Productivity growth has firmed up slightly in recent years, but employers still aren’t acting like labor costs are something they’re particularly worried about containing through investments in capital equipment or better processes.

Figure A

Nonresponse to the survey used to calculate unemployment is growing and not random

The monthly unemployment rate is calculated from the Current Population Survey (CPS), a survey of roughly 50,000 households. Household nonresponse to the CPS—people refusing to fill out the survey—has grown in recent decades. In their 2006 Center for Economic and Policy Research report, John Schmitt and Dean Baker compared the employment rate from the decennial census (which has near-universal response) and the monthly CPS for 2000, drawing on information from “a unique dataset that matched respondents to the February, March, April, and May CPS for 2000 with their Decennial Census forms” to correct for incomparability in reported employment rates between the CPS and the census. In 2000, household nonresponse to the CPS was roughly 10%, and Schmitt and Baker estimated that bias in nonresponse likely understated employment rates by roughly 1.4 percentage points compared with the near-universal census. Since 2000, nonresponse has risen by another 7–8 percentage points. This suggests significant scope for rising nonresponse since 2000 to pull down measured rates of unemployment.

More recently, a 2019 paper by Hie Joo Ahn of the Federal Reserve Board and James D. Hamilton of the University of California, San Diego, finds that a number of “inconsistencies” in CPS tabulations of unemployment—many likely driven by nonresponse—lead to understatements in the unemployment rate (and in labor force participation). Figure B shows the extent of understatements in the unemployment rate in the early 2000s relative to today. According to Ahn and Hamilton, these understatements have increased between 2001 and 2018 by roughly 0.3 percentage points (moving from 1.6 to 1.9 percentage points). Again, all else equal, the unemployment rate today likely needs to be adjusted 0.3 percentage points upward to be more usefully compared with measured unemployment rates in the late 1990s.

Figure B

Figure C

Further, in their 2014 working paper, David Blanchflower and Adam Posen of the Peterson Institute for International Economics have found that the downward pressure on wages stemming from adults not in the labor force (but in effect part of the pool of workers competing for jobs) has increased significantly after 2002 relative to the 10 years before.

Both of these measures—the quantity-side measure of how many new jobs are filled by those from out of the labor force and assessