It’s truly been a lost decade [and] [t]he only group to experience more than marginal gains in recent years has been students in the top 10th percentile” (Chingos et al. 2019).

Gains (both absolute and relative) vary by students’ background, with multiple trends visible. Carnoy and García’s 2017 research on achievement gaps between racial/ethnic groups shows that Black–white and Hispanic–white student achievement gaps have continued to narrow over the last two decades, and also that Asian students were widening the gap ahead of white students in both math and reading achievement. At the same time, Hispanic and Asian students who are English language learners (ELLs) are falling further behind white students in mathematics and reading achievement, and gaps between higher- and lower-income students persist, with some changes that vary by subject and grade. During the decade of stagnation, however, in keeping with trends in per-pupil investments over this period, these trends widened existing inequities. As National Center for Education Statistics (NCES) Associate Commissioner Peggy Carr soberly notes, “Compared to a decade ago, we see that lower-achieving students made score declines in all of the assessments, while higher-performing students made score gains” (Danilova 2018).

Finally, we have also seen marked changes in the student body composition that have implications for these trends going forward. The proportion of low-income students in U.S. schools has increased rapidly in recent decades, as has the share of students of color (NCES 2020e; Carnoy and García 2017). A student’s race/ethnicity and socioeconomic status also affects the student’s odds of ending up in a high-poverty school or a school with a high share of students of color. For example, Black and Hispanic students who are not poor are much more likely than white or Asian students who are low income to be enrolled in high-poverty schools (Carnoy and García 2017).

All of these changes point to the need for increased resources across the board, and especially in schools serving the highest-needs students. As we revisit education funding in the aftermath of the pandemic-induced recession, the new structure must make greater investments to ensure the equitable provision of education and associated supports not only in stable times but also in the context of substantial disruptions and crises (García and Weiss 2021). As the analysis above makes clear, neither equity not adequacy—and, thus, excellence in public education—will ever be possible as long as local revenues play such a central role, and as long as states are the primary vehicle to address those disparities. While we leave it to policymakers to design the specifics of this public-good investment, we emphasize that the benchmarks we should reach to determine that those investments are stable, sufficient, and equitable should reflect meaningful, consistent advances for the highest-poverty schools and schools serving students of color. In other words, when the impacts of recessions no longer fall on the backs of our most vulnerable children, we will know that we are moving in the right direction.

Public education funding could also be deployed quickly to boost the economy and serve as an automatic stabilizer

The practice of cutting school funding during recessions is not only bad for students and teachers but also hurts the economy overall. The education sector has the potential to help stabilize the economy during downturns, but historically, our policy responses have failed to provide the necessary investment, as discussed in this report.

Up to this point, we have shown the characteristics, dynamics, and consequences of the existing education funding system. We have emphasized that fixing the system’s problems and achieving an excellent, equitable, robust, and stable public education system requires more funding—not just a reshuffling of existing funding. We have presented evidence indicating the need for a significantly larger contribution to the system from the federal government on a permanent basis. We have also demonstrated that targeting additional funds to schools during the Great Recession—via ARRA funds in particular— helped offset the large cuts schools experienced due to state and local shortfalls. As stated by Evans, Schwab, and Wagner (2019), “[…] the federal government’s efforts to shield education from some of the worst effects of the recession achieved their major goal.” Based on the observed trends, we considered whether even more sustained federal investments would have better assisted the students, schools, and communities that suffered major setbacks due to the Great Recession.

There is another reason for both larger investments and a more robust federal role when state and local budgets experience shortfalls due to economic downturns: School funding can be part of the countercyclical public-spending programs that help the economy recover. While policymakers and economists have long recognized the need for, and the effectiveness of, such automatic stabilizers (programs that pump public spending into the economy just when overall spending is declining), they have not traditionally placed public education spending in this category—yet it belongs there. 20 Federal funding directed toward schools during and in the aftermath of economic downturns can further boost the economy, thereby jump-starting economic recoveries.

Stable, sufficient, and equitable education funding would give schools and districts the resources and flexibility to adapt to challenges that they need but have not had during the COVID-19 pandemic. Moreover, automatic stabilization of public education protects students and school systems against depleted school budgets during recessions and volatile business cycles (Evans, Schwab, and Wagner 2019; Allegretto, García, and Weiss 2021). In addition to averting the harms to students and teachers described above, countercyclical investments would keep the public education workforce employed. The teachers, nurses, counselors, librarians, bus drivers, cafeteria workers, and others who work in public schools made up 53.2% of all state and local public-sector workers in 2019—accounting for nearly 7.0% of total U.S. employment.21 School staff are also family and community members whose spending ripples through their local economies (known as the multiplier effect). Cuts to education revenues and employment thus also affect local communities more broadly, and retrenchment of spending acts as a type of reverse multiplier, resulting in a vicious downward cycle.

Federally provided countercyclical fiscal spending on public education set up to kick in based on defined triggers—akin to an expansion of unemployment benefits that kicks in when certain unemployment targets are reached—would have significant “bang-for-your-buck” multiplier effects. Such automatic spending constitutes smart investment that upholds public education while giving the overall economy a significant boost. Analyzing then President-elect Biden’s American Rescue Plan, which included public education spending, Zandi and Yaros (2021) reported a 1.34 fiscal multiplier for state and local government spending (the American Rescue Plan Act of 2021 was signed into law in March 2021).

Because the federal government already provides substantial support to state and local governments in such times, bolstering and further targeting that support in a defined and concerted manner would entail a relatively light lift. Despite some challenges, several programs of this nature have been shown to meet their goals in their given policy areas. For example, the federal unemployment insurance (UI) and food stamps (SNAP)22 programs are often cited as having demonstrably positive outcomes when the federal government increases their funding. Both have been heavily criticized for their structural flaws and lack of sufficient resource (Bivens et al. 2021). However, through prior recessions and the pandemic, data illustrate that UI and SNAP nonetheless prevented millions of people from falling into, or deeper into, poverty, as well as averted hunger and evictions. The CARES Act’s first allotment of the Economic Impact Payments and expanded UI benefits during the COVID-19 pandemic kept 13.2 million people out of poverty (Zipperer 2020).23 The Bureau of Economic Analysis broke out the effects of selected pandemic response programs on personal income, illustrating just how heavily Americans leaned on these benefits through the pandemic. In June 2020, UI payments accounted for 15.6% of all wages and salaries in the U.S (BEA 2020). By contrast, just prior to the pandemic UI benefits were negligible in comparison—just 0.27% of wages and salaries overall in February 2020.

We propose that policymakers create a program for funding education during downturns that is of adequate magnitude and provides immediate, sufficiently large, and sustained relief as needed.

In order to provide an immediate response, the system must have the capacity to adapt to emergencies; a key way to ensure that is to specify ahead of time the automatic triggers that prompt launching the contingency plans.24 To clarify, we are not suggesting that public education spending be treated exactly like food stamps or unemployment insurance benefits—i.e., that states amass reserves for a “rainy day” or that reserves be built up during nonrecessionary periods. Rather, we are pointing to the economic benefits of an education system that is robustly, stably, and consistently funded throughout economic ups and downs, ensuring that it also has the resources to withstand the downturns and the flexibility to adapt. And we are recommending that Congress establish a program that kicks in when needed, rather than waiting until a crisis and coming together to pass a large, responsive bill, which requires political negotiation and can thus take a lot of time.

Sufficiently large investments imply that the spending numbers are adequate to the size of the problem. As we have seen during the COVID-19 pandemic, the various public programs—even with all their flaws—have been critical to preventing a much worse disaster than the one we have experienced.25

Finally, regarding sustained assistance, it was clear that relief and recovery spending fell far short in response to the Great Recession and was cut off too soon; it took 6.2 years to recoup the jobs lost and nearly eight years for the unemployment rate to get back to its pre-recession rate of 5%. And unemployment rates for Black and Hispanic workers took much longer to return to pre-recession levels (Allegretto 2016). In education, as shown before, it was not until the 2014–2015 school year that districts’ per-student revenue, on average, recovered to 2007–2008 levels nationally—and recovery took even longer for high-poverty districts.

In sum, while the purpose of this study is not to offer guidance on how to best design a public education automatic-stabilization program, we do argue that such a program would help public education during downturns, and provide a boost to the overall economy. At later stages, proof-of-concept designs such as Medicaid and transportation grants, and some of the existing large-scale public programs already mentioned, could be a useful place to continue the discussion. Identifying best practices—in program design, financing, and implementation in the United States and elsewhere—would help to conceive a strategy.

Conclusions and next steps

For too long in this country, we have normalized the practice of underinvesting in education while expecting that schools would still function well (or at least moderately well). We have also accepted the disproportionate burden that economic recessions place on public schools and students. These norms are very costly—to individuals and to society—and they shortchange our country’s potential.

As the data and research show, this approach is backward. If we are to have a chance of providing all students in the United States with an excellent education we must  build a strong foundation—one with sufficient, adequate, and equitable funding of public schools in practice, not just in theory. Ensuring broad adequacy and equity will require increased federal investment (to more fully complement a system that relies heavily on nonfederal sources). Moreover, federal provisions that provide for automatic boosts to education spending during downturns is critical. Our education system can and should include a countercyclical designed to help stabilize the economy when it is contracting—benefiting schools and communities.

Were we to truly acknowledge the benefits, it would be hard to argue politically against making these investments a reality. Here again the data are edifying: Extensive research indicates that a stable and consistent funding system with a much higher level of investment would generate large economic and social returns.26

An increased federal investment to ensure sufficient, adequate, and equitable funding of public schools has an additional benefit: It could serve as another tool in our toolbox for faster, broader, and more equitable recoveries from recessions. Boosting school funding during downturns could boost the wider economy—and disproportionately benefit the low-income communities that tend to be hit hardest in hard economic times.

This proposal requires jettisoning the tendency to pit public policy areas against one another for resources, and to glamorize the purportedly efficient notion of “doing more with less.” The latter, often used to justify education budget cuts, actually entails a misguided denial of the need for resources and of the inevitable damage that ensues when those resources fall short—or fail to exist at all.

We are not arguing that increased access to federal resources alone will address all the issues outlined above. Simply throwing money at the goal of providing an excellent education equitably to all children won’t achieve it; we need to make the right investments.27

In addition, it is also important to distinguish funding from decision-making. While the federal government is best positioned to ensure broadly adequate and equitable education funding nationwide, it is not necessarily well suited to make decisions about policy, practice, and implementation. Evidence should guide how decision-making is allocated across the federal, state, and local levels.28

Advancing this proposal also requires that we dislodge the conversation from where it has been stuck for at least the past half-century—namely on whether the resources exist. They do. What we need to ask now is how to make those resources available, and how to deploy them to ensure that all students have the opportunities to learn, develop, and achieve their full potential—and that these opportunities are available during both ordinary and recessionary times.

About the authors

Sylvia Allegretto is a research associate with the Economic Policy Institute. She worked for 15 years at the Institute for Research on Labor and Employment at the University of California, Berkeley, where she co-founded the Center on Wage and Employment Dynamics (CWED). She received her Ph.D. in economics from the University of Colorado, Boulder.

Emma García is an economist specializing in the economics of education and education policy. She developed this study while she was at the Economic Policy Institute (2013-2021). She is now a senior researcher at the Learning Policy Institute. García received her Ph.D. in economics and education from Columbia University’s Teachers College.

Elaine Weiss is the Policy Director at the National Academy of Social Insurance, and former National Coordinator of the Broader, Bolder Approach to Education at the Economic Policy Institute (2011-2018). She received her B.A. in Political Science from the University of Maryland, J.D. from Harvard Law School, and Ph.D. in public policy from the George Washington University.

Acknowledgments

The authors are grateful to EPI Publications Director Lora Engdahl for having edited this report and for her help shepherding it to its release. The authors benefited from Ajay Srikanth’s guidance on school finance data sources at the beginning of the project. The authors appreciate EPI’s support of this project, EPI Research Assistant Daniel Perez for his assistance with the tables and figures, EPI Editor Krista Faries for her usual thoughtful insights, and EPI’s communications staff for their assistance with the production and dissemination of this study.

Appendix: Notes on the data sources and the analyses

We construct our own district-level longitudinal data set using information from three different sources:

1. the National Center for Education Statistics’s School District Finance Survey (F-33, Local Education Agency Finance Survey microdata from NCES 2007–2008 to 2017–2018 (NCES-LEAFS 2021)
2. the United States’ Census Bureau’s Small Area Income and Poverty Estimates (SAIPE) Program (for districts 2007–2018, from the Urban Institute’s Education Data Portal (Urban Institute 2021a)29
3. Stanford Education Data Archive (SEDA) Version 4.0 covariates file (Reardon et al. 2021).

The School District Finance Survey (F-33) is the source for revenues and expenditures for public elementary and secondary school districts in the country. The F-33 is a component of the Common Core of Data (CCD) and consists of local education agencies (LEA)-level finance data submitted annually to the U.S. Census Bureau by state education agencies (SEAs) in the 50 states and the District of Columbia. The entire universe of LEAs in each school year and in each state plus D.C. are included. The F-33 report includes the following types of school district finance data: revenue, current expenditure, and capital outlay expenditure totals; revenues by source; current expenditures by function and object; and revenues and current expenditures per pupil.

We use the annual data from 12 school years from 2006–2007 until 2017–2018 (the most recent available data at the time of development of this research was the data for 2017-2018, last accessed in March 2021 (NCES-LEAFS 2021) , see https://nces.ed.gov/ccd/files.asp#Fiscal:1,LevelId:5,Page:1 for updates).

We use the following variables from NCES CCD 2020:

• Total Revenue (TOTALREV)
• Total Federal Revenue (TFEDREV)
• Total State Revenue (TSTREV)
• Total Local Revenue (TLOCREV)
  • Local Rev – Property Taxes (T06)
• Fall Membership (V33 and MEMBERSCH if V33 is missing)
• Total Current Expenditures for Elementary/Secondary Education (TCURELSC)30

We calculate revenues (total and by source) and current expenditures in per-student terms.

For findings expressed “in constant 2019–2020 dollars,” all spending and revenue data are expressed in dollars corresponding with the 2019–2020 school year (average July–June as explained by NCES 2019), using the consumer price index from the Bureau of Labor Statistics (BLS-CPI 2021).

For findings involving states’ cost-of-living-adjusted (RPPs), we account for differences in the cost-of-living across states by using the Bureau of Economic Analysis’s (BEA’s) Regional Price Parities (BEA 2021).31

For analyzing metrics and outcomes by school poverty level, we link the school finance information with the poverty information.

Our preferred poverty data source is the United States Census Bureau’s Small Area Income and Poverty Estimates (SAIPE) Program for districts for school years spanning 2007–2018, which we collect from the Urban Institute’s Education Data Portal (Urban Institute 2021a). Census SAIPE district poverty data are available for the period 2007–2008 through 2017–2018 (U.S. Census Bureau 2021). The variable of interest is the poverty rate for children ages 5–17 in the district (ratio between poor children and total children in that age group). They are originally available as yearly data. To proxy for the school year (July–June) data, for a given school year, we take the average between the fall year-1 and the spring year.

We also use two other poverty data sources, which are linked to the F-33 data in sequential steps, for the following two purposes: (a) to offer sensitivity analyses of the results using alternative sources of data; and (b) to use the maximum number of observations possible, in cases in which some information is missing in one source but available in others.

Our second-preferred poverty data are SEDA’s shares of free and reduced price lunch eligible students in grades 3–8 in the districts (Reardon et al. 2021). This information is available in the covariates’ file, and it is available starting in school year 2008–2009 (which is least preferred because it is after the beginning of the Great Recession).32

As an additional source checked in our sensitivity analyses, we use the county-level information from the Census, available (by year) at: https://www.census.gov/programs-surveys/saipe/data/datasets.html (U.S. Census Bureau 2021). The information is equivalent to the district-level information, but at the county level. For this study, we use the data in the same manner (turning the year estimates into school-year equivalent estimates, etc.).

We perform the analyses using the different sources independently, plus one more in which we combine the three sources, when one is missing but the other is not (i.e., we define a poverty-all variable that “combines” sources: If Census’s SAIPE’s district poverty data are missing, SEDA’s district poverty data are used; for districts missing on both, Census’s SAIPE’s county poverty data are used).

In each case, we calculate the poverty quartiles each year by dividing the poverty variable(s) into four quartiles.33 Low-poverty districts are districts with a poverty rate for children ages 5–17 in the first quartile of the poverty distribution. Medium-low-poverty districts are districts with a poverty rate for children ages 5–17 in the second quartile of the poverty distribution. Medium-high-poverty districts are districts with a poverty rate for children ages 5–17 in the third quartile of the poverty distribution. High-poverty districts are districts with a poverty rate for children ages 5–17 in the fourth (top) quartile of the poverty distribution.

A note about analytic samples and weights: As the school finance variables of interest are in per-student terms, districts with nonmissing and nonzero numbers of students are kept in our sample. In our preferred sample, we also restrict the analyses to observations from districts serving elementary schools only, secondary schools only, or both,34 and to districts with charter information nonmissing. Results using the full number of observations (unrestricted sample) are available upon request.

A note about the final sensitivity analysis: Following the nature of F33 and the weights available in the surveys, our unit of analysis is the district, and we present unweighted averages across districts. Sensitivity analyses are also available using the student population in the district to compute weighted averages across the districts, upon request.

A note about methods: The analyses presented in this report are descriptive in nature. We are interested in providing a description of the trends in revenues and expenditures over time, by state, and by district poverty level. We produce updated estimates for the main school finance indicators and we look at trends in the main variables (per-student revenue and spending) during recessions to see the potential of a solid response from the system to respond, counter, and recover from economic recessions.

We conducted multiple sensitivity analyses in our attempt to verify that the data that we provide are not sensitive to data sources or data procedures, as well as to understand possible ways to further expand this research. Each data source offers significant advantages, but there is no source that can be used for all the purposes intended. Additionally, the evidence improves if we use multiple sources. We are confident the main findings hold and are not driven by extraneous factors. We do not use regression analyses in this version of the report.

Endnotes

1. In addition to the Department of Education, the Department of Health and Human Services, which funds the Head Start program for young children, and the Department of Agriculture, which funds the School Lunch (meals) Program are also part of the agencies that support programs or functions in education.

2. We use current expenditures instead of total expenditures when comparing education spending between states or across districts, as suggested by the agency that provides the data, the National Center for Education Statistics (NCES). This approach recognizes that current expenditures exclude expenditures for capital outlay, “which tend to have dramatic increases and decreases from year to year.” Also, “the current expenditures commonly reported are for public elementary and secondary education only. Many school districts also support community services, adult education, private education, and other programs, which are included in total expenditures. These programs and the extent to which they are funded by school districts vary greatly both across and within states and school districts.” See NCES 2008.

3. See the New Yorkers for Students’ Educational Rights backgrounder (NYSER n.d.) on Campaign for Fiscal Equity, Inc. (CFE) v. State of New York, 8 N.Y.3d 14 (2006) and Srikanth et al. 2020. Michael A. Rebell, one of the most prominent school funding litigators in the country, was co-counsel for the plaintiffs in CFE v. New York, a school funding “adequacy” lawsuit that claimed that the State of New York violated the constitutional rights of New York City students by failing to adequately fund the city’s public schools (NYSER n.d.). See also Sciarra and Dingerson 2021.

4. Since 2010, the Education Law Center (ELC), housed at Rutgers University, produces report cards that ask Is School Funding Fair? (using the data collected annually, some of which we use in our analyses below). To paraphrase their response, “Generally, no.” As the authors emphasize, “The hallmark of a fair school funding system is that it delivers more funding to educate students in high-poverty districts [since] states providing equal or less funding to high-poverty districts are shortchanging the students most in need and at risk of academic failure” (Farrie and Schiarra 2021).

5. Moreover, these wealth-based disparities are mirrored in and compounded by race/ethnicity-based gaps. The Education Trust uses data to report on disparities by both income/poverty level and race/ethnicity. As the Education Trust’s report on funding gaps in 2018 reveals, “School districts serving the largest populations of Black, Latino, or American Indian students receive roughly $1,800, or 13 percent, less per student in state and local funding than those serving the fewest students of color. This may seem like an insignificant amount, but it adds up. For a school district with 5,000 students, a gap of $1,800 per student means a shortage of $9 million per year” (Morgan and Amerikaner 2018, emphasis added).

6. Our peer Western nations view public schools as more of a national responsibility and provide resources accordingly. For example, Germany has a heavily state-based school system, France has a hybrid local–federal system in which the central government pays teachers’ salaries, and Finland’s national government takes virtually full responsibility for public education.

7. As a large study by Berry (2021) reveals, higher-income areas are taxed, on average, at just half the rate of their lower-income counterparts. Not only does this lead to structurally inequitable funding for schools, it exacts a harder toll on the residents who are least able to afford it—who pay double the taxes of their wealthier peers on much lower incomes. And, as Srikanth (2021) notes, “The study reveals structural racism at work.”

8. Funding for K–12 (21.5%) and higher education (9.4%) combined make up the largest segment of most state budgets. Spending on K–12 education alone is barely second in public budgets to public welfare spending (22.4%) (Urban Institute 2021b).

9. Bradbury (2021) explains that “the largest portion of state aid to local school districts is typically provided on a per-student basis through a ‘foundation,’ ‘power-equalizing,’ ‘flat grant,’ or ‘tiered’ program.…In addition, some states include cost adjustments in their formulas. Key attributes on which states base such cost adjustments are student poverty, English language facility, and special education or disability status.”

10. As part of his War on Poverty, which recognized the impacts of poverty on children’s well-being and the nation’s future, President Lyndon Johnson advanced the Elementary and Secondary Education Act (ESEA) in 1965. This flagship federal legislation, which has since been reauthorized multiple times and whose current iteration is the Every Student Succeeds Act, is designed principally to channel resources to schools serving low-income students. However, Title I, the largest section of ESEA, was never enough to make up for the inequities created by the local–state funding system (see Gamson, McDermott, and Reed 2015).

11. This pattern isn’t at all “inadvertent,” but is a built-in feature that is part of a pattern of systemic racism and related classism that merits attention in itself. See, for example Sosina and Weathers 2019.

12. For example, in 2018–2019, average teacher salaries ranged from less than $46,000 in Mississippi to roughly $86,000 in New York (NEA 2020). However, within New York (according to 2017 data), they ranged from as low as $55,976 in the low-income Finger Lakes region in the northern part of the state to nearly twice as high, $110,000, in the wealthiest Long Island districts (Malatras and Simons 2019).

13. We note that the Great Recession started as the 2007–2008 school year was underway, so we are using the term “pre-recession level” flexibly and assuming school budgets do not immediately respond to the economic recession.

14. See Leachman, Masterson, and Figueroa 2017; Leachman and Figueroa 2019; Baker 2018; and Allegretto 2020 for some more examples.

15. Note that we are not distinguishing here between the source of increased or decreased funding but focusing on total revenues and expenditures. Roy (2011) examined a redistributive school finance reform initiated by the state legislature in Michigan in the mid-’90s, called Proposal A. This reform, which eliminated local discretion over school spending by increasing state aid to the lowest-spending districts and limiting it in the highest-spending districts, reduced spending disparities between districts, and increased student performance (state test scores) in the lowest-spending districts, though it also had a negative effect on student performance in the highest-spending districts. For an analysis of state school finance reforms affecting Kansas (“block grant funding” that froze district revenue regardless of enrollment and reduced funding in districts where enrollment increased), see Rauscher 2020. See Biasi 2019 for an examination of the effect of equalizing revenues across public school districts on students’ intergenerational mobility; Biasi finds that equalization has a large effect on mobility of low-income students, with no significant changes for high-income students.

16. Note that these analyses are based on cross-sectional data.

17. This post-1990 period, often referred to as the “adequacy era,” represented a time in which state-court decisions in multiple states resulted in increased public-school funding, offering an opportunity for researchers to study the overall impacts of these substantial increases and to compare them to student outcomes in states that did not experience them.

18. Their preferred estimates, based on the gradient of student achievement with respect to district income, indicate that a school funding reform raises achievement in a district with log average income one point below the state mean, relative to a district at the mean, by 0.1 standard deviations after 10 years.

19. High-poverty schools found it more difficult to fill vacancies than did low-poverty schools and schools overall, and high-poverty schools experienced higher turnover and attrition rates than did low-poverty schools (García and Weiss 2019).

20. Note that in this report, our main goal is to document the need and concept for such a program, not to discuss how best to design a public education automatic-stabilization program. These considerations, including specifically raising federal supports to education, have been discussed before (Boushey, Nunn, and Shambaugh 2019; Partelow, Yin and Sargrad 2020; Ogletree et al. 2017; Sahm 2019; Schott Foundation 2022; U.S. Department of Education 2013; Washington Center for Equitable Growth 2021; etc.).

21. Author Sylvia Allegretto’s analysis based on Bureau of Labor Statistics Current Employment Statistics data for 2019 (BLS-CES 2021). Education is one of the largest single components of government spending, amassing 7.3% of GDP across federal, state, and local expenditures (OECD 2013).

22. SNAP is the abbreviation for the Supplemental Nutrition Assistance Program, also known as “food stamps.”

23. Data Household Pulse Survey (HHPS) from the U.S. Census Bureau found that 29.3% of respondents with children were food insecure in the week of April 23–July 21, 2020 (Schanzenbach and Tomeh 2020). Bauer (2020) estimates that there were almost 14 million children living in a household characterized by child food insecurity during the week of June 19–23, 2020, “5.6 times as many as in all of 2018 (2.5 million) and 2.7 times as many as during [the] peak of the Great Recession in 2008 (5.1 million).” Typically, these programs disproportionately benefit low-income communities, which are often hit the hardest, thus preventing even more damage and the exacerbation of the large existing inequities.

24. The term “contingency plans” comes from the education-in-emergencies field and is mostly applicable to international contexts, but it has also been used in the U.S. to give broader responses to crises such as Hurricane Katrina (The White House 2006). See García and Weiss 2020, 2021 for more details. The term “automatic trigger” is used to indicate what activates benefits or programs. See Mitchell and Husak 2021 and Boushey, Nunn, and Shambaugh 2019.

25. For flaws around one of those programs—unemployment insurance—see Bivens et al. 2021. Bitler, Hoynes, and Schanzenbach 2020 provide evidence for three reasons why the policy response left needs unmet: “(1) timing—relief came with a substantial delay (due to overwhelmed UI systems/need to implement new programs); (2) magnitude—payments outside UI are modest; and (3) coverage gaps—access is lower for some groups and other groups are statutorily excluded.”

26. See section summarizing the literature on the impacts of spending on education above.

27. We have discussed this point extensively in our other research on early childhood education, socio-emotional learning, and integrated student support, among others. See García 2015; García and Weiss 2017; García and Weiss 2016; Weiss and Reville 2019, among others, for guidance on smart education investments. See also Bryk et al. 2010 for a discussion on the role of context and how even after receiving funding, schools did not improve, and offering suggestions for school reform efforts.

28. California, which revamped the state’s education funding and accountability systems in the wake of the 2015 passage of the Every Student Succeeds Act, offers a valuable model. See Furger, Hernández, and Darling-Hammond 2019 and Johnson and Tanner 2018.

29. For counties 2007–2019, see U.S. Census Bureau 2021.

30. As explained earlier in the report, we use current expenditures instead of total expenditures when comparing education spending between states or across districts, as suggested by the agency that provides the data, the National Center for Education Statistics (NCES). This approach recognizes that current expenditures exclude expenditures for capital outlay, “which tend to have dramatic increases and decreases from year to year.” Also, “the current expenditures commonly reported are for public elementary and secondary education only. Many school districts also support community services, adult education, private education, and other programs, which are included in total expenditures. These programs and the extent to which they are funded by school districts vary greatly both across and within states and school districts.” See NCES 2008.

31. For 2018: https://www.bea.gov/news/2020/real-personal-income-state-and-metropolitan-area-2018, and For Time Series: https://apps.bea.gov/regional/histdata/releases/0920rpi/SARPP.zip

32. Note that we obtain the minority concentration from this source. Not used in this report.

33. Variables with the poverty quartiles are called POV_CDIST (our preferred Census SAIPE district) and povall (the one combining all sources).

34. Excluded are districts of vocational or special education system; nonoperating school system that exists for administrative purposes only and does not operate its own schools; LEAs that closed shortly before the start of the fiscal year or are scheduled to open in a future fiscal year but still reported revenue or expenditure information for the current fiscal year; and education service agency (ESA) (variable labeled schlev).

References

Allegretto, Sylvia A. 2016. California’s Labor Market: Eight Years Post-Great-Recession. Center on Wage and Employment Dynamics of the Institute for Research on Labor and Employment, University of California, Berkeley. May 2016.

Allegretto, Sylvia A. 2020. “The Critical Issues of Teacher Pay and Employment.” In Strike for the Common Good: Fighting for the Future of Public Education, edited by Rebecca Kolins Givan(1975– and Amy Schrager Lang. Ann Arbor: University of Michigan Press.

Allegretto, Sylvia A, Emma García, and Elaine Weiss. 2021. “Policymakers Cannot Relegate Another Generation to Underresourced K–12 Education Because of an Economic Recession,” Working Economics Blog, Economic Policy Institute, July 14, 2021.

Anderson, Allison. 2020. “COVID-19 Outbreak Highlights Critical Gaps in School Emergency Preparedness.” Education Plus Development (Brookings Institution blog), March 11, 2020.

Baker, Bruce D. 2018. Educational Inequality and School Finance: Why Money Matters for America’s Students. Cambridge, Mass.: Harvard Education Press.

Baker, Bruce, and Sean Corcoran. 2012. The Stealth Inequities of School Funding: How State and Local School Finance Systems Perpetuate Inequitable Student Spending. Center for American Progress, September 2012.

Baker, Bruce D., Danielle Farrie, and David G. Sciarra. 2016. Mind the Gap: 20 Years of Progress and Retrenchment in School Funding and Achievement Gaps. Educational Testing Service Research Report No. RR-16-15. https://doi.org/10.1002/ets2.12098.

Baker, Bruce D., Matthew Di Carlo, and Mark Weber. 2020. The Adequacy and Fairness of State School Finance Systems: Key Findings from the School Finance Indicators Database, Second Edition. Albert Shanker Institute and Rutgers Graduate School of Education, February 2020.

Bauer, Lauren. 2020. “About 14 Million Children in the US Are Not Getting Enough to Eat.” The Brookings Institution, July 9, 2020.

Bayer, Patrick, Peter Q. Blair, and Kenneth Whaley. 2020. “A National Study of School Spending and House Prices.” National Bureau of Economic Research Working Paper no. 28255, December 2020.

Behrman, Elizabeth. 2019. “Landmark Lawsuit Challenges How Pennsylvania Funds Its Public Schools.” Pittsburgh Post-Gazette. September 3, 2019.

Berry, Barnett, and Patrick M. Shields. 2017. “Solving the Teacher Shortage: Revisiting the Lessons We’ve Learned.” Phi Delta Kappan 98, no. 8: 8–18, https://doi.org/10.1177/0031721717708289.

Berry, Christopher. 2021. “Reassessing the Property Tax.” The University of Chicago Harris School of Public Policy and the College, March 1, 2021.

Biasi, Barbara. 2019: “School Finance Equalization and Intergenerational Mobility: A Simulated Instruments Approach.” National Bureau of Economic Research Working Paper no. 25600, October 2019.

Bivens, Josh, Melissa Boteach, Rachel Deutsch, Francisco Diez, Rebecca Dixon, Brian Galle, Alix Gould-Werth, Nicole Marquez, Lily Roberts, Heidi Shierholz, William Spriggs, and Andrew Stettner. 2021. Reforming Unemployment Insurance: Stabilizing a System in Crisis and Laying the Foundation for Equity. A joint report of the Center for American Progress, Center for Popular Democracy, Economic Policy Institute, Groundwork Collaborative, National Employment Law Project, National Women’s Law Center, and Washington Center for Equitable Growth, June 2021.

Bitler, Marianne, Hilary W. Hoynes, and Diane Whitmore Schanzenbach. 2020. “The Social Safety Net in the Wake of COVID-19.” National Bureau of Economic Research Working Paper no. 27796, September 2020.

Boushey, Heather, Ryan Nunn, and Jay Shambaugh, eds. 2019. Recession Ready: Fiscal Policies to Stabilize the American Economy. Washington Center for Equitable Growth and The Brookings Institution’s Hamilton Project, May 2019.

Bradbury, Katharine. 2021. “The Roles of State Aid and Local Conditions in Elementary School Test-Score Gaps.” Federal Reserve of Boston Research Department Working Papers, No. 21-2.

Bryk, Anthony S., Penny Bender Sebring, Elaine Allensworth, John Q. Easton, and Stuart Luppescu. 2010. Organizing Schools for Improvement: Lessons from Chicago. Chicago: University of Chicago Press.

Bureau of Economic Analysis (BEA). 2020. “Effects of Selected Federal Pandemic Programs on Personal Income, May 2020” (data table). Released June 26, 2020.

Bureau of Economic Analysis (BEA). 2021. “Real Personal Income, Per Capita Personal Income, and Regional Price Parities: State.” From Real Personal Income for States and Metropolitan Areas: 2018. Last accessed April 18, 2021.

Bureau of Labor Statistics, Consumer Price Index (BLS-CPI). 2021. CPI-U-RS public data series. Various years.

Bureau of Labor Statistics, Current Employment Statistics (BLS-CES). 2021. Employment, Hours, and Earnings (database). Accessed May 11, 2021.

Candelaria, Christopher A., and Kenneth A. Shores. 2019. “Court-Ordered Finance Reforms in the Adequacy Era: Heterogeneous Causal Effects and Sensitivity.” Education Finance and Policy 14, no. 1: 31–60. https://doi.org/10.1162/edfp_a_00236.

Carnoy, Martin, and Emma García. 2017. Five Key Trends in U.S. Student Performance: Progress by Blacks and Hispanics, the Takeoff of Asians, the Stall of Non-English Speakers, the Persistence of Socioeconomic Gaps, and the Damaging Effect of Highly Segregated Schools. Economic Policy Institute, January 2017.

Caucutt, Elizabeth M., Lance Lochner, Joseph Mullins, and Youngmin Park. 2020. Child Skill Production: Accounting for Parental and Market-Based Time and Goods Investments. National Bureau of Economic Research Working Paper no. 27838, https://doi.org/ 10.3386/w27838.

Danilova, Maria. 2018. “Fourth-graders Making No Improvements in Math or Reading, Study Says.” USA Today, April 10, 2018.

Duncan, Greg J., and Richard. J. Murnane. 2016. “Rising Inequality in Family Incomes and Children’s Educational Outcomes.” RSF: The Russell Sage Foundation Journal of the Social Sciences 2, no. 2, 142–158.

Chingos, Matthew M., Sandy Kress, Matthew Ladner, Susanna Loeb, Paul E. Peterson, Michael J. Petrilli, and Morgan S. Polikoff. 2019. “What to Make of the 2019 Results from the ‘Nation’s Report Card.’” Education Next. October 30, 2019.

Evans, William N., Robert M. Schwab, and Kathryn L. Wagner. 2019. “The Great Recession and Public Education.” Education Finance and Policy 2019, vol. 14, no. 2: 298–326. doi: https://doi.org/10.1162/edfp_a_00245

Farrie, Danielle, and David G. Sciarra. 2021. Making the Grade 2020: How Fair Is School Funding in Your State? Education Law Center. January 2021.

Filardo, Mary, Jeffrey M. Vincent, and Kevin J. Sullivan. 2019. “How Crumbling School Facilities Perpetuate Inequality.” Phi Delta Kappan, April 29, 2019.

Furger, Roberta. C., Laura E. Hernández, and Linda Darling-Hammond. 2019. The California Way: The Golden State’s Quest to Build an Equitable and Excellent Education System. Learning Policy Institute. February 2019.

Gamson, David A., Kathryn A. McDermott, and Douglas S. Reed. 2015. “The Elementary and Secondary Education Act at Fifty: Aspirations, Effects, and Limitations.” Russell Sage Foundation Journal of the Social Sciences 1, no. 3 (December 2015).

García, Emma. 2015. Inequalities at the Starting Gate: Cognitive and Noncognitive Skills Gaps Between 2010–2011 Kindergarten Classmates. Economic Policy Institute, June 2015.

García, Emma, and Elaine Weiss. 2016. Making Whole-Child Education the Norm: How Research and Policy Initiatives Can Make Social and Emotional Skills a Focal Point of Children’s Education. Economic Policy Institute, August 2016.

García, Emma, and Elaine Weiss. 2017. Education Inequalities at the School Starting Gate: Gaps, Trends, and Strategies to Address Them. Economic Policy Institute, September 2017.

García, Emma, and Elaine Weiss. 2019. U.S. Schools Struggle to Hire and Retain Teachers: The Second Report in the ‘Perfect Storm in the Teacher Labor Market’ Series. Economic Policy Institute, April 2019.

García, Emma, and Elaine Weiss. 2020. COVID-19 and Student Performance, Equity, and U.S. Education Policy: Lessons from Pre-pandemic Research to Inform Relief, Recovery, and Rebuilding. Economic Policy Institute, September 2020.

García, Emma, and Elaine Weiss. 2021. “Learning During the Pandemic: Lessons from the Research on Education in Emergencies for COVID-19 and Afterwards.” Working Economics Blog (Economic Policy Institute), February 18, 2021.

Gibbons, S., Sandra McNally, and Martina Viarengo. 2018. “Does Additional Spending Help Urban Schools? An Evaluation Using Boundary Discontinuities.” Journal of the European Economic Association 16, no. 5, 1618–1668.

Gould, Elise. 2017. “Local Public Education Employment May Have Weathered Recent Storms, but Schools Are Still Short 327,000 Public Educators” (economic snapshot). Economic Policy Institute, October 6, 2017.

Gould, Elise. 2019. “Back-to-School Jobs Report Shows a Continued Shortfall in Public Education Jobs” (economic snapshot). Economic Policy Institute, October 10, 2019.

Greaves, Ellen, and Luke Sibieta. 2019. “Constrained Optimisation? Teacher Salaries, School Resources and Student Achievement.” Economics of Education Review 73, https://doi.org/10.1016/j.econedurev.2019.101924.

Griffith, Michael. 2021. “An Unparalleled Investment in U.S. Public Education: Analysis of the American Rescue Plan Act of 2021.” Learning in the Time of COVID-19 (Learning Policy Institute blog series), March 11, 2021.

Hussar, W.J., and T.M. Bailey. 2020. Projections of Education Statistics to 2028 (NCES 2020-024). U.S. Department of Education, National Center for Education Statistics, May 2020. 

Hyman, J. 2017. “Does Money Matter in the Long Run? Effects of School Spending on Educational Attainment.” American Economic Journal: Economic Policy 9, no. 4: 256–80. 

Jackson, C. Kirabo. 2018. “Does School Spending Matter?: The New Literature on an Old Question.” National Bureau of Economic Research Working Paper no. 25368. December 2018.

Jackson, C. Kirabo, and Claire Mackevicius. 2021. “The Distribution of School Spending Impacts.” National Bureau of Economic Research Working Paper no. 28517, February 2021. https://doi.org/10.3386/w28517. 

Jackson, C. Kirabo, Cora Wigger, and Heyu Xiong. 2020. “Do School Spending Cuts Matter? Evidence from the Great Recession.” American Economic Journal: Economic Policy 13, no. 2: 304–335. https://doi.org/10.1257/pol.20180674.

Jackson, C. Kirabo, Rucker C. Johnson, and Claudia Persico. 2016. “The Effects of School Spending on Educational and Economic Outcomes: Evidence from School Finance Reforms.” Quarterly Journal of Economics 131, no. 1: 157–218. https://doi.org/10.1093/qje/qjv036.

Johnson, Rucker C., and Sean Tanner. 2018. Money and Freedom: The Impact of California’s School Finance Reform (research brief). Palo Alto, Calif.: Learning Policy Institute, February 2018..

Jordan, Phyllis W. 2021. “What Congressional Covid Funding Means for K–12 Schools.” FutureEd (McCourt School of Public Policy, Georgetown University). August 24, 2021.

Kornrich, Sabino. 2016. “Inequalities in Parental Spending on Young Children: 1972 to 2010.” AERA Open, vol. 2, no. 2: 1–12. https://doi.org/10.1177/2332858416644180.

Lafortune, Julien, Jesse Rothstein, and Diane Whitmore Schanzenbach. 2018. “School Finance Reform and the Distribution of Student Achievement.” American Economic Journal: Applied Economics 10, no. 2: 1–26. https://doi.org/10.1257/app.20160567.

Leachman, Michael, and Eric Figueroa. 2019. K–12 School Funding Up in Most 2018 Teacher-Protest States, But Still Well Below Decade Ago. Center on Budget and Policy Priorities, March 2019.

Leachman, Michael, Kathleen Masterson, and Eric Figueroa. 2017. A Punishing Decade for School Funding. Center on Budget and Policy Priorities, November 2017.

Malatras, Jim, and Nicholas Simons. 2019. “A Preliminary Analysis of Teacher Salaries in New York by Region and Wealth.” Rockefeller Institute of Government, April 17, 2019.

Mishel, Lawrence, and Richard Rothstein. 2003. The Class Size Debate. Washington, D.C.: Economic Policy Institute.

Mitchell, David, and Corey Husak. 2021. How to Replace COVID Relief Deadlines with Automatic ‘Triggers’ That Meet the Needs of the U.S. Economy. Washington Center for Equitable Growth, February 2021.

Morgan, Ivy, and Ary Amerikaner. 2018. “An Analysis of School Funding Equity Across the U.S. and Within Each State.” Education Trust, February 27, 2018.

National Assessment Governing Board (NAGB). 2019. “Reading Scores on the Nation’s Report Card Lower Than in 2017, While Math Results Vary” (news release). October 30, 2019.

National Center for Education Statistics (NCES), U.S. Department of Education. 2008. “Appendix A: Methodology and Technical Notes,” in Revenues and Expenditures for Public Elementary and Secondary Education, School Year 2005–06 (Fiscal Year 2006). Institute of Education Sciences, U.S. Department of Education, NCES 2008-328.

National Center for Education Statistics (NCES), U.S. Department of Education. 2019. “Table 106.70. Gross Domestic Product Price index, Consumer Price Index, Education Price Indexes, and Federal Budget Composite Deflator: Selected Years, 1919 Through 2019.” Digest of Education Statistics: 2019, Institute of Education Sciences, U.S. Department of Education. Table prepared in February 2020.

National Center for Education Statistics (NCES), U.S. Department of Education. 2020a. “Table 235.10. Revenues for Public Elementary and Secondary Schools, by Source of Funds: Selected Years, 1919–20 Through 2017–18.” Digest of Education Statistics: 2020, Institute of Education Sciences, U.S. Department of Education. Table prepared in August 2020.

National Center for Education Statistics (NCES), U.S. Department of Education. 2020b. “Table 235.20. Revenues for Public Elementary and Secondary Schools, by Source of Funds and State or Jurisdiction: 2017–18.” Digest of Education Statistics: 2020, Institute of Education Sciences, U.S. Department of Education. Table prepared in August 2020.

National Center for Education Statistics (NCES), U.S. Department of Education. 2020c. “Table 236.25. Current Expenditures for Public Elementary and Secondary Education, by State or Jurisdiction: Selected Years, 1969–70 Through 2017–18.” Digest of Education Statistics: 2020, Institute of Education Sciences, U.S. Department of Education. Table prepared in November 2020.

National Center for Education Statistics (NCES), U.S. Department of Education. 2020d. “Table 208.40: Public Elementary and Secondary Teachers, Enrollment, and Pupil/Teacher Ratios, by State or Jurisdiction: Selected Years, Fall 2000 Through Fall 2018.” Digest of Education Statistics: 2020, Institute of Education Sciences, U.S. Department of Education. Table prepared in August 2020.

National Center for Education Statistics (NCES), U.S. Department of Education. 2020e. “Table 203.50. Enrollment and Percentage Distribution of Enrollment in Public Elementary and Secondary Schools, by Race/Ethnicity and Region: Selected Years, Fall 1995 Through Fall 2029.” Digest of Education Statistics: 2020, Institute of Education Sciences, U.S. Department of Education. Table prepared in September 2020.

National Center for Education Statistics, Common Core of Data (NCES CCD), U.S. Department of Education. 2020. Documentation for the NCES Common Core of Data School District Finance Survey (F-33), School Year 2017–18 (Fiscal Year 2018). Institute of Education Sciences, U.S. Department of Education, NCES 2020-309.

National Center for Education Statistics, Local Education Agency Finance Survey (NCES-LEAFS). 2021. 2007–2008 to 2017–2018 microdata from the Local Education Agency Finance Survey (F-33). NCES, U.S. Department of Education. Accessed March 2021.

National Education Association (NEA). 2020. Rankings of the States 2019 and Estimates of School Statistics 2020. July 2020. Downloaded from https://www.nea.org/research-publications.

New Yorkers for Students’ Educational Rights (NYSER). n.d. “Background: CFE v. State of New York” (web page). Accessed July 2022.

Ogletree, Charles J., Jr., Kimberly Jenkins Robinson, Alfred A. Lindseth, Rocco E. Testani, and Lee A. Peifer. 2017. “Rodriguez Reconsidered: Is There a Federal Constitutional Right to Education?” Education Next, Spring 2017.

Organisation for Economic Co-operation and Development (OECD). 2013. Education at a Glance 2013: OECD Indicators. OECD Publishing/.http://dx.doi.org/10.1787/eag-2013-en.

Özek, Umut. 2020. “Examining the Educational Spillover Effects of Severe Natural Disasters: The Case of Hurricane Maria.” American Institutes for Research/Calder Working Paper no. 233-0320, March 2020.

Partelow, Lisette, Jessica Yin, and Scott Sargrad. 2020. Why K–12 Education Needs More Federal Stimulus Funding. Center for American Progress, July 2020.

Raikes, Jeff, and Linda Darling-Hammond. 2019. “Why Our Education Funding Systems Are Derailing the American Dream.” Money Matters (Learning Policy Institute blog series), February 18, 2019.

Rauscher, Emily. 2020. “Does Money Matter More in the Country? Education Funding Reductions and Achievement in Kansas, 2010–2018.” AERA Open 6, no. 4. https://doi.org/10.1177/2332858420963685.

Reardon, Sean F. 2011. “The Widening Academic Achievement Gap between the Rich and the Poor: New Evidence and Possible Explanations.” In Whither Opportunity?: Rising Inequality, Schools, and Children’s Life Chances, edited by Greg J. Duncan and Richard Murnane. New York: Russell Sage Foundation.

Reardon, S.F., A.D. Ho, B.R. Shear, E.M. Fahle, D. Kalogrides, H. Jang, H., and B. Chavez. 2021. Stanford Education Data Archive (Version 4.0). Retrieved from http://purl.stanford.edu/db586ns4974.

Roy, Joydeep. 2011. “Impact of School Finance Reform on Resource Equalization and Academic Performance: Evidence from Michigan.” Education Finance and Policy 6, no. 2: 137–167. http://www.jstor.org/stable/educfinapoli.6.2.137.

Sahm, Claudia. 2019. “Direct Stimulus Payments to Individuals.” In Recession Ready: Fiscal Policies to Stimulate the American Economy, edited by Heather Boushey, Ryan Nunn, and Jay Shambaugh. Washington, D.C., The Hamilton Project, the Washington Center for Equitable Growth, and the Brookings Institution.

Schanzenbach, Diane Whitmore. 2020. “The Economics of Class Size.” In The Economics of Education (Second Edition), A Comprehensive Overview, edited by Steve Bradley and Colin Green, 321–331. Chicago: University of Chicago Press. https://doi.org/10.1016/B978-0-12-815391-8.00023-9.

Schanzenbach, Diane Whitmore, and Natalie Tomeh. 2020. “Visualizing Food Insecurity: App Offers Snapshot of Weekly National and State-by-State Averages.” Institute For Policy Research, Northwestern University, July 14, 2020.

Schneider, Daniel, Orestes P. Hastings, and Joe LaBriola. 2018. “Income Inequality and Class Divides in Parental Investments.” American Sociological Review 83, no. 3, 475–507. https://doi.org/10.1177/0003122418772034.

Sciarra, David, and Leigh Dingerson. 2021. From Courthouse to Statehouse—and Back Again: The Role of Litigation in School Funding Reform. Education Law Center, February 2021.

Schott Foundation for Public Education. 2022. “Equity and Excellence Commission” (web page). Accessed May 18, 2022.

Shores, Kenneth, and Steinberg, Matthew. 2017. “The Impact of the Great Recession on Student Achievement: Evidence from Population Data.” August 28, 2017. https://doi.org/10.2139/ssrn.3026151.

Sosina, Victoria E., and Ericka S. Weathers. 2019. “Pathways to Inequality: Between-District Segregation and Racial Disparities in School District Expenditures.” AERA Open 5, no. 3. https://doi.org/10.1177/2332858419872445.

Srikanth, Anagha. 2021. “Huge New Study Shows Homes in Poor Areas Are Taxed at Twice the Rate as Rich Neighborhoods.” The Hill, March 12, 2021.

Srikanth, Ajay, Michael Atzbi, Bruce D. Baker, and Mark Weber. 2020. How States Fund Education: The Oxford Handbook of U.S. Education Law. Edited by Kristine L. Bowman. New York and Oxford, UK: Oxford University Press. https://doi.org/10.1093/oxfordhb/9780190697402.013.10.

Turner, Corey, Reema Khrais, Tim Lloyd, Alexandra Olgin, Laura Isensee, Becky Vevea, and Dan Carsen. 2016. “Why America’s Schools Have a Money Problem.” NPR’s Morning Edition, April 18, 2016.

U.S. Census Bureau. 2021. “Small Area Income and Poverty Estimates Program (SAIPE) State and County Estimates,” 2007–2018. From SAIPE Datasets. Last accessed April 18, 2021.

U.S. Department of Education. 2013. For Each and Every Child—A Strategy for Education Equity and Excellence. The Equity and Excellence Commission, February 2013.

Urban Institute. 2021a. “Small Area Income and Poverty Estimates Program (SAIPE) School District Estimates,” 2007–2018. From the Urban Institute Education Data Portal (Version 0.12.0), made available under the ODC Attribution License. Last accessed April 18, 2021.

Urban Institute. 2021b. “State and Local Financial Expenditures, Fiscal Year 2018.” Accessed August 2021.

Washington Center for Equitable Growth. 2021. Executive Action to Coordinate Federal Countercyclical Regulatory Policy (fact sheet). February 2021.

Weiss, Elaine, and Paul Reville. 2019. Broader, Bolder, Better: How Schools and Communities Help Students Overcome the Disadvantages of Poverty. Cambridge, Mass.: Harvard Education Publishing Group.

The White House. 2006. The Federal Response to Hurricane Katrina: Lessons Learned. February 2006.

Zandi, Mark, and Bernard Yaros Jr. 2021. The Biden Fiscal Rescue Package: Light on the Horizon. Moody’s Analytics, January 15, 2021.

Zipperer, Ben. 2020. “Over 13 Million More People Would Be in Poverty Without Unemployment Insurance and Stimulus Payments.” Working Economics Blog (Economic Policy Institute), September 17, 2020.

What this report finds: The U.S. unemployment insurance (UI) system has historically underperformed as a macroeconomic stabilizer. While UI, like other automatic stabilizers, is designed to automatically spur aggregate demand when private spending falls (in UI’s case by temporarily replacing some lost wages of jobless workers), the boost is weaker than it could be. The UI system’s fuller potential was highlighted by the extraordinarily large but temporary UI expansions enacted by Congress during the COVID-19 pandemic, which made more workers eligible for benefits, raised benefit levels, and lengthened the duration of benefits. With these expansions, UI benefits as a share of wage and salary income provided an economic boost roughly four times as great during the pandemic as during any previous recession.

Why it matters: Weak automatic stabilizers mean that recessions last longer and inflict more damage than they need to—unless Congress and the president act nimbly and in concert to pass discretionary relief. Even then, the discretionary programs end earlier than they should. Consider for example the UI expansions enacted during the Great Recession that were turned off in 2014— well before a full recovery had taken hold. The less that American families have to rely on ad hoc relief offered only when there is political comity, the better it is for their economic security. As the pandemic UI programs showed, more forceful UI interventions are possible during recessions. If these expansions were set on autopilot, then future recessions would be shorter and less painful, and recoveries would come more quickly.

What we can do about it: Make UI a more powerful macroeconomic stabilizer by enacting reforms along three key dimensions or margins: eligibility, duration, and benefit levels. For example, program parameters could be strengthened to ensure that a larger share of unemployed workers are eligible for benefits, that benefits last long enough to bridge a jobless spell, and that benefits replace a high-enough share of previous earnings to minimize hardship.

The unemployment insurance (UI) system provides critical support during economic downturns, with cash benefits bolstering both the incomes of working people who have lost jobs as well as a flagging macroeconomy (Bivens et al. 2021; Hickey 2021). Signed into law as part of the Social Security Act in 1935 during the Great Depression, the system has historically been one the first lines of response to a downturn, providing immediate financial relief to households whose spending helps stabilize the economy  by boosting economywide consumer spending.

However, weaknesses in the UI system have limited its effectiveness as an automatic stabilizer relative to its potential. Automatic stabilizers are parts of the federal budget—either spending increases or tax cuts—that boost aggregate demand when private spending falls even with no change in legislation. Optimal stabilizers trigger on in a timely fashion as private spending begins slowing, provide a larger boost to aggregate demand as private spending falls further, and only begin ramping down as private spending begins recovering. Today’s UI system is not automatic enough.  

The UI system also has serious flaws as a social safety net program, including troubling racial disparities in recipiency, stringent work requirements, and more. The focus of this report, however, will be UI’s potential as a macroeconomic stabilizer during downturns. In this paper, we highlight three aspects of the UI system that can be augmented to make the system a more-effective macroeconomic stabilizer. Specifically, these areas where—or margins along which— the UI system’s stabilizing effects can be enhanced are the duration of UI benefits (how many weeks benefits last), the generosity of UI benefits (the benefit level), and the eligibility of UI benefits (which occupations or classes of workers can get benefits). We use evidence from the response to the COVID-19 pandemic—when Congress enacted temporary emergency measures that significantly raised benefit amounts, added additional weeks of benefits, and extended eligibility to a much greater share of workers—to show the macroeconomic benefits of permanent, versus ad hoc, expansions to UI. Our main findings are:

• The muted UI response to economic downturns before the COVID-19 shock show that it has long underperformed its potential as a macroeconomic stabilizer, due to short duration, low generosity, and limited eligibility.

• The emergency extended UI benefits that Congress provided in response to the pandemic provided a far larger boost to personal income during the COVID-19 crisis than any previous recession—probably ever, but certainly since personal income data began being systematically collected in 1960. This large boost to personal income meant UI had a far larger effect as a macroeconomic stimulus in 2020. It has the potential to do so again in future downturns.
  • UI as a share of personal income was four times as high in the year after the 2020 recession than the year following 2007–2009 Great Recession.

• • UI as a share of total wage and salary income—a different measure than personal income— reached 13% in 2020, compared with just 2.5% in the aftermath of the Great Recession in 2010.

• Pandemic UI programs—most notably Pandemic Unemployment Assistance (PUA), extending eligibility to workers who previously could not receive UI, and Pandemic Unemployment Compensation (PUC), providing an additional $600 per week on top of existing UI benefits—met the urgent need and filled gaps traditional state UI could not meet.
  • Traditional state UI made up just 20% of all UI by June 2021
  • At its height in the summer of 2020, the PUA program covered nearly 15 million workers who accounted for half of all UI claimants.
  • PUA and other pandemic UI programs were transferring more than $60 billion into personal incomes per percentage of unemployment within a few months after the March 2020 passage of the CARES Act.
• A key barrier to structurally reforming UI to make it a more-powerful macroeconomic stabilizer is that the need for reform is most apparent when the reforms look most expensive in terms of how the Congressional Budget Office (CBO) would score them. In downturns there is more need, which costs more, which makes reforms seem costly if undertaken during recessions. If this skewed cost score prevents policymakers from taking action now, they should consider these reforms when the overall economy begins a strong recovery.

Background on UI—and potential margins of improvement to UI as a macroeconomic stabilizer

The UI program—funded by states and the federal government and mostly administered by states— serves as both social insurance and a macroeconomic stabilizer. In addition to providing immediate relief to struggling households in the form of cash benefits covering a share of lost income, unemployment insurance stimulates a contracting economy by providing unemployed workers with benefits income they can spend in their communities and the broader economy. However, the macroeconomic stabilization function— in other words, the ability for UI to cushion against recessions and spur faster recoveries—has never lived up to its potential.

Since its inception in 1935, the UI system has had clear shortcomings in its ability to deliver robust macroeconomic stabilization in the face of an economic downturn. These shortcomings, which largely fall under three key margins of UI coverage, are insufficient duration of benefits (too few weeks of benefits), inadequate generosity of benefits (low benefits amounts), and limited eligibility (key worker occupations and classes are excluded from benefits). While the federal government sets some basic parameters for the program, state governments are in charge of the details, such as how long benefits last, the benefit amount, and the kind of work history people must prove to claim benefits. Benefits duration and generosity have proven inadequate during normal economic times and fail to automatically ramp up sufficiently during downturns.

Over the decades following its inception, legislative fixes at the federal level tinkered with these gaps in the UI system and offered some improvements. For example, previously excluded workers, most notably agricultural and domestic workers, were finally included. The longtime exclusion of farm and domestic workers, along with the uneven state and local role in administering UI, meant that in practice millions of Black and Hispanic workers were denied and excluded from any UI relief. Also, the duration of benefits got an automatic extension during periods of economic distress with the Extended Unemployment Compensation Act of 1970, which created a mandatory permanent program of extended benefits (EB) (Price 1985). Under the EB program, special EB benefits would be “triggered on” if a certain unemployment rate was reached. Despite this automatic program, Congress still saw a need to address major economic distress among U.S. households after major recessions such as the downturns in 1974, 1982, 1991, 2002, and significantly, 2008, by enacting ad hoc federal temporary programs of supplemented UI, such as the Emergency Unemployment Compensation (EUC08) program (CRS 2014).

However, the most dramatic changes to the UI system came in the wake of the recent 2020 crisis. This sharp downturn, driven by the COVID-19 pandemic, forced millions out of work in mere weeks, and spurred a rapid congressional response to temporarily reinforce and expand existing UI programs. The temporary expanded unemployment insurance programs created in the Coronavirus Aid, Relief, and Economic Security (CARES) Act of 2020 expanded traditional UI on three margins: duration, generosity, and eligibility.

Duration: The Pandemic Emergency Unemployment Compensation (PEUC) program provided up to 53 weeks of additional UI payments that laid-off workers could tap into after exhausting traditional UI benefits and EB.

Generosity: The Pandemic Unemployment Compensation (PUC) program provided an additional $600 per week on top of existing UI benefits. This program was allowed to expire in July 2020. The Lost Wages Assistance (LWA) program, which provided six weeks of additional $300 weekly UI from disaster relief funds, was authorized until September 2020. Congress renewed the PUC at $300 per week in an appropriations bill in December 2020 and then in the American Rescue Plan (ARP) Act of 2021.

Eligibility: The Pandemic Unemployment Assistance (PUA) program extended UI eligibility to workers who previously could not receive UI, such as workers classified as independent contractors, app-based and “gig” workers, part-time workers, and workers with short or irregular work histories. PUA also extended eligibility to workers who voluntarily left jobs in response to public health fears spurred by the pandemic and the closures of schools and day care centers.

This paper focuses on the UI program’s role as a macroeconomic stabilizer and will explore the ways UI could be augmented as an automatic stabilizer. Research from before the 2020 crisis shows the relatively modest boost that the UI system provided at the outset of past economic downturns. For example, Chodorow-Reich and Coglianese (2019) estimate that all of the extended UI benefit programs—both standard EB programs and the ad hoc emergency unemployment compensation programs passed in 2008—probably served to lower the overall unemployment rate by only about 0.2% in 2010 (the labor market trough of the Great Recession, when unemployment was 5 percentage points higher than in the pre-recession years of 2006 and 2007).

However, the 2020 UI modifications and improvements were the most significant in history, and they likely provided a much larger potential stabilizing role for the economy.1 This paper will analyze the extended UI programs enacted in response to the 2020 crisis to compare the counter-cyclical fiscal boost provided by the 2020 UI programs with UI response in previous downturns. Even as policymakers allowed the pandemic UI to cease completely by September 2021, the performance of UI as a potential macroeconomic stabilizer in 2020 ought to be examined carefully as a lesson for future downturns if policymakers return to the issue of long-term reforms of the UI system.

UI as an income stabilizer during the COVID-19 crisis compared with previous downturns

UI supplements household income after a job loss and provides a buffer to economywide consumption spending in the face of sudden earnings losses. Looking at UI’s share of household personal income (which includes wage and salary income as well as government social benefits such as Social Security, Medicare, Medicaid, and UI) before and after recessions can give a rough estimate of the boost provided solely by UI. Figure A shows the average boost from UI to personal income across business cycles from 1960.

Figure A

The first bar in each pair indicates UI income as a share of personal income during the official recession period, relative to the year prior to the recession. The second bar compares the share of personal income accounted for UI during the first year of recovery relative to the year prior to the recession. As can be seen, because in many recessions, employment losses have lagged behind other measures of recession, the boost from UI is often greater in the first year of recovery than during the official recession.

The most striking finding of the figure, however, is that expanded UI produced a dramatically larger boost to personal income both during and after the COVID-19 crisis compared with prior recessions. In past downturns, UI provided only a very modest boost to personal income. For example, even in the Great Recession of 2008–2009, UI boosted personal incomes only by about 0.3% in the depths of the recession, compared with just under 1% in the COVID-19 crisis. Critically, in 2020, UI also boosted personal incomes in the immediate recovery: UI’s share of personal income rose 2.3% (in the first year of recovery from the Great Recession, UI’s share rose a mere 0.6 percentage points).

In some sense, looking at the UI’s boost to overall personal income during the COVID-19 crisis can understate how transformational it was for the lives of workers. COVID-19 relief legislation included many large transfers besides UI expansion, such as the stimulus checks and expanded Child Tax Credit (CTC), which boosted personal income as well. Given that UI serves explicitly as a replacement for lost labor earnings, Figure B isolates this role by looking at UI as a share of wage and salary income plus UI payments. Examined this way, the UI response to the COVID-19 recession far overshadows any previous downturn since 1979.

Figure B

The impact of the expanded UI provisions from the CARES Act passed in late March 2020 is most evident from May to July 2020 when UI reached a staggering 13% of wage and salary income. The sharp fall after July mostly reflects the failure of Congress to extend the supplemental UI programs.2

From a macroeconomic point of view, UI boosting personal incomes (and labor earnings) both during a recession and immediately after is supremely valuable. With UI payments, households headed by those who have lost jobs have more funds to cover their rents, living expenses and debts, and hence consumption throughout the economy is buttressed even as earnings fall. This dynamic also works in reverse: If UI is cut prematurely when the labor market is still weak, the reductions in household incomes put downward pressure on consumption spending, which then slows economic growth.

What we learn from looking at UI support during and after recessions since 1960 is that policymakers have never used UI as effectively for macroeconomic stabilization as they did for the 2020 COVID-19 crisis. By expanding UI so significantly in duration, generosity, and eligibility during the crisis, federal policymakers greatly augmented UI’s potential role as a macroeconomic stabilizer. Given how the U.S. economy has taken longer and longer to regain pre-recession health after each recession since the early 1980s, any lessons on improving automatic stabilizers and fostering more rapid recoveries should be examined closely (Freeman 2013).

In the next section, we provide some rough quantification of how important changes to each of the three critical elements—duration, generosity, and eligibility—are to UI’s outsized performance in stabilizing incomes during the COVID-19 crisis. We then evaluate how expansions in these three areas could be incorporated into a long-term reform of UI. It is well-known by now that UI generally does not respond automatically enough or at sufficient scale to downturns (Bivens et al. 2021). Even more glaringly, sometimes recession-driven expansions are pulled back before full economic recovery is reached (Bivens 2016). Enhancing the margins along which UI can effectively stabilize the macroeconomy and having those margins respond automatically to downturns could provide a much better buffer against future recessions and too-slow recoveries.

How the pandemic UI programs expanded benefit duration, generosity, and eligibility

Historically, the changes to UI duration, generosity, and eligibility have been relatively modest, even in the face of recessions. Regarding eligibility, some states have opted to relax eligibility requirements during recessions (CRS 2020; Congdon and Vroman 2021). The federal “extended benefit” (EB) program operating in all states is meant to trigger-on automatically as the unemployment rate rises, but it has serious flaws (Bivens et al. 2021). Largely due to these flaws, UI duration is often extended on an ad hoc basis by Congress during national recessions. Finally, benefit levels have traditionally been very modest in standard UI programs (generally replacing substantially less than 50% of workers’ wages) and have been only rarely boosted in response to recessions, and even then, only modestly. For example, the American Recovery and Reinvestment Act of 2009 boosted weekly UI benefits by $25.

Gauging the need for expanded eligibility 

The large pandemic changes to UI eligibility can be seen clearly in Figure C, which shows weekly claimants of UI from 1986, with special programs highlighted.

Looking at UI program by claimants tells us a few things, even before we get to issues of eligibility. First, before the 2008–2009 Great Recession, nonstandard UI programs (either EB or EUC) provided only very small shares of total UI coverage. Second, the 2020 COVID-19 recession, in both severity and federal fiscal response, was unprecedented in nature compared with previous downturns since 1986. From 1986 through 2007, weekly claimants never rose above 7 million. While the downturns between 1986 and 2007 were certainly less severe, many potential claimants were likely shut out of UI or undercompensated due to the lack of extended benefits duration and eligibility and lack of expanded benefit amounts.

Figure C

Crucially for eligibility, the impact of the PUA program (the pandemic program expanding eligibility to workers not traditionally covered by UI) in 2020 and 2021 is striking. With over 30 million claiming unemployment insurance at the height of the downturn, traditional UI declined heavily as a source of UI coverage, falling from 100% of all UI (pre-CARES Act) claims to just 20% by June 2021. At its peak in August 2020, PUA was covering 15 million workers, and made up half of all UI claimants. PUA recipients are generally workers who just would not have been covered at all under traditional UI, and who would hence have had no income support to buffer their spending as jobs dried up.

Figure D provides another way of highlighting the importance of PUA and the eligibility expansion to macroeconomic stabilization. It shows the total dollar contribution of UI to personal income divided by the unemployment rate. This is a measure of how much UI adds to personal income for each percentage-point increase in the overall unemployment rate.3 The figure separates out the PUA program contribution from all other UI programs. Within a few months following the CARES Act, non-PUA UI programs were transferring more than $60 billion into personal incomes per percentage point of measured unemployment, and even as of spring 2021 were transferring more than $40 billion per percentage point of unemployment. PUA programs, however, were transferring almost exactly as much as non-PUA UI in the summer of 2020 and the winter of 2021, effectively doubling the effectiveness of the entire pandemic UI system. This highlights just how important modernizing the eligibility component could be for boosting the UI system as a macroeconomic stabilizer.4

Figure D

Gauging the need for benefits that last longer 

Looking at claimants in the 2008–2009 crisis in Figure C also offers key insights about potential benefit duration (i.e., the maximum number of weeks of UI benefits that applicants meeting the criteria could obtain). First, extended programs have provided some nontrivial expansions of UI coverage in the past. At its peak in 2010, the EUC program enrolled an average of 4.6 million workers (the EUC program is reflected in the Nonstandard (discretionary) federal programs stack in Figure C). However, a sharp cutoff of the EUC program in 2014 is clearly visible as well—this cutoff, which happened due to congressional whim rather than any serious assessment of labor market health—occurred with the unemployment rate still over 7%, a higher level than occurred at any point during the labor market recession of 2001–2003. This 2014 cutoff likely had serious effects on the pace of recovery in subsequent years (Shierholz and Mishel 2013).

One reason why the EUC program was so important was because the automatic state-administered EB programs were so flawed, and many states saw the EB benefits trigger-off even at quite-high unemployment rates. In Figure C, this can be seen in how small EB enrollments were relative to EUC in the pre-2014 years of the recession and recovery.

Despite its importance, the EUC program was cut off too early, depriving workers of benefits they would have gotten were the UI program designed to extend benefits while the labor market is still in serious distress. Figure E demonstrates just how premature the 2014 cutoff of EUC was by showing how many states would have allowed workers access to extended potential benefit durations if these extended benefits triggered off only when the state unemployment rate fell to 6%, 5.5%, or 5%.

Figure E

The results are striking. By January 2014, EB programs had all triggered off and the EUC program had lapsed. In 2014, workers in 37 states would have been able to access extended benefit durations if 5% was the benchmark for triggering-off these provisions, and 27 states would have kept extended benefits available to unemployed workers even with a 6% benchmark. Even in 2016, a 5% benchmark would have allowed 21 states to continue offering extended benefits, with seven states keeping them with a 6% benchmark.

Gauging the need for higher benefit levels

Perhaps the most well-known changes to UI made during the pandemic concerned the level of benefits. The FPUC program within the CARES Act provided a $600 boost to weekly benefits. The $600 figure was chosen to ensure at least 100% wage replacement for essentially all workers. This high replacement ratio arguably made economic sense in this context. During the peak spread of the virus in 2020, it was essential from a public health standpoint that people not work in person; non-employment was actually a policy goal during this brief period and hence any moral hazard concerns regarding the effect of UI recipiency on incentives to search for jobs were rightly considered insignificant. The original $600 boost was cut off in August 2020. In January 2021, a $300 boost provided by congressional appropriators in December 2020 was codified in the American Rescue Plan. In July and August, a number of states chose to end the PUC programs early. By early September all pandemic UI programs had lapsed, and as of early October, prospects for any resuscitation of these programs seem extremely remote.

There is no evidence the original $600 top-up in additional UI benefits throttled economic recovery, as the extraordinarily rapid bounceback from the first wave of COVID-19 shutdowns began before the $600 benefit expired in summer 2020. Additionally, significant economic research has emerged showing that states that cut off the $300 boost early, claiming that it dissuaded job search activity, have not seen sustained job growth or hiring either. Knowing that the benefit expansions did not cut off economic recovery can inform UI reforms that incorporate large benefit extensions for future downturns.

Parsing out how expanded benefits eligibility, duration, and levels contributed to stabilizing the pandemic economy 

All three pandemic UI programs and the expansions they provided along crucial margins—PEUC (duration), PUC (generosity), and PUA (eligibility)—boosted the income support provided by UI enormously. Figure F shows the relative contribution of each pandemic UI program to wage and salary income since March 2020.

Figure F

We can see the efficacy of the three programs in tandem as well as the dramatic impact of letting the $600 weekly PUC program expire in summer 2020 and resume at a lower level in 2021. While the impact of PUA (expanded eligibility) is relatively steady throughout 2020 and 2021, the importance of PEUC (additional weeks) appears only from October 2020 onward as more claimants exhausted other UI options and the number of long-term unemployed persons increased. The extended duration of the PEUC program also made a larger impact than the EB program, which shows a much smaller contribution to income in late 2020 and 2021. As shown before in Figures A and B, the cumulative impact of these programs was huge, and each program provided a crucial role.

These dramatic expansions and their salutary macroeconomic effect should inform policymakers as they ponder long-term reforms to the nation’s UI system.

Issues in assessing the likely 10-year fiscal costs of UI reforms 

The UI system’s weaknesses as a macroeconomic stabilizer have been known for some time—and were particularly apparent during the long and slow recovery following the Great Recession. There are likely many political reasons why these weaknesses have not been addressed—many not unique to UI and likely related to why the U.S. has such a small fiscal footprint across-the-board. But some oddities in how the fiscal cost of these reforms might be scored add to the difficulties of reform.

Policy efforts to make automatic stabilizers like the UI system more responsive and more effective in supporting aggregate demand during economic downturns suffer greatly from a problem of timing inconsistency—the minds of the public and policymakers are focused on this need most when undertaking a permanent reform will look expensive as scored by the Congressional Budget Office (CBO). As explained below, reform will only look substantially cheaper in CBO scores precisely when the need for reform seems less urgent (during expansionary periods).

House Speaker Nancy Pelosi summarized the issue when asked why Congress had not taken up permanent reform to the UI system as part of efforts to respond to the economic shock caused by COVID-19.

At a May 14 press conference, House Speaker Nancy Pelosi laid it out. ‘I’m a big supporter of having stabilizers in the bill,’ she said. She blamed their absence on the Congressional Budget Office (CBO), which estimates the costs of legislation, because under CBO’s rules, the likely cost of the stabilizers ‘counts in the bill today.’” (Klein 2020)

The reason for this time-inconsistency issue is straightforward: the CBO essentially assumes the economy moves from its current state of slack (weak economic demand and weak demand for labor) to a state of full employment within a few years, and that developments four years or more out cannot be precisely forecast. In practice, this means if the economy is currently experiencing high unemployment and a permanent reform to the UI system was proposed, the CBO would (sensibly) forecast unemployment to be elevated for the next few years before settling down closer to full employment. In these first few years with elevated unemployment, a substantially more-generous UI system would be scored as being quite expensive in the short-run as many unemployed workers would be drawing benefits. Conversely, if the economy were currently experiencing quite low unemployment and a permanent reform of UI was proposed, the CBO would forecast low unemployment over the entire 10-year budget window, having no real capacity to forecast otherwise more than a few years down the road. A low unemployment rate over the entire 10-year budget window would in turn make reforms to UI look significantly cheaper when implemented in this hypothetical low-unemployment year than if implemented during a recession.

Of course, it should be noted that when the national unemployment rate rises during and after recessions, Congress has tended to do something to boost the generosity of the UI system, even with no “automatic” spending that the CBO could reliably put into a budget forecast. Further, these ad hoc UI enhancements (longer benefit durations generally, along with an occasional small boost to weekly benefits) have added to federal spending. So the refusal to pass a permanent change to UI during recessions makes little sense in real-world fiscal terms: Automatic or not, UI spending rises during recessions, regardless of what the CBO has previously forecast for such spending. Any reluctance to undertake structural reform to automatic triggers during times of labor market distress really seems to be a case where costs forecast by the CBO are somehow more daunting to policymakers than costs accompanying the passage of real-time legislation during recessions.

Would the CBO score UI reform as “free” if undertaken during an expansion? 

It is not quite the case that a structural reform to UI that increased UI payments during labor market downturns would be forecast as essentially free by the CBO if scored when unemployment was low. The CBO has little basis to forecast the timing of recessions outside of the next year or two, but it does (sensibly) recognize that recessions are likely to occur in any such 10-year window.

Consider, for example, a scenario in which the unemployment rate is 4.5% in the current year and a UI reform is proposed that only provides higher levels of UI funding (longer durations or more-generous benefits) if the unemployment rate rises above 5.5%. In this situation, the baseline CBO forecast will show essentially constant 4.5% unemployment over the next 10 years. But the CBO will draw on historical experience to estimate a probability that unemployment will rise over 5.5% for a given period of time over that window. In a paper explaining this process, the CBO refers to this as “estimating the costs of one-sided bets.”5

In practice, if a UI reform that contained the elements we highlighted above were passed today, the CBO score of its cost over the next 10 years would likely add up its “normal” cost (cost increases during time when the national unemployment rate was below any “trigger”) and then would add on the expected value of recession-driven costs. The rest of this section aims to provide a very rough estimate of how these issues would be estimated in the context of ambitious UI reforms.

What would UI expansions along all three margins cost?

The rest of this section addresses these questions of a UI reform’s fiscal cost and how a reform package might be scored by the CBO. For an archetype reform, we look at the budgetary cost of a reform that makes the following changes:

• doubles UI recipiency rates (share of unemployment workers receiving UI benefits) during times of non-elevated unemployment
• increases UI benefit levels by a factor of 1.75
• provides for automatic triggering of extended potential benefit durations during times of high unemployment, with the longest maximum potential benefit duration rising to 95 weeks when unemployment hits 10%
• boosts benefit levels during recessions by an average of $100 per week (over already-augmented benefit levels in normal economic times). These reforms are very roughly in line with a set of reforms suggested by Dube (2021) and Bivens et al. (2021).

These parameters are slightly more generous than those suggested in a recent policy white paper released by the offices of Sens. Ron Wyden (D-Ore.) and Michael Bennet (D-Colo.).

Likely budgetary costs of making UI a more effective macroeconomic stabilizer

Rough budgetary costs for the first two margins—expanded eligibility and more-generous normal benefit levels (bullets one and two above)—are relatively straightforward to calculate for periods of low unemployment (i.e., before any recession-driven triggers kick-in). However, assessing the cost of longer potential benefit durations and increases in benefit generosity that rise as labor market conditions deteriorate (bullets three and four) requires drawing on others’ research.

Costs of expanded eligibility and increased standard benefit levels 

A number of UI reform proposals (Dube 2021 and Bivens et al. 2021) include measures both to expand eligibility and to raise benefit levels even during periods of low unemployment. To get a very rough estimate of the budgetary cost of proposals like this, we can look at average UI spending between 2016 and 2019 and then adjust it for a counterfactual where eligibility requirements were expanded such that the recipiency rate doubled, and where there was an across-the-board increase in benefit levels.

Table 1 provides most of the information needed for this calculation. It shows that in 2016–2019, the unemployment rate averaged 4.2%, and average spending on UI was $29.6 billion annually. The recipiency rate averaged 27.3% and the average replacement rate for benefits was 39.2%. If the recipiency rate doubled and replacement rates increased by a factor of 1.75 (from 39.2% to 68.6%), then spending would increase an estimated $36.3 billion in those years ($65.9 billion minus $29.6 billion), or, roughly 0.15% of gross domestic product.

Table 1

Adding the costs of automatic expansion of potential benefit durations and higher weekly benefits during labor market recessions 

Assessing costs for new parameters that depend upon the state of the business cycle is a much more complicated task. Luckily, Chodorow-Reich and Coglianese (2019) have done extensive work in simulating how the cost of various UI reform proposals would vary depending on the severity and length of potential recessions. Table 2 uses their findings as a baseline to assess costs of the UI reform outlined in the four bullets earlier, but then shows how the probability of recession affects these costs. A fuller explanation of how we derived costs in this table is provided in the appendix.

Table 2

The first row of Table 2 simply shows the 10-year cost of current law regarding UI and the 10-year cost of the reform detailed above if no recession occurs over those 10 years. We include two columns for the “current law” estimate because we assess this cost under two assumptions: that Congress passes no emergency boost to UI during recessions, or that Congress provides the same emergency boost that it has typically legislated in past recessions. In the first row, because no recession is assumed, these costs are identical in the two “current law” scenarios. The cost under reform starts from the $65.9 billion cost estimated in Table 1 for the first year, and then accounts for inflation and labor force growth.

The next two rows provide estimates of the incremental 10-year UI costs spurred by either a mild or severe recession. Under current law, assuming no discretionary response from Congress, the incremental cost is driven by the fact that the federal government finances half of the extended benefit (EB) programs that trigger-on at the state level when unemployment rises.

For a severe recession, we assess the costs of current law assuming a discretionary response by applying the incremental boost to UI spending provided between 2008 and 2013—the period of labor market distress caused by the Great Recession. For the mild recession, we mark-down the spending in the severe recession by 45%, a ratio we obtain from the Chodorow-Reich and Coglianese (2019) estimates of UI costs during recessions of different intensity. Finally, for the incremental cost of recession under reform, we take the Chodorow-Reich and Coglianese (2019) estimate of their proposed reforms and wedge them up to reflect the increased costs of the archetype 2021 reform relative to their proposals. For example, their reform calls for a $50 per week increase in benefit levels during recessions, but our enhancements call for an average increase of $100 during recessions (over already-augmented benefit levels in normal economic times). Accordingly, we double their estimate of the cost of a benefit increase during mild or severe recessions.

Over the next 10 years, assuming no recession, UI spending would be $330 billion under current law, but would rise to $690 billion under our reform. In the case of a mild recession at some point during the decade, under current law and with no discretionary action from Congress, UI spending would be $373.6 billion over the next 10 years ($330.0 billion plus the incremental cost of recessions of $43.6 billion shown in Table 2). Under current law but with discretionary actions by Congress similar to past recessions, UI spending would be $547.7 billion over the next 10 years if there were a mild recession during that period. Under our archetype reforms, spending over the next 10 years would be $924.1 billion over the next 10 years if there were a mild recession during that period.

The last row translates these scenarios to an average annualized cost over the next decade when factoring in the 1-in-3 chance that the economy experiences no recession, the 1-in-3 chance that it goes through a mild recession, and the 1-in-3 chance it suffers a severe recession over the next 10 years. Under current law but assuming no emergency spending measures enacted by Congress during recessions, average annual costs would be $37.1 billion for the next decade. Under current law but assuming Congress acts as it has in the (pre-COVID-19) past during recessions, average annual costs would be $53.3 billion. Under the archetype reforms outlined earlier (doubled UI recipiency rates and almost doubled benefit levels during times of non-elevated unemployment, maximum potential benefit duration rising to 95 weeks when unemployment hits 10%, and an additional $100 per week boost to benefit levels during recessions), average annual costs would be $106.6 billion.

Conclusion

In previous economic downturns, benefits paid out under the current unemployment insurance system provided only modest boosts to aggregate demand, and thus has had a limited role as an automatic stabilizer. However, the pandemic UI programs greatly boosted the contribution that UI benefits made to personal income. These programs enhanced the UI system’s effectiveness in boosting personal income along three crucial margins: expanding eligibility to more workers, extending the potential number of weeks that eligible workers could claim benefits, and increased benefit levels. Policymakers going forward should examine this episode closely to see how eligibility, benefit levels and duration enhancements could be part of a structural reform of the UI system to make UI a more effective macroeconomic stabilizer.

Appendix: Table 1 and 2 methodology

The first column of Table 1 reports the average values over 2016–2019 for the unemployment rate, the share of unemployed workers receiving UI benefits (the recipiency rate), the average share of wages replaced by UI benefits (the replacement rate), and annual UI spending (in billions of dollars). Between 2016 and 2019, overall unemployment was low by historical standards, so the annual UI spending can be interpreted as what could be expected in years when the labor market is not seriously damaged by current or recent recessions.

The next column shows what the recipiency rate, the replacement rate, and average annual spending would have been in those years if the broad reforms described in the paper were made. Note that in practice this means what the rates and spending would be under implementation of the two of the four broad reforms that have to do with eligibility and benefits levels during standard times. The table assumes reforms that would boost recipiency during nonrecessionary times by 100% (pushing the recipiency rate to 54.5%) and would boost the replacement rate of UI benefits by 75% (boosting the replacement rate to 68.6%). Given more people collecting higher benefits, annual UI spending would more than double, rising from just under $30 billion to almost $66 billion.

We use these numbers as inputs for the calculations made in Table 2, which shows in very broad strokes how the CBO might be likely to score large reforms to UI. The first row of Table 2 shows the likely 10-year cost of UI spending under current law and under the reform if no recession occurs over the 10-year window. The current law trajectory includes two different scenarios: one where Congress provides no emergency response to a recession with discretionary spending measures, and one in which Congress provides a discretionary response that is similar to the congressional response to past recessions. This second scenario is necessary for a realistic assessment of the incremental cost of UI reform that strengthens the system’s automatic response to recessions. In the absence of automatic change in UI parameters, the realistic alternative is not no change at all to UI during recessions—Congress routinely steps in and provides some extra boost to UI during recession (even if this discretionary boost is often insufficient and too short-lived).

The next two rows in Table 2 show the incremental cost over and above the “no recession” scenario that would be imposed by a severe or mild recession. Under this scenario, the two additional reform measures outlined in our report (maximum potential benefit duration rising to 95 weeks when unemployment hits 10%, and an additional $100 per week boost to benefit levels during recessions) would kick in. To assess the costs of these reforms, we draw on calculations in Table 2 of Chodorow-Reich and Coglianese (2019), who assess the incremental costs of current UI law under a range of recessionary scenarios. They then assume three UI reforms which are largely in line with our reforms. We use the ratio of their reform costs to current law costs in recessionary scenarios as a “multiplier” to apply to our own reform costs during recessions. Further, when our reforms are more expensive than the Chodorow-Reich and Coglianese reforms, we inflate estimates of our reforms appropriately. For example, Chodorow-Reich and Coglianese (2019) call for a $50 weekly supplement to UI during recessions, while under our reform the boost would be $100 a week. Similarly, because our reforms call for modestly longer potential benefit durations during recessions than do Chodorow-Reich and Coglianese, we boost the estimated costs of these by 20% relative to their estimates.6

Finally, the last row calculates the annualized cost of each of the three policy regimes: current law with no emergency or discretionary response, current law with emergency response during recessions, and the archetypal reform outlined in our report. To calculate these annualized costs we assume there is a one-third probability each of: no recession during the next 10 years, a mild recession during that time, or a severe recession. Under the current-law no-emergency response policy scenario, annualized UI costs would average $37.1 billion each year over the next decade. Under the current-law, emergency response scenario, these costs would rise to $53.3 billion annually. Finally, under the enhancements outlined in this report, the costs of UI would average just under $107 billion annually.

Acknowledgments

This report was made possible by support provided by the Peter G. Peterson Foundation (PGPF). Melat Kassa and Jori Kandra provided research assistance and Lora Engdahl edited.

Endnotes

1. We say “potential” stabilizing role because the large UI expansions in the CARES Act were not entirely meant to stabilize macroeconomic measures like GDP. In most recessions, a prime goal of expanding UI would be precisely to stimulate economic activity. But in a pandemic-driven recession where much economic activity was shut down due to public health concerns, the primary role of UI was social insurance and redistribution. That said, the very rapid bounceback of economic activity after the first wave of pandemic shutdowns was certainly aided by the income boost provided by the CARES Act UI expansions—and that is true for the rapid bounceback of activity so far in 2021, following the UI expansions in the American Rescue Plan (ARP).

2. A small share of the drop-off after July is due to improving labor market conditions.

3. This measure allows us to get a measure of UI generosity by controlling for the fact that UI mechanically rises as the unemployment rate rises.

4. Alongside independent contractors and gig workers, the PUA program also expanded eligibility to those who were unemployed or unable to work due to COVID-19, including caring for someone with the virus, providing care to a child or family member whose school or care facility was closed, or refusing to work in an unsafe work environment. This array of eligibility extensions meant that a not-insignificant share of PUA recipients were those who quit their jobs due to fear of the virus, contagion, and unsafe work conditions or who left the workforce due to school closures and lack of affordable and safe child care. These eligibility criteria were specifically included due to the particular nature of the public health crisis directly impacting employment. Future UI reform programs expanding eligibility will very likely not contain such broad eligibility conditions. Therefore, we would not expect structural UI reform at the federal level to boost eligibility as much as the 2020 PUA program. Unfortunately, specific recipient and eligibility breakdowns within the pandemic UI programs are not available. While the U.S. Department of Labor compiles the aggregate PUA figures, more specific tracking within programs has not been possible, partially due to state administering of UI and variance of reporting requirements, as well as the state administrative burden of implementing the pandemic programs.

5. The “natural rate” of unemployment is the rate below which further increases in economywide spending will mostly lead to accelerating inflation rather than greater output. In a well-managed macroeconomy, any time spent 1% over the economy’s natural unemployment rate should be matched by an equivalent amount of time spent 1% below the economy’s natural unemployment rate. But, because in most proposed reforms UI benefits do not get cheaper or less expansive as the unemployment rate falls beneath the natural rate, this does not provide one-for-one countervailing savings that cancel out the fiscal effect of UI benefit expansions that kick in as unemployment rises.

6. This 20% is likely an overestimate. Chodorow-Reich and Coglianese (2019) show that potential benefit durations of over 46 weeks did very little to boost overall UI spending during the Great Recession.

References 

Bivens, Josh. 2016. Why is Recovery Taking So Long—And Who’s to Blame? Economic Policy Institute, August 2016.

Bivens, Josh, Melissa Boteach, Rachel Deutsch, Francisco Díez, Rebecca Dixon, Brian Galle, Alix Gould-Werth, Nicole Marquez, Lily Roberts, Heidi Shierholz, and William Spriggs. 2021. Reforming Unemployment Insurance: Stabilizing a System in Crisis and Laying the Foundation for Equity. Center for American Progress, Center for Popular Democracy, Economic Policy Institute, Groundwork Collaborative, National Employment Law Project, National Women’s Law Center, and Washington Center for Equitable Growth, June 2021.

Bureau of Economic Analysis (BEA). 2021b. “Table 2.1: Personal Income and its Disposition”, National Income and Product Accounts. Accessed July 2021.

Bureau of Economic Analysis (BEA). 2021a. “Effects of Selected Federal Pandemic Response Programs on Personal Income.” March 2021.

Bureau of Labor Statistics (BLS). 2021a. “State Unemployment Rates Over the Last 10 years, Seasonally Adjusted: June 2011 to June 2021.” State Employment and Unemployment Chart Package, 2021.

Bureau of Labor Statistics (BLS). 2021b. Unemployment Rate Data Series ID LNS14000000 [Excel file], Accessed July 2021.

Chodorow-Reich, Gabriel, and John Coglianese. 2019. “Unemployment Insurance and Macroeconomic Stabilization.” In Recession Ready: Fiscal Policies to Stabilize the American Economy, edited by Heather Boushey, Ryan Nunn, and Jay Shambaugh, 153–179. Washington, D.C. Brookings Institution.

Chodorow-Reich, Gabriel, John Coglianese, and Loukas Karabarbounis. 2019. “The Macro Effects of Unemployment Benefit Extensions: A Measurement Error Approach.” Quarterly Journal of Economics 134, no. 1: 227–279.

Congdon, William J. and Wayne Vroman. 2021. Extending Unemployment Insurance Benefits in Recessions: Lessons from the Great Recession. Urban Institute, February 2021.

Congressional Research Service (CRS). 2014. Extending Unemployment Compensation Benefits During Recessions. CRS Report RL34340, October 2014.

Congressional Research Service (CRS). 2020. Comparing the Congressional Response to the Great Recession and the COVID-19-Related Recession: Unemployment Insurance (UI) Provisions. CRS Report R46472, July 2020.

Dube, Arindrajit. 2021. “Aggregate Employment Effects of Unemployment Benefits During Deep Downturns: Evidence from the Expiration of the Federal Pandemic Unemployment Compensation.” National Bureau of Economic Research Working Paper no. 28470, February 2021. https://doi.org/10.3386/w28470.

Freeman, Richard B. 2013. “Failing the Test? The Flexible U.S. Job Market in the Great Recession.” National Bureau of Economic Research Working Paper no. 19587, October 2013.

Hickey, Sebastian. 2021. “New Personal Income Data Show the Need for Broad and Permanent Unemployment Insurance Reform.” Working Economics Blog (Economic Policy Institute), April 23, 2021.

Klein, Ezra. 2020. “The Vital Missing Piece of the Democrats’ Stimulus Bill.” Vox, May 28. 2020.

National Bureau of Economic Research (NBER). 2021. “US Business Cycle Expansions and Contractions.” NBER Public Use Data Archive, July 19, 2021.

Price, Daniel N. 1985. “Unemployment Insurance, Then and Now, 1935-85.” Social Security Bulletin 48, no. 10: 22–32.

Shierholz, Heidi, and Lawrence Mishel. 2013. Labor Market Will Lose 310,000 Jobs in 2014 if Unemployment Insurance Extensions Expire. Economic Policy Institute, November 2013.

U.S. Department of Labor Employment and Training Administration (U.S. DOL-ETA). 2021. Unemployment Insurance Data: Continuing Claims, All Programs [Excel file]. Accessed April 2021.

This report assesses the potential macroeconomic impact of two pieces of legislation pending in Congress: the Infrastructure Investment and Jobs Act (IIJA), which incorporates elements of the American Jobs Plan (White House 2021a, 2021c, 2021d), and Congress’s $3.5 trillion (over 10 years) budget reconciliation bill, still being written in Congress, which incorporates measures proposed under the Biden administration’s American Families Plan of the Build Back Better agenda (White House 2021b). The investments and social insurance expansions provided for by these plans will boost productivity and provide key relief to family budgets in coming years. The benefits from these policies will be realized even if the economy has reached full employment when they take effect. Moreover, these plans also provide a valuable backstop against the possibility that the economic recovery falters after 2022 as the effect of the American Rescue Plan (ARP) fades, as was the case when fiscal support for recovery ended prematurely following the Great Recession. This report highlights just how strong a fiscal backstop the plans will provide.

The two pieces of legislation, amounting to just over $4 trillion in new spending over 10 years, reflect versions of the Biden-Harris administration’s economic agenda that have been scaled back in order to strike political compromises necessary to earn support for passage in Congress. In departing from more ambitious plans, Congress will reduce the level of insurance the legislation provides against a flagging recovery in coming years (Zandi and Yaros 2021). Nonetheless, together these two pieces of legislation would provide much needed support to a still-recovering job market, enhance equity and long-term economic performance, and take serious steps toward addressing the climate crisis we can already see unfolding. The report finds:

• Combined, the IIJA and budget reconciliation package would provide fiscal support for more than 4 million jobs per year, on average, over the course of the 10-year budgeting window, through direct spending and increased indirect demand in related industries. The analysis does not account for dynamic effects of the planned investments, though these policies are also certain to raise business and worker productivity, and to create faster and more equitably distributed long-run economic growth and increased tax revenues.
• The budget reconciliation package under consideration would support a far greater number of jobs than the IIJA. On its own, the IIJA will provide fiscal support for 772,400 jobs per year, or 19% of the total jobs supported by the combined package. In comparison, the budget reconciliation is expected to support more than 3.2 million jobs per year, or 81% of the total jobs. The budget reconciliation’s outsize economic impact flows from its more significant financial commitment to public investments.
• Manufacturing industries would see a significant boost under these combined plans, with more than 556,000 jobs supported annually.
• The planned investments would support more than 312,000 jobs annually in construction industries.
• The budget reconciliation would vastly expand caregiving jobs to address unmet needs in child care and elder care, supporting 1.1 million jobs per year. Such investments in universal pre-K, child care, and long-term care would not only disproportionately provide jobs for women—and particularly for women of color—but would also enable millions to participate more fully in the workforce at higher productivity and to earn higher compensation. This social infrastructure investment would facilitate increased accumulation of human capital critical for America’s long-term economic growth prosperity; and it would provide crucial relief to family budgets straining to balance work with the costs of caregiving.
• Climate-related and other environmental provisions in the legislation would support more than 763,000 jobs annually. This includes jobs supported by investments in electric vehicle infrastructure and federal procurement of clean technologies, public transit, power infrastructure, climate resilience, agriculture and forestry innovations, environmental remediation, and scientific research and development, among other measures.

In the sections that follow, we first explain the spending plans embodied in the IIJA and Congress’s anticipated budget reconciliation bill. Next, we explain the methodological approach we use to assess what job impacts can be expected from the proposed legislation, and we present the detailed results of our analysis. Finally, we summarize the opportunities legislators now have to rebuild the American economy stronger and better than before.

Plans to rebuild the economy for all

Our analysis assesses the impacts of just over $4 trillion in new spending on a range of policy initiatives over a 10-year budget window, specified in Table 1. First, the Infrastructure Investment and Jobs Act reflects a bipartisan compromise—modeled on the American Jobs Plan—negotiated by President Biden and a bipartisan group of senators to expand investments in surface transportation, public transit and rail, water, and broadband internet infrastructure, along with new investments in renewable energy and electric vehicles (White House 2021a, 2021c, 2021d). Although the media often report this bill as having a $1.2 trillion price tag, the package encompasses much previously budgeted and paid-for infrastructure spending; thus, the analysis here focuses only on the nearly $550 billion in net new infrastructure investments the bill would authorize.1

Table 1

These investments target sorely needed renewal and expansion of America’s physical infrastructure, which has been allowed to deteriorate for more than a decade (Ayres Steinberg and Hersh 2013; Bivens 2017). The American Society of Civil Engineers (ASCE 2021) estimates that the depreciation of existing public infrastructure assets alone will cost the United States $10 trillion in gross domestic product, 3 million jobs, and $2.4 trillion in lost exports as a result of increased costs of doing business by 2039.

Spending on infrastructure yields immediate benefits due to the labor- and capital-intensive demands of these investment projects, and it continues to yield economic dividends for years to come by allowing people, goods, and ideas to move around more efficiently (Bivens 2019). Estimates of the longer-term economic impacts of infrastructure spending find that returns on investment range from 17% to 73% (Bivens 2017; Heintz 2010; Berechman, Ozmen, and Ozbay 2006) as businesses more efficiently reach markets, workers access more job opportunities, and families find it easier to access quality education and health care.

Although the bipartisan IIJA agreement tees up significant new infrastructure investments, the compromise still falls roughly $2.5–3 trillion short of the U.S. economy’s actual infrastructure demands over the next decade, in addition to $400–600 billion per year needed to achieve carbon net neutrality (Pollin, Chakraborty, and Wicks-Lim 2021). This shortfall leaves significant jobs and economic growth potential on the table. In fact, in some of the states where political opposition to green investment has been highest—and most influential on national policy—such investments in infrastructure and climate change could add a significant number of jobs: These include West Virginia (41,000 jobs), Ohio (235,000 jobs), and Pennsylvania (243,000 jobs) (PERI 2021).

Separate from the IIJA, a budget framework being advanced by congressional Democrats would provide $3.5 trillion in spending over 10 years for priorities set out in the Biden-Harris administration’s American Families Plan (White House 2021b) of the Build Back Better agenda, with offsetting revenues largely derived from tax increases on America’s wealthiest individuals and largest corporations. Congress is still formalizing this legislation; for the purpose of our analysis, we infer the composition of new spending from the original American Families Plan proposal and the scale of the budget resolution that passed both the House and Senate in August and that is expected to be formalized in a forthcoming budget reconciliation bill this month (Table 1). As the budget reconciliation is still under negotiation, it is possible that the resulting legislation could differ significantly from assumptions employed here, with the resulting job impacts differing in turn.

This Build Back Better agenda expands upon more traditional infrastructure investments in the IIJA in ways that promise to be transformative for social equity, manufacturing renewal, energy efficiency, and environmental sustainability. To highlight several themes of these initiatives:

• Expanded Child, Earned Income, and Child and Dependent Care Tax Credits (CTC, EITC, and CDCTC) would provide a boost in financial security to families, providing stability that is shown to increase academic performance, attainment, and lifetime economic mobility (Sherman et al. 2021).
• Investments in child and elder care and universal prekindergarten would address the inequities and inadequacy of America’s caregiving infrastructure, which leave children without the opportunities for early learning and development and leave parents—primarily women of color—with diminished opportunities for work and career advancement that have come to define the pandemic’s “she-cession” (Glynn 2021; Savage 2019). Providing aid for child and elder care also lowers the costs of some of the key stressors of family budgets (Gould and Blair 2020). Finally, creating quality jobs in the caregiving economy would provide substantial economic benefits to a long-marginalized workforce and enhance productivity in the overall economy (Gould, Sawo, and Banerjee 2021; Palladino and Lala 2021).
• Proposed investments in the manufacturing sector and a broad range of technological research and development, alongside commitments to invest in renewable energy generation and climate change resilience and mitigation strategies, carry the potential to revitalize America’s industrial base, reduce energy costs for businesses and households, and prevent future catastrophic losses and economic disruptions from extreme climate events.

Even though it largely finances new expenditures with current revenues, the budget reconciliation plan would still provide an important macroeconomic backstop to aggregate demand in coming years by taking advantage of what economists refer to as the “balanced budget multiplier”: By shifting expenditures from areas with a low propensity to stimulate additional activity elsewhere in the economy to areas with a high propensity to promote downstream economic activity, it is possible to achieve a stimulative effect without fundamentally altering overall fiscal balances.

Although tax increases generally may dampen economic activity, there are two strong reasons to expect the measures proposed in the budget reconciliation plan to impose only a very small fiscal drag. First, the vast majority of families, small businesses, and farm holders would be exempt from the tax increases (Buffie and Lord 2021). The measures would instead focus on raising revenues from America’s highest earners, including the wealthiest individuals and largest corporations—many of whom avoid paying taxes altogether. The plan would exempt those with earnings below $400,000 annually and, in the case of provisions pertaining to untaxed capital gains, those with less than $1,000,000 in income from paying higher taxes.

Second, these rich individuals and big corporations exhibit exceptionally low propensities to spend from additional increases in income. The extremely muted effect on aggregate demand stemming from tax changes could be seen following the passage of the 2017 Tax Cuts and Jobs Act (TCJA), which focused primarily on cutting taxes for corporations and top income earners. The TCJA largely failed to fulfill its defenders’ stated goal of boosting investment because excess economic slack remained in the U.S. economy after its passage; instead of making real investment, beneficiaries of the tax cuts gorged on corporate stock repurchases (Troise 2019). The failure of a quite large (in fiscal terms) tax cut to take up the remaining slack in 2018 and 2019 highlights just how weakly top-end tax changes affect aggregate demand. Gale and Haldeman (2021) document the failure of the TCJA to boost investment; this record is in line with previous experiments with supply-side tax cuts aimed at the top of the income and wealth distribution (Hungerford 2011). Thus, revenues raised from incomes of exceedingly rich individuals and the largest corporations will yield much bigger economic effects spent through this plan than parked in their bank and brokerage accounts.

Analyzing support for widespread employment in good jobs

In assessing the likely employment impacts of any macroeconomic policy change, it is important to be explicit in the concept of jobs being modeled. This report employs the concept of jobs “supported”—the labor inputs required in various industries of the economy to fulfill a given level of economic activity—rather than the concept of jobs “created,” or net increases in the overall level of employment. This distinction reflects the complicated nature of considering the employment effects from significant macroeconomic changes sustained over a relatively long time frame (more than two years, for example).

When an economy operates with pervasive unemployment, an increase in net aggregate demand—typically from additional government investments or from an increase in net exports—requires increased employment of idled labor and capital resources. Thus, the boost to demand generates net increases in total employment. Such excess unemployment and growth constrained by too-slack aggregate demand defines the U.S. economy today. Despite marked recovery in employment from the depths of the COVID-19 recession, labor and capital underutilization persist (BLS 2021b; Federal Reserve 2021). The economy still exhibits an employment shortfall of 6.5 to 9 million jobs (EPI 2021), and—even before the pandemic—conventional measures of unemployment tended to understate economic slack, relegating workers of color to perpetually disproportionately high unemployment (Hersh and Paul 2021).

At some point in the future, it is likely that recent past policy interventions (the American Rescue Plan, most notably) and the expansive policies under consideration with the IIJA and forthcoming budget reconciliation bill will eliminate current economic slack. After this point, the macroeconomic effect of additional spending would largely be to reallocate some employment from one industry to another, rather than adding net new jobs to the economy. Although recent data on consumer prices have some commentators and analysts concerned about the potential for inflation, sober analysis of the available evidence suggests that current inflation is transitory in nature and unlikely to persist once pandemic-related supply-chain bottlenecks are relaxed (Bivens and Thompson 2021). But even at full employment, additional benefits can be derived from spending to support economic activity: Such spending could help shift the composition of overall employment toward better-compensated jobs; eliminate labor market slack, increasing the bargaining power of workers to achieve real wage gains; and direct workers and capital resources into higher-productivity uses that expand America’s economic potential. Further, the broader economic benefits of the Build Back Better agenda—increased productivity through public investment and relief for families through more expansive social insurance—will be realized regardless of the state of labor market slack.

Given these considerations, the best way to view the economic impact of the IIJA and the Build Back Better agenda is to assess the number of jobs its spending supports and the insurance it provides to sustain high growth and tight labor markets in coming years. If economic growth is strong even absent this fiscal boost, then the jobs supported will mostly be reallocations that lead to a fairer and more productive economy. If growth outside this fiscal boost begins to flag, the jobs supported by these programs will constitute net new additions to employment, providing a macroeconomic buffer against worsening unemployment.

To obtain our empirical measures of jobs supported, we utilize the Department of Labor’s domestic Employment Requirements Matrix (ERM: BLS 2020) to estimate the number of jobs from spending in the IIJA and expected budget resolution. The ERM breaks down the economy into 206 sectors and tabulates the number of full-time jobs required for a given level of economic output in a given sector, as well as the jobs required to produce intermediate inputs (in other industries) that are used by that industry. We map the underlying policies proposed in the IIJA and the budget plan (Table 1) onto these 206 industries. We follow the work of Pollin and Chakraborty (2020) to analyze data on renewable energy and a range of other climate change-related investments. For other industries, we analyze data from the Bureau of Economic Analysis’s national income accounting input-output tables (BEA 2021).

Jobs supported by policy

Table 2 presents the results of our analysis, indicating the number of jobs that would be supported by the legislation under consideration. As the timing of this proposed spending is uncertain, we report the average number of jobs supported per year over the 10-year budgeting window, and we disaggregate the results to show the discrete contributions of each underlying policy. The results include both jobs supported directly in each industry in which spending occurs and jobs supported indirectly in industries that supply key inputs required for production in the primary industry.

In total, these policies would support more than 4.0 million jobs annually, with 772,400 jobs supported per year by the IIJA and more than 3.2 million jobs supported per year by budget reconciliation. To be clear, these average annual number of jobs supported cannot be summed together over 10 years. If, for example, all of the spending ramped up in Year 1 and then persisted, then 4.0 million jobs would be supported in the first year and then this number would persist but not grow. Over the 10-year window, one could cumulate these job numbers and classify them as “job-years”—a measure of total hours of work supported by this spending over the next decade.

A portfolio of investments in the Build Back Better agenda will tackle head-on the climate change crisis unfolding before our eyes, with efforts to mitigate and prepare for climate change that would support more than 763,000 jobs annually. These policies would combine investments in hard infrastructure with investments to develop new green technologies to make the economy cleaner and more efficient across a range of industries, and investments to make infrastructure, agriculture, and other key sectors of the economy resilient to potential climate-related disruptions. These measures include investments in electric vehicles infrastructure, public transit, power infrastructure and electric grids, environmental remediation and resilience, clean energy tax incentives, creation of a national infrastructure investment bank and a Civilian Conservation Corps, federal procurement of clean technologies, agriculture and forestry investments, weatherization upgrades to commercial and residential buildings, place-based clean energy economic development initiatives, and a range of research and development initiatives.

Another suite of policies would address America’s ongoing caregiving crises, with investments in child care, elder care, and early learning and development. Investments to make pre-K schooling universal and expand access to quality, affordable child care and long-term care would support 1.1 million jobs per year.

Table 2

Jobs supported by industry

Table 3 provides an alternative view of the analysis, summarizing the number of jobs supported by industry. To simplify interpretation of the results, we compile the detailed 206 industries identified by the Bureau of Labor Statistics into 23 larger sectoral groupings, with underlying detail provided for a selection of key manufacturing industries. Overall, the health care and social assistance sector would see the largest number of jobs supported, nearly 1.1 million annually, owing to significant new investments to expand access to quality health care, child care, and elder care services. The legislation would support 556,300 jobs annually in manufacturing industries and 312,200 jobs annually in construction industries, owing to the significant investments in physical infrastructure, electric vehicles, renewable energy generation, installation of new climate change-related technologies, and initiatives to strengthen critical manufacturing supply chains. The manufacturing industries poised to see the most benefit include electrical equipment, industrial machinery, construction products, fabricated metals, and motor vehicles and parts—together making up half of all manufacturing jobs supported by the policies.2

Table 3

Conclusion

Supporting more than 4 million jobs annually, the proposed Infrastructure Investment and Jobs Act, combined with Congress’s anticipated budget reconciliation, would provide a significant boost to America’s job market as it recovers from the pandemic economic shock and would sustain high-pressure labor markets critical to broadly rising wages. This could well turn out to be a vitally needed backstop to growth in coming years as the fiscal boost from the ARP winds down. These policies would accomplish much more than the immediate boost to employment, transforming the U.S. economy to be more efficient, equitable, sustainable, and prosperous for the long run.

Acknowledgments

The author would like to thank Zane Mokhiber and Jori Kandra for meticulous research assistance, Rob Scott and Josh Bivens for thoughtful input, and Krista Faries for editorial assistance.

Notes

1. The analysis here follows Congressional Budget Office (2021) methodology mandated by the Balanced Budget and Emergency Deficit Control Act of 1985 in assuming that expenditures authorized for less than the 10-year budgeting window will continue to be funded at the same level in each subsequent year.

2. These manufacturing industries include: Other electrical equipment, appliances, and components; HVAC and miscellaneous industrial machinery; architectural and structural products; boiler, tank, and shipping containers; other fabricated metal products; and motor vehicles and motor vehicle parts.

References

American Society of Civil Engineers (ASCE). 2021. A Comprehensive Assessment of America’s Infrastructure. March 3, 2021.

Ayres Steinberg, Sarah, and Adam Hersh. 2013. New Ryan Budget Cuts Investments in America’s Future. Center for American Progress, March 2013.

Berechman, Joseph, Dilruba Ozmen, and Kaan Ozbay. 2006. “Empirical Analysis of Transportation Investment and Economic Development at State, County and Municipality Levels.” Transportation, vol. 33, pp. 537–551, https://doi.org/10.1007/s11116-006-7472-6.

Bivens, L. Josh. 2017. The Potential Macroeconomic Benefits from Increasing Infrastructure Investment. Economic Policy Institute, July 2017.

Bivens, L. Josh. 2019. Updated Employment Multipliers for the U.S. Economy. Economic Policy Institute, January 2019.

Bivens, L. Josh, and Stuart A. Thompson. 2021. “179 Reasons You Probably Don’t Need to Panic About Inflation.” New York Times, August 18, 2021.

Buffie, Nick, and Bob Lord. 2021. The American Families Plan Taxes Billionaires and Protects Family Farms and Businesses. Center for American Progress, August 2021.

Bureau of Economic Analysis (BEA). 2021. Input-Output Accounts Data (Excel spreadsheets). Accessed August 6, 2021.

Bureau of Labor Statistics (BLS). 2020. Employment Requirements Matrix (Excel spreadsheets). Last modified June 10, 2020.

Bureau of Labor Statistics (BLS). 2021a. Current Employment Statistics National Databases. Accessed August 6, 2021.

Bureau of Labor Statistics (BLS). 2021b. Labor Force Statistics from the Current Population Survey. Accessed August 6, 2021.

Congressional Budget Office. 2021. Cost Estimate: Senate Amendment 2137 to H.R. 3684, the Infrastructure Investment and Jobs Act. August 1, 2021.

Economic Policy Institute (EPI). 2021. “Measuring the Job Shortfall Since February 2020: Actual and Counterfactual Employment, September 2019–August 2021” (chart). Economic Policy Institute.

Federal Reserve. 2021. “Industrial Production and Capacity Utilization – G.17” (monthly statistical release). Board of Governors of the Federal Reserve System, August 17, 2021.

Gale, William G., and Claire Haldeman. 2021. Searching for Supply-Side Effects of the Tax Cuts and Jobs Act. Brookings Institution, July 2021.

Glynn, Sara Jane. 2021. Breadwinning Mothers Are Critical to Families’ Economic Security. Center for American Progress, March 2021.

Gould, Elise, and Hunter Blair. 2020. Who’s Paying Now?: The Explicit and Implicit Costs of the Current Early Care and Education System. Economic Policy Institute, January 2020.

Gould, Elise, Marokey Sawo, and Asha Banerjee. 2021. “Care Workers Are Deeply Undervalued and Underpaid: Estimating Fair and Equitable Wages in the Care Sectors.” Working Economics Blog (Economic Policy Institute), July 16, 2021.

Heintz, James. 2010. “The Impact of Public Capital on the U.S. Private Economy: New Evidence and Analysis.” International Review of Applied Economics 24, no. 5: 619–632, https://doi.org/10.1080/02692170903426104.

Hersh, Adam S., and Mark V. Paul. 2021. Room to Run: America Has Ample Fiscal Space and Should Use It to Tackle Pressing Economic and Climate Challenges. Groundwork Collaborative, April 2021.

Hungerford, Thomas. 2011. Changes in the Distribution of Income Among Tax Filers Between 1996 and 2006: The Role of Labor Income, Capital Income, and Tax Policy. Congressional Research Service, December 2011.

Palladino, Lenore, and Chirag Lala. 2021. The Economic Effects of Investing in Quality Care Jobs and Paid Family and Medical Leave. Political Economy Research Institute, June 2021.

Political Economy Research Institute (PERI). 2021. Green Economy Transition Programs for U.S. States. February 2021.

Pollin, Robert, and Shouvik Chakraborty. 2020. Job Creation Estimates Through Proposed Economic Stimulus Measures. Political Economy Research Institute (PERI), September 2020.

Pollin, Robert, Shouvik Chakraborty, and Jeanette Wicks-Lim. 2021. Employment Impacts of Proposed U.S. Economic Stimulus Programs: Job Creation, Job Quality, and Demographic Distribution Measures. Political Economy Research Institute (PERI), March 2021.

Savage, Sarah Ann. 2019. High-Quality Early Child Care: A Critical Piece of the Workforce Infrastructure. Federal Reserve Bank of Boston, May 2019.

Sherman, Arloc, Ali Safawi, Zoë Neuberger, and Will Fischer. 2021. Recovery Proposals Adopt Proven Approaches to Reducing Poverty, Increasing Social Mobility. Center on Budget and Policy Priorities, August 2021.

Troise, Damian J. 2019. “US Companies’ Tax Windfall Fuels Record Share Buybacks.” AP News, April 4, 2019.

White House. 2021a. “Fact Sheet: The American Jobs Plan.” March 31, 2021.

White House. 2021b. “Fact Sheet: The American Families Plan.” April 28, 2021.

White House. 2021c. “Fact Sheet: Historic Bipartisan Infrastructure Deal.” July 28, 2021.

White House. 2021d. “Updated Fact Sheet: Bipartisan Infrastructure Investment and Jobs Act.” August 2, 2021.

Zandi, Mark, and Bernard Yaros, Jr. 2021. Macroeconomic Consequences of the Infrastructure and Budget Reconciliation Plans. Moody’s Analytics, July 2021.

TABLE OF CONTENTS

Reforming Unemployment Insurance

• Foreword
• Executive summary
• Introduction
• Primer
• Statement of the problem
• Section 1. Universal standards
• Section 2. Financing
• Section 3. Eligibility
• Section 4. Benefit Duration
• Section 5. Benefit levels
• Appendix

Key proposals

• Federal financing: Address chronic underfunding and incentives to reduce outlays at the expense of workers by paying for all regular and extended UI benefits with federal dollars. Absent a switch to a wholly federally financed system, reform tax rate calculations and state trust fund targets to strengthen support for workers and stabilize the system.
• Taxable wage base: To make UI’s finances fairer and more efficient, broaden the taxable wage base for unemployment insurance to be equal to the wage base for Social Security.
• Tax rate: In a federally financed system, reduce employer incentives to block worker claims by reforming “experience rating” to determine the employer tax rate based on changes in hours employees work, not unemployment insurance claims. In a federal–state financing system, reconsider “experience rating” and base state trust fund targets on industry-adjusted per capita targets, not “high cost multiple” systems that encourage states to slash awards to replenish trust fund accounts.
• Recessions: In a federally financed system, acknowledge the economy-boosting benefits of unemployment insurance benefits during recessions by paying for them with general revenue. In a federal–state financing system, use automatic triggers to provide federal financial assistance to states.
• Classification of workers: Require states to implement “ABC” tests that prevent employers from circumventing taxes by reclassifying employees as contractors, and require large businesses that use a lot of contractors to pay unemployment insurance tax on their contractors.

Introduction

The problem

States and the federal government jointly finance unemployment insurance (UI) in the United States through state and federal taxes imposed on wages and paid by employers, with a portion of those taxes likely passed through to workers in the form of lower pay. During normal economic times, states pay for the benefits their workers receive, and the federal government provides grants to pay for the cost of administering the program. The federal government also finances 50% of the Extended Benefits (EB) program.1 In recent recessions, it has paid 100% of EB costs as well as the costs for additional non-EB benefit extensions.

The way we pay for unemployment insurance benefits explains many of the program’s problems today. The desire to attract businesses and jobs puts pressure on states to keep their tax rates low. To do so, they limit benefits recipiency (the share of unemployed workers receiving benefits), decrease the amount of benefits workers receive, or both. Employers also have an incentive to make it harder for workers to claim benefits—for example, by filing challenges and appeals to worker claims—because the UI taxes firms pay rise when their workers collect UI benefits, a structure known as “experience rating.”

For both employers and states, pressure to reduce the benefits flowing to workers is especially intense when benefits are needed most: during and just after major recessions. When profit margins are slim, employers may be more likely to try to save money by keeping their UI tax rate as low as possible.

Federal law requires states to save their UI funds in a trust fund account and encourages them to keep adequate funds in it. During past recessions, these accounts have run empty in a number of states. To refill their accounts and get back into balance, many states slash benefits. In the wake of the Great Recession, for example, nine states cut the number of weeks individuals could get benefits from the traditional 26 to as few as 12 (Leachman 2015).2 As of April 2021, state legislators in at least three states (Iowa, Louisiana, and Tennessee) had proposed cutting the benefit duration to 12 weeks in response to low trust fund balances following the economic crisis caused by the COVID-19 pandemic (Golshan and Delaney 2021). The consequences of inadequate program financing fall squarely on the most disadvantaged workers, disproportionately affecting communities of color, as described in Section 3, covering eligibility.

The solution

Current financing of the UI system incentivizes employers and states to obstruct the delivery of UI benefits. Redesigning the way benefits are financed could create a system that would more effectively deliver UI.

Implementation

Financing UI benefits at the federal, rather than the state, level is the most logical way to ensure adequate funding for and the equitable distribution of UI benefits. If policymakers choose to leave benefit financing primarily to the states, however, significant improvements could still be made.

Federal policymakers could approximate “first-best” reforms by (a) raising the federal taxable wage base (which would require states to raise their own taxable wage bases), (b) reforming experience rating, (c) making federal financial assistance to states for funding UI benefits more predictable by tying it to automatic triggers, (d) ensuring that firms do not have incentives to outsource or misclassify workers as contractors, and (e) basing state trust fund targets on industry-adjusted per capita targets, not the current “high cost multiple” system, which encourages tax and benefit cuts.

Funding unemployment benefits with federal financing

In normal economic times, state UI benefits are the only support available to unemployed workers; during economic crises, they are often the only support preventing workers from falling into distress. But rather than ensure that these benefits are adequate, states compete in a race to the bottom, in an effort to attract firms. Employers shift production and workers to low-tax states (Guo 2020). They are especially likely to exploit differences across states in the taxation of part-time workers (U.S. GAO 1993). These moves hurt women, who are disproportionately likely to work part time, and people of color, who are disproportionately likely to work part time but want full-time work (Golden and Kim 2020).

Cutting taxes requires cutting benefits, making the race to the bottom evident in every aspect of the UI system (Galle 2019). After the 2009 recession, 10 states dramatically cut the duration of their UI benefits, and others failed to adjust the maximum benefit level for inflation for more than a decade, eroding benefits (CRS 2019; Smith, Wilson, and Bivens 2014). States that have most aggressively courted business by cutting benefits the most are also those states granting UI benefits to lower shares of separated workers (Desilver 2020; U.S. DOL-ETA 2021a; U.S. Census Bureau 2019). Indeed, historical data on the recipiency rate suggest that states tightening requirements in times of economic distress have contributed to subsequent declines in recipiency rates that have never ascended back to mid-last century highs exceeding 50%, and hover around 28% today (as of 2019) (U.S. DOL-ETA 2021b). By capitulating to employer threats to move jobs out of state, local officials may see themselves as winning a competition with their neighbors. But the result is a less effective automatic stabilizer to assist displaced workers during recessions, leaving the federal government holding the bag to pay for other safety-net programs.

Policy proposal: Fund regular unemployment benefits and the full cost of extended benefits with federal financing

No solution short of federal financing will fully overcome these problems; if states can compete with one another to lower taxes, they will. If their revenue is lower, they will find new ways to reduce recipiency rates. Following the Great Recession, for example, when federal law temporarily constrained states from reducing benefit amounts, they cut the duration of benefits (CRS 2019; Smith, Wilson, and Bivens 2014). Some states also erected new administrative barriers (Wentworth 2017). For example, a new online system in Massachusetts increased one form of benefit denial by 950%, and delayed claims processing by more than six weeks on average. Florida increased one category of procedural denials by 180%, and another by 400%.

Financing at least the cost of the minimum regular UI benefits we have described with federal money—building on the existing federal financing of program administration and extended benefits—could eliminate this race to the bottom.3 Removing this competition would remove the greatest barrier to adequate benefit financing.4 A federally financed regular UI benefit could be financed in several ways, as shown below.

Increasing the taxable wage base

Currently, states and the federal government finance UI almost exclusively through taxes on employers. The total tax an employer owes is based on the product of its tax rate and the taxable wage base, the portion of each employee’s salary that is considered when calculating the tax owed. For the current federal UI tax—the Federal Unemployment Tax Act (FUTA) tax—the federal government taxes only the first $7,000 of each employee’s wages. State wage bases cannot be lower than the federal wage base; they currently range from $7,000 to about $53,000, with most of them less than $15,000 (U.S. DOL-ETA 2020c). The taxable maximums for UI wage bases are far lower than the taxable maximum wage bases for other programs. For example, the wage cap for the Social Security portion of the Federal Insurance Contributions Act (FICA) for 2021 was $142,800. The UI federal wage base has failed to keep up with inflation; it was $3,000 in 1939, or more than $55,000 in 2021 dollars, almost eight times its current value of $7,000 (Gould-Werth 2020a).

Policy proposal: Set the taxable wage base for unemployment insurance equal to the wage base for Social Security

We propose increasing the federal UI taxable wage base to a level equal to the current Social Security taxable wage base and indexing it to inflation. Raising the wage base threshold would allow for much lower rates, which would be more economically efficient, because they would reduce the incentive to attempt to avoid the tax. If UI financing were shifted to the federal level and the taxable federal wage base broadened, revenues could increase without an accompanying rate hike.

A higher wage base cap would also better align the constituencies that pay the tax and draw benefits from it. For example, a worker in Vermont who earns $10,000 a year has taxes levied on 100% of her wages and has 100% of her wages considered when setting benefits, whereas a worker who earns $50,000 has taxes levied on only 28% of her earnings, but has 92% of her wages used when determining her benefit level (U.S. DOL-ETA 2020b; Vermont DOL 2021). Such a system thus imposes a greater relative burden on the lower-wage worker.

Raising the wage cap would also reduce the incentive to prefer a small number of high-earning workers over a large number of low earners. With a low cap, an employer incurs greater tax liability if she hires two $20,000 earners than if she hires one $40,000 earner. For instance, in Vermont, where there is a 2021 cap of $14,100, an employer with a 1% rate would pay $141 in state UI tax for the $40,000 earner, but $282 for the two $20,000 earners. The system thus disincentivizes employers to employ low-wage workers.5

Under the current federal–state UI system, state wage bases cannot be lower than the federal wage base by law. Raising the federal UI cap to the Social Security cap would therefore also raise all state caps, generating all of the benefits described above.6

Changing the tax rate formulation

Average UI tax rates are low in the United States. The federal rate is 0.6%,7 and most state rates average 1%–2% (U.S. DOL-ETA 2020a). Employers pay the federal government $42 a year for each full-time, year-round employee (0.6% times the $7,000 federal wage base). In Missouri, a fairly low-tax state, the average employer would pay another 1% tax on the state wage base of $11,500, for a state-tax total of $115 per worker. Nationwide, the total average tax was about $267 per worker in 2020; the average over several recent years was a bit higher, at $350 (U.S. DOL-OUI 2021; Pavosevich 2020).

Within a state, different employers face different tax rates, depending on the employer’s experience with unemployment—a method of setting the tax rate known as “experience rating.” In all U.S. jurisdictions except Alaska, the rate is based on the UI benefits collected by separated workers. Every benefit recipient must be traced back to one (or more) previous employers. Employers with a higher ratio of benefits awarded to payroll pay higher rates than others up to a cap.8 In Alaska, employers instead pay a higher tax when they cut payroll, regardless of whether their employees successfully claim benefits.

The basic logic of experience rating is similar to the rationale for adjusting auto or home insurance rates upward after the insured files a claim. Layoffs impose costs on society, including the expense of providing UI benefits. Without experience rating, employers would not take account of these costs, leading to excessive job turnover (Alessie and Bloemen 2004; Karni 1999). There is strong evidence that experience rating reduces layoffs as well as seasonal employment (Albertini and Fairise 2018; Anderson and Meyer 2000; Baicker, Goldin, and Katz 1998; Ratner 2013; Woodbury 2004).

As currently designed, though, experience rating also encourages employers to prevent their employees from receiving adequate benefits. Benefit denials increased by 50%–66% in Washington state after it adopted experience rating (Anderson and Meyer 2000). Employer tactics include failing to provide information about benefits; encouraging employees not to file; challenging employee claims;9 or structuring their workforce to reduce eligibility, such as by using part-time workers or contractors, and shifting job sites to states where taxes are lower and benefits less generous (Anderson and Meyer 2000; de Raaf, Motte, and Vincent 2005; Gould-Werth 2016; Hyatt and Kralj 1995; Thomason and Pozzebon 2002; Vroman et al. 2017).

Grocery worker fired for taking health leave is initially denied benefits

Justin Grevencamp, Maine

I worked at an organic grocery-cafe in Brewer for two years. It was a beloved, locally owned small business. I developed new recipes and handled important catering orders. I took a two-week unpaid leave to deal with health issues, and when I returned I was told there was no more work for me. When I filed for UI, the store owners told the agency I had quit. At the appeal, one owner was upset that their taxes would go up because I had claimed UI. I received my benefits, but I was stunned that people I considered friends had tried to block me from receiving them.

Policy proposal: In a federally financed system, base the unemployment insurance tax rate on changes in hours employees work, not unemployment insurance claims

Because the U.S. lacks the employment protections common in most other developed economies, there is a strong case for a system that still imposes costs on employers when they terminate workers but does not encourage employers to prevent their workers from obtaining unemployment benefits. We suggest removing incentives to prevent UI claims for separated workers by using the full-time equivalent (“FTE”) or “hours-worked variation” experience rating method, similar to the method used in Alaska.10 In the FTE method, employers are rated based on how much the hours their employees worked changed, on a quarterly basis, over a three-year period. Increases in hours are not given as much weight as decreases.

Because employers are charged based on changes to payroll rather than benefit claims, these methods should make employers indifferent about whether their separated workers obtain benefits.11 Notably, until the tenure of its latest governor, Alaska’s recipiency rate was among the highest in the country (as of 2020, it had declined to around the national mean).

To calculate their tax rate in a given quarter under the FTE method, employers compute the average of the total hours worked in each of the 12 preceding quarters. They then calculate the change from each of these quarters to the next, dividing this change by the average quarterly total hours over the 12-quarter measuring period.12 For increases, the resulting product is discounted by a special weighting factor. The taxpayer’s final rating factor is the average of these 12 changes over the three-year rating period.13 Employers are then ranked and tax rates assigned as in existing experience rating systems. For more details, refer to the text box, “Calculating the tax rate under the FTE method.” Under federal financing, employers should be ranked regionally (by state or Metropolitan Statistical Area, for example), rather than nationally, to account for local variations in industry, seasonality, and other economic conditions.

Using hours worked as the base unit of measure offers a few modest advantages over Alaska’s method (based on payroll) or one based on the number of employees. Using the number of employees on the payroll cannot account for reduction in hours. It either weights full- and part-time employees equally or discounts employees who work part time, allowing employers to game the system by reducing work hours. Payroll-based systems may favor high-wage over low-wage workers, rewarding business models with a small number of highly skilled employees (Vroman et al. 2017). Such a system could result in lower levels of employment for less-educated workers.

Many states do not presently collect hours-worked data, but former U.S. Department of Labor officials have described this as only a minor obstacle (Miller and Pavosevich 2019). As explained later, expanding state collection of hours data would have other advantages as well, such as improving the data available through the Quarterly Census of Employment and Wages, and allowing for implementation of our proposal to base worker eligibility on hours, not wages.

Policy proposal: In an experience rating system, penalize employers who consistently remain at the tax rate cap

In contemporary experience rating systems, some firms are indifferent to additional layoffs because the tax rate is capped—that is, the rate they pay for UI will not rise if additional workers are laid off. We recommend adopting the suggestion of Vroman et al. (2017) of imposing a penalty on firms that persistently remain at the cap (for three years or more, for example).

Policy proposal: In a federal–state financing system, base state trust fund targets on industry-adjusted per capita targets

In a system with significant state financial contributions, state savings targets should be reformed to be based on industry-adjusted per-capita targets, not the current “high cost multiple” system that favors cutting benefits rather than raising revenues when trust fund balances are depleted. Currently, in good times, states deposit excess UI revenues in a trust fund account, to be drawn on during recessions. Regulations encourage states to maintain an adequate balance, based on a calculation known as the “average high cost multiple” (AHCM). Because the AHCM is based on expected benefit awards, states can achieve their fiscal targets by slashing awards instead of adding revenue (ACUC 1996).14

Thus, in a system with significant state contributions, the AHCM should be replaced with a system in which state savings targets are based on industry-adjusted per capita targets. That is, state trust fund adequacy would be calculated using nationwide projected benefit costs per capita multiplied by the state population (see Galle 2019 for a complete description). This figure could then be modified to reflect each state’s mix of local industries and those industries’ historic nationwide turnover rates.

It may also be desirable to eliminate experience rating in a system that is mostly state financed. Although a portion of UI taxes are passed along to workers in the form of lower wages, in a competitive industry an employer cannot offer lower wages than its rivals, so that with experience rating higher-taxed businesses must bear the costs themselves (Gruber 1997; Anderson and Meyer 2000). This encourages business owners to seek out lower rates, contributing to the race to the bottom that federalizing the financing system would avoid.

Paying for unemployment benefits during recessions

UI payouts during recessions provide large positive spillovers to the economy, including to workers and businesses that do not draw on UI (Elmendorf 2011). The cost of financing this economic stimulus should therefore be widely shared, though the method for doing so may vary depending on whether UI is wholly federally financed or is partly left to states.

Policy proposal: In a federally financed system, pay for benefits with general revenue during recessions

If funding for UI is made wholly federal, we recommend that benefits be paid out of general federal revenues during recessions, as they have been in recent episodes. As with Social Security, there are economic and political advantages to “benefit taxation,” in which taxes are explicitly linked to a particular social insurance product. In recessions, however, the UI system should be able to borrow from general Treasury funds without having to repay its borrowing via higher UI taxes later.

Policy proposal: In a federal–state financing system, use automatic triggers to provide federal financial assistance to states for unemployment insurance

Even if states are left to bear some of the costs of standard UI during normal times, the federal government should pay all of the cost during recessions. Federal dollars already pay for half of extended benefits, and in recent recessions that has been expanded to 100%, so extending federal support to base-period UI is not a dramatic shift and is economically justified. Recessions hit some regions harder than others; national financing allows states to share risk. States also face structural obstacles to countercyclical spending (Galle 2019). For example, it is politically difficult to maintain reserve funds at the state level, because today’s taxpayers may be living or doing business elsewhere when the rainy day arrives.

Federal financing for base-period UI can be expanded without major changes to the current structure. Following the Great Recession, Congress eventually granted states significant financial relief. Under current law, states accessing federal financing must risk the possibility that they will face swift repayment obligations as well as significant interest and added charges before Congress eventually provides them with relief. To eliminate this risk, we recommend that when EB is triggered (as described in Section 5, benefits duration), states become automatically eligible for low-interest or interest-free financing for base-period UI as well, repayable over a period that does not coincide with the state’s recovery from its downturn. Even greater federal support is possible; for example, severe downturns could trigger automatic loan forgiveness. That would in effect offer federal reinsurance to state trust funds, relieving some of the financial pressure on them to compete over low benefits.

Including self-employed and misclassified workers

State and federal UI taxes apply only to the wages of employees. To circumvent the tax, many employers therefore claim that their workers are contractors, not employees. Legal tests to distinguish genuine employees from contractors are usually complicated and thus costly and time-consuming to enforce. Other employers respond to UI incentives through actual (not merely nominal) changes in their workforce, shifting traditional full-time employees into more-precarious, less-rewarding contract work. The rise of large digital platform-based businesses, such as Airbnb and Uber, has drawn a growing attention to the plight of workers in these types of positions (Ravenelle 2019).

Policy proposal: Require states to implement the “ABC” test

To ensure that enterprises pay their fair share of UI taxes, rather than shift their burdens to others, federal law should make it more difficult for employers to recharacterize taxable wages as untaxed contractor payments. An important starting point is to require states to implement a version of the “ABC” test. Under this legal rule, a service provider to a business is presumed to be an employee unless the individual (a) is free from the direction and control of the business, (b) provides labor outside the usual course of the business, and (c) is customarily engaged in their own independently established business. A few states already apply versions of the ABC test, at least for select industries (Doroghazi 2019). It should be a national gold standard. Alternative legal approaches, such as the 20-factor balancing test applied by the IRS, have proven to be unpredictable, highly manipulable, and expensive and burdensome to administer.

Policy proposal: Require large 1099 payers to pay unemployment insurance tax on their contractors

As an additional disincentive to misclassification and efforts to replace the security and benefits of a steady job with the uncertainty of contractor status, we recommend taxing contractor payments at the same rate as employee hours. Imposing UI taxes on contractor payments would also bring in contributions from a self-employed population that, under our proposal in the benefits chapter, would receive greatly expanded benefits under the jobseeker’s allowance (See Section 3, on eligibility, in this report).

To reduce the administrative burden on small businesses and households, only relatively large businesses would be subject to the UI tax, and even they would be required to pay it on the wages of only some workers. Individuals who receive only very small payments from a given payer or who provide services only for the nonbusiness purposes of the payer would be exempt. The UI tax would be imposed only on payments to individuals for whom the payer is obligated to issue a Form 1099 (currently contractors paid at least $600 a year for the payer’s business purpose). For simplicity, only large firms, such as companies with more than $1 million in annual payroll or profits, might be required to pay the UI tax on contractors. If this threshold is considered too low, an additional threshold of 50 total combined employees and contractors could be applied. Both of these measures could be phased in, with the UI tax on 1099 earners starting at 10% of the full tax rate and stepping up 10% for each additional $1 million in payroll/profits to a cap of 100%. If a worker threshold is also used, the rate could step up by 10% per 25 workers, with the actual tax rate imposed being the greater of the payroll/profit- or worker-determined rate. Whichever threshold is used, related entities such as companies that share the same owners would be counted together when determining whether the firm crosses the threshold, to prevent gaming the system.

Endnotes

1. According to the U.S. Department of Labor, “Extended Benefits are available to workers who have exhausted regular unemployment insurance benefits during periods of high unemployment. The basic Extended Benefits program provides up to 13 additional weeks of benefits when a State is experiencing high unemployment. Some States have also enacted a voluntary program to pay up to seven additional weeks (20 weeks maximum) of Extended Benefits during periods of extremely high unemployment. Extended Benefits may start after an individual exhausts other unemployment insurance benefits (not including Disaster Unemployment Assistance or Trade Readjustment Allowances). Not everyone who qualified for regular benefits qualifies for Extended Benefits…. The weekly benefit amount of Extended Benefits is the same as the individual received for regular unemployment compensation. The total amount of Extended Benefits that an individual could receive may be fewer than 13 weeks (or fewer than 20 weeks).” U.S. DOL-ETA 2021c.

2. The states were Arkansas, Florida, Georgia, Illinois, Kansas, Michigan, Missouri, North Carolina, and South Carolina.

3. We do not mean to rule out the possibility that states would voluntarily impose their own taxes to pay for benefits more expansive than those we propose. Added benefits might also be fully or partially federally funded.

4. If the taxable wage base were raised, workers in high-wage states would cost their employers more in UI taxes than workers in low-wage states. But equal tax rates would ensure that this difference was driven by labor costs, not taxes. The marginal dollar of salary would cost the same no matter what state it was paid in.

5. The other side of this argument is that raising the cap could discourage pay raises and potentially full-time work. In many jurisdictions, the cap is now so low that it is effectively a fixed tax per employee, leaving employers indifferent to wage levels. (For a summary of the evidence that rated taxes affect only the “extensive,” or hiring, margin, see Guo and Johnston 2020.) With a higher cap, pay raises potentially lead to a UI tax increase. Depending on other UI rules, this principle can also translate into a preference for part-time employees: Once it is no longer advantageous to hire one $40,000 salary worker instead of two $20,000 salary workers, an employer might split a single position into two, perhaps to take advantage of other provisions that turn on full-time versus part-time status (Guo and Johnston 2020).

6. If states continue to pay UI benefits through state trust funds, expanding the wage base could help strengthen state finances (ACUC 1996; Vroman et al. 2017). Although states could cut rates in response to an expansion of their wage base, analysts predict that many states would still bring in additional revenue on net (CBO 2012).

7. The full federal rate is 6%, but it is reduced to 0.6% in states that are in compliance with federal law. The Department of Labor has never found a state to be so out of compliance, so the 6% rate has never been triggered.

8. Most states still use an antiquated variant (the “reserve ratio” method), in which long-standing employers can become insensitive at the margin to further changes in employment. Under this system, high-turnover firms are strongly incentivized to outsource work to contractors (Pavosevich 2020).

9. It does not appear that there are any strong reasons to encourage employers to challenge employee benefits. McLeod and Malcolmson (1989) argue that if employers are not rated based on their employees’ benefits, they will not be incentivized to share with government officials information that could indicate a fraudulent claim. We could not identify any evidence that this is the case. Moreover, fraud reduction is small compared with the direct impacts of experience rating. For example, recipiency rates vary dramatically between high- and low-recipiency states, with workers in high-recipiency states receiving benefits more than twice as often as workers in low- recipiency states (Desilver 2020). It is unlikely that differences in fraud detection explain a meaningful portion of these differences.

10. Our proposal builds on work by Vroman et al. (2017) and Miller and Pavosevich (2019). Miller and Pavosevich (2019) propose alternative systems modeled on Alaska’s payroll decline method. Vroman et al. (2017) conclude that none of the alternatives they modeled was clearly superior to the best options now in practice, but they did not consider the refinements Miller and Pavosevich propose, such as the use of the “average quarterly hours” measure we describe below.

11. Municipal and nonprofit employers often are not formally experience rated, but instead may opt to “self-insure,” which means that they pay into the system only when their employees claim benefits. Our proposal would therefore leave these employers with strong incentives to prefer lower benefits and to contest claims. It is likely all employers should be taxed by the same rules.

12. Although the use of average quarterly total hours—as Miller and Pavosevich (2019) propose—adds some complexity, it solves a key problem Vroman et al. (2017) identify. Consider a 10-worker firm that lays off all 10 workers in Q1 and rehires them in Q3. There is a 100% decline from Q1 to Q2 and an infinite percentage increase between Q2 and Q3. If the firm lays off and rehires only nine workers, there is a 90% reduction between Q1 and Q2 and a 900% increase between Q2 and Q3. Both these scenarios would produce highly inaccurate results for the rehiring employers. Using average total hours smooths these changes, so that in the second scenario, for instance, there is about a 130% reduction (9/[(10+1+10)]/3)) in the number of employees followed by a 130% increase. Applying the discount factor, W, to the increase would result in a net reduction in hours, reflecting the fact that the firm made some use of the UI system.

13. For small employers, the quarterly components could be weighted by the quarter’s total hours, a step that would tend to reduce the volatility of the measure (Vroman et al. 2017).

14. For example, suppose a state is projected to average $1 billion in UI spending during high-cost periods but it has only $750 million in its trust fund. A target AHCM of 90% would require the state to hold at least $900 million in its trust fund. If the state cuts benefits, its projected spending might fall to $800 million, allowing the target trust fund balance to be only $720 million. The state has cut its way to its target.

References

Advisory Council on Unemployment Compensation (ACUC). 1996. Collected Findings and Recommendations: 1994–1996. Washington, D.C.: Advisory Council on Unemployment Compensation.

Albertini, Julien, and Xavier Fairise. 2018. “Layoffs, Recalls and Experience Rating.” TEPP Working Paper 2018-10.

Alessie, Rob, and Hans Bloemen. 2004. “Premium Differentiation in the Unemployment Insurance System and the Demand for Labor.” Journal of Population Economics 17, no. 4 (December): 729–765.

Anderson, Patricia M., and Bruce D. Meyer. 2000. “The Effects of Unemployment Insurance Payroll Taxes on Wages, Employment, Claims and Denials.” Journal of Public Economics 78, nos. 1–2 (October): 81–106 .

Baicker, Katherine, Claudia Goldin, and Lawrence Katz. 1998. “A Distinctive System: Origins and Impact of U.S. Unemployment Compensation.” In The Defining Moment: The Great Depression and the American Economy in the Twentieth Century, edited by Michael D. Bordo, Claudia Goldin, and Eugene N. University of Chicago Press.

Congressional Budget Office (CBO). 2012. Unemployment Insurance in the Wake of the Recent Recession. Publication no. 4525, November 2012.

Congressional Research Service (CRS). 2019. Unemployment Insurance: Consequences of Changes in State Unemployment Compensation Laws. R41859, updated October 23, 2019.

Elmendorf, Douglas W. 2011. “Policies for Increasing Economic Growth in 2012 and 2013.” Testimony of Douglas W. Elmendorf, Director, Congressional Budget Office, before the Committee on the Budget, United States Senate, November 15, 2011.

de Raaf, Shawn, Anne Motte, and Carole Vincent. 2005. “A Literature Review of Experience-Rating Employment Insurance in Canada.” SDRC Working Paper Series No. 05–03, May 2005.

Desilver, Drew. 2020. “Not All Unemployed People Get Unemployment Benefits; In Some States, Very Few Do.” FactTank Blog (Pew Research Center), April 24, 2020.

Doroghazi, Lauren. 2019. Beyond California AB 5 – States Address Independent Contractors. MultiState, November 2019.

Galle, Brian. 2019. “How to Save Unemployment Insurance.” Arizona State Law Journal 50: 1009–1064.

Golden, Lonnie, and Jaeseung Kim. 2020. Underemployment Just Isn’t Working for U.S. Part-Time Workers. CLASP, May 2020.

Golshan, Tara, and Arthur Delaney. 2021. “The Looming Showdown Over Unemployment Benefits.” HuffPost, April 12, 2021.

Gould-Werth, Alix. 2016. “Workplace Experiences and Unemployment Insurance Claims: How Personal Relationships and the Structure of Work Shape Access to Public Benefits.” Social Service Review 90, no. 3 (June): 305–352.

Gould-Werth, Alix. 2020a. “Fool Me Once: Investing in Unemployment Insurance Systems to Avoid the Mistakes of the Great Recession During COVID-19.” Washington Center for Equitable Growth, April 30, 2020.

Gould-Werth, Alix. 2020b. The Long-Run Implications of Extending Unemployment Benefits in the United States for Workers, Firms, and the Economy. Washington Center for Equitable Growth, December 2020.

Gruber, Jonathan. 1997. “The Incidence of Payroll Taxation: Evidence from Chile.” Journal of Labor Economics 15, no. 53 (July): S72–S101.

Guo, Audrey, 2020. The Effects of Unemployment Insurance Taxation on Multi-Establishment Firms. MPRA Paper 97919, University Library of Munich, Germany.

Guo, Audrey, and Andrew C. Johnston. 2020. “The Finance of Unemployment Compensation and Its Consequence.” IZA Discussion Papers 13330, Institute of Labor Economics (IZA). June 2020.

Hyatt, Douglas E., and Boris Kralj. 1995. “The Impact of Workers’ Compensation Experience Rating on Employer Appeals Activity.” Industrial Relations 34, no 1 (January): 95–106. https://doi.org/10.1111/j.1468-232X.1995.tb00362.x.

Karni, Edi. 1999. “Optimal Unemployment Insurance: A Survey.” Southern Economic Journal 66, no. 2 (October): 442–465.

Leachman, Michael. 2015. “States Cutting Jobless Benefits Hadn’t Adequately Prepared for Recession.” Off the Charts Blog (Center on Budget and Policy Priorities), May 28, 2015.

McLeod, W. Bentley, and James M. Malcolmson. 1989. “Implicit Contracts, Incentive Compatability, and Involuntary Unemployment,” Econometrica 57: 447–480.

Miller, Michael, and Robert Pavosevich. 2019. “Alternative Methods of Experience Rating Unemployment Insurance Taxes.” Public Budgeting & Finance 39, no. 4 (Winter): 28–47.

Organisation for Economic Co-Operation and Development (OECD). 2019. The OECD Tax-Benefit Model for Canada: Description of Policy Rules for 2019.

Pavosevich, Robert. 2020. “The Cost of Layoffs in Unemployment Insurance Taxes.” BLS Monthly Labor Review, April 2020.

Ratner, David D. 2013. “Unemployment Insurance Experience Rating and Labor Market Dynamics.” Board of Governors of the Federal Reserve System Finance and Economics Discussion Series No. 2013-86, December 2013.

Ravenelle, Alexandrea J. 2019. Hustle and Gig: Struggling and Surviving in the Sharing Economy. Oakland: University of California Press.

Smith, Joshua, Valerie Wilson, and Josh Bivens. 2014. State Cuts to Jobless Benefits Did Not Help Workers or Taxpayers. Economic Policy Institute Briefing Paper #380, July 2014.

Thomason, Terry, and Silvanna Pozzebon. 2002. “Determinants of Firm Workplace Health and Safety and Claims Management Practices.” Industrial and Labor Relations Review 55, no. 2 (January): 286–307.

U.S. Census Bureau. 2019. “Table DP05. 2019 American Community Survey 1-Year Estimates Selected Population Profiles.” Accessed June 2021.

U.S. Department of Labor Employment and Training Administration (U.S. DOL-ETA). 2020a. “Estimated Employer Contribution Rates Calendar Year 2020” (data table). September 2020.

U.S. Department of Labor Employment and Training Administration (U.S. DOL-ETA). 2020b. “Monetary Entitlement” in Comparison of State Unemployment Laws 2020.

U.S. Department of Labor Employment and Training Administration (U.S. DOL-ETA). 2020c. “Significant Provisions of State Unemployment Insurance Laws,” January 2020.

U.S. Department of Labor Employment and Training Administration (U.S. DOL-ETA). 2021a. UI Replacement Rates Report (online database). Accessed June 8, 2021.

U.S. Department of Labor Employment and Training Administration (U.S. DOL-ETA). 2021b. Unemployment Insurance Chartbook (online database), “Category 12 Regular Program Insured Unemployment as a Percent of Total Unemployment.” Accessed June 2, 2021.

U.S. Department of Labor Employment and Training Administration (U.S. DOL-ETA). 2021c. “Unemployment Insurance Extended Benefits” (web page), accessed May 14, 2021.

U.S. Department of Labor Office of Unemployment Insurance Division of Fiscal and Actuarial Services (U.S. DOL-OUI). 2021. State Unemployment Insurance Tax Measures Report 2020. April 2021.

U.S. General Accounting Office (U.S. GAO). 1993. GAO/HRD-93-107, Unemployment Insurance: Program’s Ability to Meet Objectives Jeopardized. Report to the Chairman, Senate Committee on Finance, U.S. Senate, September 1993.

Vermont Department of Labor (Vermont DOL). 2021. Quarterly Reporting & Taxable Wage Information (online database). Accessed May 21, 2021.

Vroman, Wayne, Elaine Maag, Christopher O’Leary, and Stephen Woodbury. 2017. A Comparative Analysis of Unemployment Insurance Financing Methods. Washington, D.C. Department of Labor, Chief Evaluation Office, prepared by the Urban Institute, December 2017.

Wentworth, George. 2017. Closing Doors on the Unemployed: Why Most Jobless Workers Are Not Receiving Unemployment Insurance and What States Can Do About It. National Employment Law Project, December 2017.

Woodbury, Stephen A. 2004. Layoffs and Experience Rating of the Unemployment Insurance Payroll Tax: Panel Data Analysis of Employers in Three States. October 2004.

TABLE OF CONTENTS

Reforming Unemployment Insurance

• Foreword
• Executive summary
• Introduction
• Primer
• Statement of the problem
• Section 1. Universal standards
• Section 2. Financing
• Section 3. Eligibility
• Section 4. Benefit Duration
• Section 5. Benefit levels
• Appendix

By Rebecca Dixon and William Spriggs

The unemployment insurance system is the country’s only automatic income support program for individuals who have lost work. It kicks in automatically when job losses start—without the delays and political and policy wrangling about if, when, or how to respond. The program serves as a key stabilizer during economic downturns by buttressing consumer spending, demand, and sentiment—preventing people’s fears of job loss from constricting private demand for spending on goods and services faster than the initial job losses.

As we saw these past 15 months, however, when so many businesses had to shut down to help stop the spread of COVID-19, the unemployment insurance program has big holes in who normally gets benefits. Faced with massive job losses, the program would have proven an insufficient automatic stabilizer because of its huge gaps in coverage. Thankfully, Congress acted quickly to plug (albeit temporarily) some of those holes, and unemployment insurance served as the primary vehicle to help people who lost work and were left with no income. Since the start of the pandemic, the program has injected hundreds of billions of dollars into the U.S. economy, keeping millions of families in their homes and out of poverty.1

Because unemployment insurance is a federal–state hybrid program, states have broad flexibility to determine eligibility and benefit levels. This has led to notable coverage variations across states. For example, southern states with large Black populations have among the lowest benefit payments.

The UI program’s shortcomings, laid bare by the pandemic, meant that the injection of income support was not automatic for the millions of workers locked out of the program by outdated eligibility rules; it was not automatic for people caught in their state’s “race to the bottom” to slash benefits. But when the scale of the pandemic’s unemployment crisis became clear, Congress had to step in, to supplement benefit amounts and expand the program temporarily to cover workers not normally eligible for benefits due to their nonemployee status or because they worked part time or for low wages.

In the wake of lessons learned from the pandemic, unemployed workers and their advocates are speaking up, organizing, and demanding long-term reforms to fundamentally transform the unemployment system. This report provides a brief history of race- and gender-based exclusions and analyzes the program’s coverage, performance, and revenue base, pinpointing gaps and offering a vision of a revamped UI system with minimum federal standards that will more effectively and expansively provide income support to unemployed workers while stabilizing the macroeconomy.

The unemployment insurance system was conceived during the Great Depression in the 1930s by Labor Secretary Frances Perkins as part of President Franklin D. Roosevelt’s New Deal. As the program was created in the context of a broader system of patriarchy and racial hierarchy, Congress made compromises in its design that were laser-focused on protecting the earnings of white male breadwinners in manufacturing. Congress also compromised to give states huge discretion in implementing the program so they could have control to mirror their practices of racial exclusion. The program’s architects designed a system that attempted to reconcile opposing goals: to support the intended workers while excluding enough other workers to get it across the finish line in Congress. Federal unemployment insurance law has never meaningfully addressed those exclusions.

The composition of the U.S. workforce has shifted enormously since the 1930s. In just the past two decades, the United States lost about 4 million manufacturing jobs while gaining about 4 million food service jobs. In February 2020, there were roughly the same number of manufacturing workers as restaurant workers. But the unemployment insurance system, designed for full-time, higher-wage manufacturing workers, was never amended or adapted to industries where many workers are part time and are paid poverty wages. The almost 6 million payroll positions lost in March and April 2020 in food services was greater than the entire nondurable manufacturing workforce in February 2020 (BLS-CES 2021).

Clearly, the rules about who has access to unemployment insurance need to catch up to our racial and gender equity values and to the realities of working people and families today.

On the revenue side, the financing of the unemployment insurance system no longer matches the source of labor market collapses. In the 1930s, the state unemployment insurance systems on which the federal program was modeled taxed companies with an eye to swings in inventories. Manufacturers would hire workers when the economy expanded, guess wrong on the strength of growth, and then, when inventories piled up from weak sales, lay workers off.

Today, the companies that create excessive churn in the labor market (relying on low wages and high labor turnover rates) often do not pay more into the system, because workers under this style of low-road management—temporary, contract staffing, part-time, and low-paid workers—too often do not qualify for unemployment insurance. Since the 1980s, recessions have been more related to financial markets than inventory cycles, and they have been more severe than state unemployment trust funds were designed to handle. The disruptions are greater, causing more permanent job losses than cyclical temporary layoffs. The average duration of unemployment is trending upward, and it takes longer to return to previous peak employment levels (Freeman 2013). For Black workers who face systemic employment discrimination, the recovery is always longer.

Unfortunately, the system has no protection for workers in this “race to the bottom.” Women and Black, Latinx, Indigenous, and immigrant workers will bear the brunt of the pain unless we transform the system. Black and Latinx workers have lower recipiency rates than white workers, despite facing higher unemployment rates. Women have a lower recipiency rate than men, because they are more likely to work in low-paid, part-time dominated industries (Nichols and Simms 2012). Every severe downturn mars the early earnings history of young people, disrupts women’s labor force participation, and makes it harder to return to the labor market.2 Race, gender, and intergenerational equity fault lines in the unemployment insurance program must be addressed.

The politics around unemployment insurance are another nagging problem. Since 1980, with each new downturn drawing down state unemployment trust fund balances, conservative state lawmakers and policymakers have responded not by replenishing funds but by slashing eligibility and benefits. Already, as this report is being released, a conservative “revolt” has taken place, with half of states announcing plans to “secede” from the federal government’s expanded unemployment insurance programs. These misguided efforts will remove a half billion dollars a week from the nation’s economic recovery, while deepening inequities felt by workers of color and hurting families that need the aid to get by (Stettner 2021). Some of these states are now considering legislation to cut the duration of benefits and erect more barriers to benefits—further weakening the system and leaving those states ill-prepared for the next downturn.

In the wake of the COVID-19 crisis, Congress swiftly enacted changes to expand the program and improve eligibility and benefit sufficiency. Now the challenge is how to learn from those lessons and adopt permanent changes to strengthen the unemployment insurance system. The urgency of understanding the shortfalls and reforming the system are clear. U.S. payrolls remain down over 8 million jobs from their February 2020 peak (Gould 2021). Even if the economy generates a record-setting 1 million jobs each month until September, Labor Day will arrive with millions facing long-term unemployment and having exhausted their lifeline of unemployment benefits. Others will find themselves defined out of being eligible, as their low earnings and part-time status will not qualify them for benefits; but for the expiring Pandemic Unemployment Assistance program, they would have had nothing. The withdrawal of support will complicate, if not halt, the economic recovery.

This report examines the problems that resulted in an unemployment benefit system that struggled to rise to the crisis we faced in 2020 and will struggle again as states end emergency benefits, and when federal pandemic unemployment provisions expire in September. In addition, this report proposes critical first steps on the road to systemic unemployment insurance reform—fixes that state legislatures and governors can enact to ensure equity and adequate standards regarding just financing, eligibility, duration of benefits, and benefit levels and amounts—so that no one is left behind.

The Biden-Harris administration has proposed significant reforms to the unemployment insurance system in its fiscal year 2022 budget. Congress must seize this opportunity to begin to fix a failed system before more workers—particularly women and workers of color and their families—are hit again with a potential double dip in the labor market. If these fixes are not implemented now, the race to the bottom that has already started will leave us with a wholly inadequate system for the next downturn. Now is the time to fix UI.

—Rebecca Dixon is executive director of the National Employment Law Project. William Spriggs is chief economist at the AFL-CIO and a professor in, and former chair of, the Department of Economics at Howard University.

Endnotes

1. According to a recent audit by the U.S. Department of Labor’s Office of the Inspector General, as of January 2, 2021, states had drawn down a total of $392 billion to pay UI benefits for the PUA, PEUC, and FPUC programs. See U.S. DOL-OIG 2021. According to researchers at the Economic Policy Institute, the UI expansions in the CARES Act and regular UI payments reduced the number of those in poverty by 7.2 million in June 2020. See Zipperer 2020.

2. Since the 2001 downturn, women’s labor force participation has been relatively flat at near 60% (not counting the sharp drop in this rate during the COVID-19 pandemic). It dropped during the Great Recession and never fully recovered for white women. See BLS CPS 2021.

References

Bureau of Labor Statistics, Current Employment Statistics (BLS-CES). 2021. “All Employees, Food Services and Drinking Places” [CES7072200001], retrieved from FRED, Federal Reserve Bank of St. Louis, https://fred.stlouisfed.org/series/CES7072200001, June 7, 2021; “All Employees, Manufacturing (MANEMP)” [CES3000000001], retrieved from FRED, Federal Reserve Bank of St. Louis, https://fred.stlouisfed.org/series/MANEMP, June 7, 2021; and “All Employees, Nondurable Goods (NDMANEMP)” [CES3200000001], retrieved from FRED, Federal Reserve Bank of St. Louis, https://fred.stlouisfed.org/series/NDMANEMP, June 7, 2021.

Bureau of Labor Statistics, Current Population Survey (BLS-CPS). 2021. “Labor Force Participation Rate – 20 Yrs. & Over, White Women” [LNS11300029], retrieved from FRED, Federal Reserve Bank of St. Louis; https://fred.stlouisfed.org/series/LNS11300029, June 7, 2021.

Freeman, Richard B. 2013. “Failing the Test? The Flexible U.S. Job Market in the Great Recession.” National Bureau of Economic Research Working Paper no. 19587, October 2013.

Gould, Elise. 2021. “The labor market is down 7.6 million jobs since February 2020, but the total jobs shortfall should take into account pre-pandemic labor market trends,” Twitter, @eliselgould, June 4, 2021, 9:02 a.m.

Nichols, Austin, and Margaret Simms. 2012. Racial and Ethnic Differences in Receipt of Unemployment Insurance Benefits During the Great Recession. Urban Institute, June 2012.

Stettner, Andrew. 2021. Fact Sheet: What’s at Stake as States Cancel Federal Unemployment Benefits. The Century Foundation, May 13, 2021.

U.S. Department of Labor, Office of Inspector General (DOL-OIG). 2021. COVID-19: States Struggled to Implement Cares Act Unemployment Insurance Programs, May 2021.

Zipperer, Ben. 2020. “Over 13 Million More People Would be in Poverty Without Unemployment Insurance and Stimulus Payments,” Working Economics Blog (Economic Policy Institute), September 17, 2020.

TABLE OF CONTENTS

Reforming Unemployment Insurance

• Foreword
• Executive summary
• Introduction
• Primer
• Statement of the problem
• Section 1. Universal standards
• Section 2. Financing
• Section 3. Eligibility
• Section 4. Benefit Duration
• Section 5. Benefit levels
• Appendix

Key proposals

• Minimum standards: Establish federal minimum standards for which workers are eligible for unemployment insurance benefits, the minimum length of time benefits can last during regular economic times and during downturns, the share of lost wages replenished by UI benefits, and a weekly wage floor that benefits amounts cannot drop below.
• Tax penalties: Encourage the use of statutory tax penalties by allowing the federal government to apply incremental increases to the federal tax rate on employers in states that fail to meet minimum universal standards and other key obligations of a more equitable UI system.
• Noncompliance: Provide workers with a pathway to report state noncompliance to the federal government.
• Processing delays: Ensure that workers with pending applications that are not being processed in a timely way receive at least the minimum weekly benefit.

Introduction

The problem

The unemployment insurance system is a cornerstone of our economic infrastructure. It exists to support working people who have lost their jobs through no fault of their own with cash benefits while steadying the economy during crises. But the UI system in many states now fundamentally fails in its core missions. Under the current system, the federal government sets certain basic ground rules, while states can fill in most of the details. State eligibility formulas exclude core groups of workers that include disproportionate shares of workers of color, such as low-wage workers, and women, such as part-time workers. Also unfairly left out are many gig workers who are misclassified as contractors and thus denied the benefits afforded only to employees in traditional employment relationships. It is up to states to determine whether workers and their employers have complied with state and federal rules, and many states lack effective tools for important tasks such as determining which employees are being misclassified as contractors.

States compete in a race to the bottom on UI taxes and, in times of economic distress, slash their way to replenishing depleted trust funds by shortening benefit duration and tightening eligibility. As a result, the once-standard, but by no means adequate, 26 weeks of benefits in normal economic times is eroding and some states offer just 12 weeks of benefits. Today, weekly benefits replace just about 40% of average wages, which is not nearly enough for workers, particular low-wage workers with families, to get by. These deteriorated standards are particularly harmful to Black workers. UI data show that states with high shares of Black workers tend to have lower UI recipiency rates—shares of unemployed workers who actually receive unemployment benefits—and lower average weekly UI benefits.

The solution

Require states to meet minimum federal standards for UI eligibility, duration of benefits, and benefit levels.

Implementation

To ensure that states meet minimum federal standards for UI eligibility, duration of benefits, and benefit levels, deploy carrots—assistance in the form of federal data collection enabling delivery of broader and better targeted benefits—but rely on sticks—incremental increases in tax rates for employers that fail to meet the minimum requirements, supported by mechanisms for workers to report state noncompliance.

Setting universal minimum guarantees across three big policy areas

Many of the problems we’ve diagnosed within the UI system are driven at least in part by large structural choices that the designers of the program made when it launched in the 1930s. In the long run, a sustainable UI program for the 21st century may require fundamental rethinking of those choices. Given the political reality that interest in UI reform peaks in the wake of an economic crises and ebbs thereafter, failure to pursue major revisions now—in the midst of the coronavirus pandemic—risks squandering a rare opportunity. Our effort here therefore aims at the most urgently needed repairs.

Policy proposal: Enact a federal law establishing universal minimum guarantees across three key big policy areas: eligibility for benefits, duration of benefits, and benefit levels

To roll back the states’ race to the bottom, and guard against further erosion, we propose that federal law establish universal minimum guarantees across three key big policy areas: eligibility for benefits, duration of benefits, and benefit levels. States that wanted to establish more robust or more expansive benefits than the minimum guarantee could still do so. Guaranteeing universal minimum standards nationwide will make the system easier to navigate, and make it easier for workers to understand their eligibility.

Ensuring that states meet the minimum requirements

To ensure that states can afford to deliver on these guarantees, we propose a set of financing reforms—fair, efficient taxes that strongly encourage employers to be good partners with workforce agencies in administering the system.

Foremost, we propose requirements with real consequences for falling short, not just incentives, because recent efforts to gently nudge states to modernize have not resulted in meaningful or lasting changes. For example, although federal changes in the wake of the Great Recession sought to encourage states to extend benefits to part-time workers, and states did make some technical changes on that front, “monetary eligibility” rules still effectively bar many part-timers from collecting any UI benefits. In the current political environment, in which some states are outright refusing to deliver even benefits fully paid for with federal dollars (Stettner 2021), we doubt that meaningful reforms can be achieved with the kinds of modest carrots and unthreatening sticks previously on offer (Galle 2019).

Instead, we would update and expand the list of obligations states must meet in order for their businesses to qualify for low federal unemployment taxes, and give the Department of Labor more flexibility to set federal tax rates.

Policy proposal: Allow the federal government to apply incremental increases to the federal tax rate on employers in states that fail to meet minimum universal standards and other key obligations

The federal tax rate, the Federal Unemployment Tax Act (FUTA) rate, is 0.6%, but this rate rises tenfold to 6.0% if a state fails to meet its statutory obligations. This penalty is so large, however, that the Department of Labor has never been willing to impose it. Decades ago, Congress wrestled with a similar problem with the charitable contribution deduction: Because revocation of a charity’s eligibility to receive deductible donations was effectively a “death sentence,” the IRS very rarely imposed it, even in situations where individuals were exploiting charitable resources for their own use (Manny 2007). In response, in 1998 Congress enacted a system of “intermediate sanctions,” under which the IRS is empowered to impose much smaller penalty taxes on organization managers. We would similarly recommend that Congress authorize the Department of Labor to increase the federal tax rate on employers in small increments within the range between 0.6% and 6.0%, depending on the degree of state noncompliance. Certain minimum penalties for clear breaches of federal guarantees, such as for potential benefit duration of less than the federal minimum, should be set by statute.

Providing workers with a voice and options when universal standards are not met

Intermediate sanctions might help ensure state consistency with universal guarantees in the long run, but additional measures are needed to provide workers with a voice in ensuring that universal guarantees are met and to hold them harmless when they are not. Even an expanded sanctions regime depends on Department of Labor willingness to use it. Under current administrative law doctrine, agencies can foil most enforcement regimes by refusing to initiate action (Barkow 2016). Private parties often have no avenue for judicial recourse to challenge agency inaction (Metzger 2015).

Policy proposal: Provide workers with a pathway to report state noncompliance to the federal government

Workers should have a pathway to report state noncompliance to the federal government. We therefore propose that the sanctions statute include opportunities for the public, including labor organizations and other community groups, to petition the Department of Labor to initiate a sanctions action, patterned on existing administrative law procedures that allow private parties to petition for rulemaking (CRS 2020). As in the petition for rulemaking process, DOL would be obligated to respond to nonfrivolous petitions, and if it chooses not to take action, it must explain why not. A decision not to impose sanctions would be reviewable in court as “final agency action.” Petitioners would be entitled to recover costs and attorneys’ fees for nonfrivolous petitions and court challenges.

Policy proposal: Ensure that workers with pending applications that are not being processed in a timely way receive at least the minimum weekly benefit

In states that fail to deliver guaranteed benefits in a timely way, workers should be entitled to at least the minimum weekly benefit while their application is pending. As explained in Section 5, the benefit levels section of the report, the minimum benefit in normal economic times is the greater of $250 or 30% of the state’s average weekly wage, and in downturns the greater of $300 per week or 40% of the state’s average weekly wage.

Benefit lapses leave mother scrambling to pay bills

Amy Cabrera-Cabello, Arizona

I am a single mother trying to put my son through college and then medical school. I was working at the largest provider of meeting and travel services in the world, until COVID. After taxes my state benefits are a little over $200 a week. I am not sure how that little bit of money is supposed to help the average worker.

Then [after receiving benefits for a while] things got worse. On December 26, 2020, I didn’t receive a check. I called and after waiting on hold for two hours, spoke to a woman who was convinced [the missed check] was due to Trump signing the relief bill late. I continued calling, often waiting on hold for hours. I was told to file an extension, but that the system was so backed up it could take 11 weeks.

After more than two months, I received some payments, but on March 20th they stopped again. I was told to file for another extension. I can barely pay my bills on the little benefits in Arizona. Not getting it at all, and having this happen frequently, is even more nerve-wracking.

Enacting universal guarantees as a path to bigger reforms

Not all the signatories think these steps go far enough to solve unemployment insurance’s deep-seated challenges. We all agree, though, that the universal guarantees we propose below are major improvements over the status quo, may help to set UI on a path to bigger reforms, and are well worth pursuing in their own right.

References

Barkow, Rachel. 2016. “Overseeing Agency Enforcement.” George Washington Law Review 84: 1129–1186.

Congressional Research Service (CRS). 2020. Petitions for Rulemaking: An Overview, CRS Report R46190, January 2020.

Galle, Brian. 2019. “How to Save Unemployment Insurance.” Arizona State Law Journal 50: 1009–1064.

Manny, Jill. 2007. “Nonprofit Payments to Insiders and Outsiders: Is the Sky the Limit?” Fordham Law Review 76: 735–763.

Metzger, Gillian. 2015. “The Constitutional Duty to Supervise.” Yale Law Journal 124: 1836–1933.

Stettner, Andrew. 2021. “Fact Sheet: What’s at Stake As States Cancel Federal Unemployment Benefits.” The Century Foundation, May 13, 2021.

TABLE OF CONTENTS

Reforming Unemployment Insurance

• Foreword
• Executive summary
• Introduction
• Primer
• Statement of the problem
• Section 1. Universal standards
• Section 2. Financing
• Section 3. Eligibility
• Section 4. Benefit Duration
• Section 5. Benefit levels
• Appendix

Unemployment insurance (UI) is a pillar of the social safety net. During normal economic times, UI helps families to protect against unexpected drops in income. In recessions, it takes on an added role as economic booster, helping to maintain spending against the downward spiral that is typical of most downturns. It was introduced as a national policy in the United States in the wake of the Great Depression, after several states had tried and largely failed to establish their own programs.

The core aspects of UI haven’t changed much since 1937. In essence, UI is a government program jointly funded by states and the federal government, and mostly administered by states. The federal government sets certain basic ground rules, while states can fill in most of the details. States also determine whether workers and their employers have complied with state and federal rules.

UI provides a partial replacement of wages for some recently unemployed workers. A worker who is fired, or is forced by certain compelling circumstances to leave, can claim benefits, while workers who quit voluntarily usually cannot. Eligible workers must submit a claim to a local unemployment office, and in most cases must show the office that they are available for and seeking a replacement job. Claimants also must meet “monetary eligibility” targets, meaning that they must show that they earned some minimum amount in the year or so before submitting their claim. This rule makes lower-earning part-time workers effectively ineligible for UI in many states. Self-employed individuals, such as those who work as contractors rather than employees, are also usually ineligible, although in 2020 a special program (Pandemic Unemployment Assistance, or “PUA”) enacted by Congress as part of the CARES Act in response to the COVID-19 pandemic temporarily granted benefits to these workers.

The fraction of separated workers (unemployed, for any reason) who receive benefits from the regular (i.e., nonemergency programs) is known as the “recipiency rate.” Because states vary so widely in their rules, so too do their recipiency rates. In 2019, although the national average was 28%, New Jersey topped the nation at 59%, while in many states fewer than one in six separated workers got any UI at all. Florida was lowest, at 11% (U.S. DOL-ETA 2021b). The recipiency rate in the United States is far below average for the rich countries of the Organisation for Economic Co-operation and Development (OECD 2018).

If a worker is found eligible, she is paid a fraction of her old wages, known as the “replacement rate,” up to a statutory dollar-value cap. Both this fraction and the cap are set by states. At the end of 2019, weekly benefits averaged $371 nationwide; before tax this represented a replacement of about 38% of the average wage (U.S. DOL-ETA 2021a). Again, the United States replaces a much smaller fraction of worker wages than many of its rich country peers in the OECD, and the U.S. replacement rate is below the OECD average (OECD 2021b).

Workers can only claim benefits for a limited period, 26 weeks in most states, although since 2010 at least 10 states have cut their benefit duration (some to as short as 12 weeks). Federal law requires states to also offer extended benefits (EB) for an additional period when certain adverse economic conditions are met. And Congress enacted additional, temporary, extensions in 2009 and 2020.

Both state and federal governments impose taxes to pay for the UI system. In both cases, UI taxes are nominally imposed on employers, but economists believe that in the long run employers pass some of the cost of these taxes on to workers in the form of lower salaries. The federal government collects state and federal taxes and holds each state’s proceeds in a trust fund account. Federal law encourages states to require that employers are “experience rated,” so that employers whose workers file more successful claims pay a higher rate of tax.1 Firms with very high experience-rated tax rates relative to their industry are less likely to be able to pass through these taxes to workers, which means that experience rating creates real incentives for firms to either reduce turnover or find ways to deny benefits to workers (see Section 2. Financing).

State revenues pay for benefits, while the federal revenues fund grants to cover most of the direct costs of UI administration. States also must pay 50% of the cost of extended benefits when those are active, but in recent recessions the federal government has covered 100% of those expenses, as well as the full cost of the emergency extensions.

Additional details on UI finances, eligibility, benefit levels, and benefit duration

We now offer additional detail for readers who may be interested in particulars, with the summary of federal rules drawn primarily from 26 U.S. Code Chapter 23—Federal Unemployment Tax Act (https://www.law.cornell.edu/uscode/text/26/subtitle-C/chapter-23) and state rules based largely on the U.S. Department of Labor’s “Significant Provisions of State Unemployment Insurance Laws” (U.S. DOL-ETA 2020b). In the current UI system, most rules are set individually by states. Although federal law imposes a handful of rules, the main influence of federal law on states is by way of the tax system and other financial incentives. We therefore begin with some added information about how UI is funded.

Finances

The taxable wage base. States and the federal government both collect unemployment taxes from employers on the “taxable wage base.” Each employer pays a tax on each dollar of wages paid to employees, up to a cap. The portion of wages under the cap is the wage base. For example, the federal tax is imposed only on the first $7,000 of wages paid to each employee. The wage base on which the state tax is imposed cannot be lower than the federal wage base but can be higher, and state wage bases today range from $7,000 to about $53,000, with the majority under $15,000.

The federal tax rate. Federal taxes are usually 0.6%, but only if a state is in compliance with federal requirements. If the U.S. Department of Labor were to find a state out of compliance, the tax rate would increase by a factor of 10, to 6%. To our knowledge no state has ever triggered this penalty; the list of requirements is short and DOL has little discretion to reject state rules. Thus, the current federal tax is $42 per worker annually: 0.6% times the wage base of $7,000.

The state tax rate. An employer’s total state tax payment per worker depends not only on the wage base but also on the applicable tax rate, which can vary from employer to employer. Employers with more worker turnover pay higher rates, a practice known as “experience rating.” The total payment per employee is just the product of the employer’s rate times the wage base. In Missouri, a fairly low-tax state, the average employer would pay to the state a 1% tax on the state wage base of $11,500, for a state tax total of $115 per worker. Nationwide in 2020, the total average state UI tax was about $277 per worker, a decrease from an average of $350 over the 2013 to 2018 period (U.S. DOL-OUI 2021; BLS 2020).

State trust funds. Because UI expenditures usually exceed revenues during recessions, states are required to maintain a “trust fund” account with the federal government. State tax revenues are deposited in the trust fund and then state benefits are paid out of the fund. States that run out of money can borrow from the federal government, although these debts must usually be repaid fairly quickly and at relatively high rates. States that maintain a high enough balance in their accounts can qualify for a very short period of interest-free borrowing.

Federal grants for administration and operations. Federal revenues are used mostly to pay for state administration of the UI program. The U.S. Department of Labor makes annual grants to states out of the federal unemployment insurance trust fund to cover personnel and other expenses, and it has some limited discretion to impose conditions on these grants. States also receive money under other federal grant programs for operations related to UI, such as funding for job centers that may serve to connect individuals to UI.

Eligibility for benefits

States impose both initial and continuing requirements for receiving UI benefits. Initial requirements are usually subdivided into two categories, “monetary” and “nonmonetary” requirements. Each week after receiving initial benefits, workers must typically also establish that they are “able and available” for work.

Monetary requirements for initial benefits. Monetary requirements are intended to establish that the individual is already attached to the workforce—that is, that they are experiencing a temporary interruption in earnings, not permanent unemployment. Typically, applicants must show that they have earned some minimum amount of wages in the year to year and half prior to their application. This testing period is known as the “base period,” and in most states is the first four of the five most recent completed quarters prior to application; some states are more flexible for some applicants (U.S. DOL-ETA 2020b).

Determining the minimum total wage that has to be earned during the base period is complex and varies widely from state to state. For instance, in Connecticut all that is needed is the greater of $600 or 40 times the worker’s weekly benefit amount over the base period. Next door in New York, a worker must have earned at least $2,400 in one of the base period quarters, and $3,600 overall. As a practical matter, some part-time workers are effectively disqualified from UI in states with higher minimums, such as Arizona, Kansas, Michigan, and New York.

Nonmonetary requirements for initial benefits. In addition to needing to meet these dollar thresholds, separated workers also must show that they left work for the right reasons, usually known as the “nonmonetary” requirement. Like other insurance, UI programs attempt to limit an insured’s ability to control when they receive payment, and this is the primary function of nonmonetary rules. While states vary considerably, for the most part workers cannot claim benefits if they left a job voluntarily, unless they can establish a compelling and relatively involuntary reason for their departure. Many workers lose benefits on these grounds, with nonmonetary denials averaging more than 1.2 million per year nationally (U.S. DOL-ETA 2021c).

Continuing requirements. Lastly, even after satisfying these initial hurdles workers must also establish, on a weekly or biweekly basis, that they are searching for work and available to accept a job. In most states an individual who is only able to accept part-time work is not considered “available for work” unless that worker qualified for UI benefits with part-time hours (U.S. DOL-ETA 2020a).

Together, these rules create several notable gaps in who is able to depend on UI. Self-employed individuals, including workers who in reality are employees but have been misclassified by their employer as contractors, cannot claim UI because they lack qualifying wages. Part-time workers, as we’ve noted, often fail monetary eligibility and may also be unable to establish that they are available for work. Most noncitizens without work authorization are denied benefits by federal law, although the exact boundaries of that prohibition are contested. Those who are just entering the workforce or rejoining after an extended interruption, such as mothers or recently incarcerated individuals, also cannot meet monetary eligibility standards (U.S. DOL-ETA 2020a).

Benefit levels

For the most part, states are free to decide what benefits to provide to eligible UI claimants. Key factors that determine a worker’s benefits are the share of pretax wages paid, also known as the “replacement rate,” as well as minimum and maximum weekly benefit amounts. Net benefit amounts are also affected by state and federal government policies on whether UI benefits are taxed.

State formulas for determining weekly benefits are complex and vary considerably from state to state. A plurality approach is to compute the worker’s highest quarterly wage during the base period (again, the base period is usually the four quarters preceding the most recent completed quarter). Among these states, many of them award workers one-half of their average weekly wage during that highest-wage quarter. Beneficiaries always get at least a minimum weekly benefit, with the median state offering around $50 (U.S. DOL-ETA 2020b).

Weekly benefits are capped, usually at some multiple of the average weekly wage for all workers in the state. In 2020, there were six states (Alabama, Arizona, Florida, Louisiana, Mississippi, and Tennessee) where the maximum weekly benefit was less than $300; in Mississippi it was just $235. A few states additionally grant a small allowance for each dependent, which also can increase the maximum benefit amount. For example, in 2020 the maximum benefit for any household, including the dependent allowance, was the $1,234 allotted in Massachusetts.

In general, a worker who is receiving benefits loses them when returning to work. To encourage part-time work among benefit recipients, most states allow a small amount of part-time earnings in addition to UI. Any earnings above this amount then reduce benefits dollar for dollar: effectively, a 100% marginal tax rate. A typical earnings disregard is 20% to 25% of the weekly benefit amount (U.S. DOL-ETA 2020b). So, for instance, with a disregard of 20%, a worker receiving $200 per week in benefits could earn up to $40 in added wages without reducing their benefits, but then a worker who earned $50 would get only $190 in benefits.

Unemployment insurance benefits are included in federal taxable income for income tax purposes, but are not subject to the Social Security payroll tax (Congress also exempted the first $10,200 of 2020 UI benefits from tax for lower-earning households). Most states follow federal law with respect to taxation of UI benefits, although California and Pennsylvania both expressly exempt UI benefits from state taxable income (Galle, Pancotti, and Stettner 2021; Pennsylvania Office of Unemployment Compensation 2021).

Taxation affects the net replacement rate. Suppose a worker is in the 10% federal tax bracket. If their pretax wages are $10,000, they take home $9,000. Suppose the state intends to provide workers with a net replacement rate of 50%. If UI benefits are not federally taxed, the state can offer a UI benefit of $4,500: that is the amount that leaves workers with 50% of their working (net) wages. In contrast, if UI benefits are subject to federal tax, the state must pay a benefit of $5,000, resulting in the same $4,500 in benefits received after tax. For any given replacement rate target, the state must pay more (and impose higher UI tax rates) when UI benefits are subject to federal tax.

Benefit duration

As with the weekly benefit amount, states have almost complete control over the duration of ordinary UI benefits. Traditionally, states offered a maximum of 26 weeks, a figure that is among the lowest in the developed world. Again the United States lags far behind its OECD peers, many of which offer up to two years or more (OECD 2021a). Further, since 2009, many states have reduced maximum benefit durations further, in a couple of cases to as few as 12 weeks of benefits.2 Massachusetts currently has the longest potential benefit duration, topping out at 30 weeks for some beneficiaries.

As described briefly above, the federal Extended Benefits program, or EB, mandates additional weeks of benefits, and these benefits are triggered automatically by economic conditions. The EB program provides two tiers of extended potential benefit durations: tier 1 adds 13 weeks and tier 2 adds an additional 7 weeks. The triggers for tier 1 are based on the level and/or the change in the insured unemployment rate (IUR)—a measure of how many in the labor force are currently receiving UI benefits.3 This measure can obviously vary by labor market conditions, but also can vary based on state-level eligibility criteria. That is, in states where it is hard to qualify for UI benefits, the state is less likely to hit the IUR trigger. States can opt to use another trigger that is a much more focused measure of labor market distress: the total unemployment rate (TUR), which is simply the official unemployment rate, a measure that includes workers who are unemployed but not receiving UI benefits.4

The EB program is also designed to automatically trigger off benefits when economic conditions improve or do not worsen. Once the trigger on criteria are no longer met, EB is turned off. In addition, an EB “look-back” provision triggers benefits off whenever there has been no significant increase in unemployment over the past two years. Persistently high unemployment does not prevent the EB benefits from triggering off.

Endnotes

1. 26 U.S. Code § 3303

2. In some cases states do not directly determine how long benefits will last, but instead first calculate a total benefit amount and weekly benefit amount, both based on wages during the base period. For example, if the total benefit were $10,000 and the weekly amount were $500, the household would have 20 weeks of benefits.

3. The IUR trigger is 5% plus 20% above the average level in the same quarters of the two previous years, or optionally a flat 6% IUR. Most states and the District of Columbia using the IUR trigger use the optional 6% trigger.

4. At press, 19 states used the TUR trigger, which is 6.5% and a 10% increase over the same quarter of either of the previous two years.

References

Bureau of Labor Statistics (BLS). 2020. “The Cost of Layoffs in Unemployment Insurance Taxes.” Monthly Labor Review. April 2020.

Galle, Brian, Elizabeth Pancotti, and Andrew Stettner. 2021. “Expert Q&A About the Unemployment Provisions of the American Rescue Plan.” The Century Foundation, March 10, 2021.

Organisation for Economic Co-Operation and Development (OECD). 2018. “Pseudo-coverage Rates of Unemployment Benefits.” Social Benefit Recipients (SOCR) Annual Data by Country. Pooled 2007–2018 data, accessed June 8, 2021.

Organisation for Economic Co-operation and Development (OECD). 2021a. “Comparative Policy Tables 2020” (online database), OECD Tax-Benefit Data Portal, accessed 2021.

Organisation for Economic Co-Operation and Development (OECD). 2021b. “Net Replacement Rate in Unemployment” (online table). In OECD.Stat (database), accessed 2021.

Pennsylvania Office of Unemployment Compensation. 2021. “Taxes on Benefits.” Accessed May 2021.

U.S. Department of Labor Employment and Training Administration (U.S. DOL-ETA). 2020a. “Monetary Entitlement” and “Nonmonetary Eligibility.” In Comparison of State UI Laws. Accessed June 2021.

U.S. Department of Labor Employment and Training Administration (U.S. DOL-ETA). 2020b. “Significant Provisions of State Unemployment Insurance Laws.” July 2020.

U.S. Department of Labor Employment and Training Administration (U.S. DOL-ETA). 2021a. UI Replacement Rates Report (online database), fourth quarter 2019 data accessed June 8, 2021.

U.S. Department of Labor Employment and Training Administration (U.S. DOL-ETA). 2021b. Unemployment Insurance Chartbook (online database), “Category 13: Recipiency Rates by State.” Accessed June 8, 2021.

U.S. Department of Labor Employment and Training Administration (U.S. DOL-ETA). 2021c. Unemployment Insurance (UI) Nonmonetary Determinations [downloadable data sets]; summing 2019 quarterly figures from the “UI Nonmonetary Determinations” and column in spreadsheet for separation denials. Updated April 1, 2021.

U.S. Department of Labor Office of Unemployment Insurance (U.S. DOL-OUI). 2021. State Unemployment Insurance: Tax Measures Report 2020. April 2021.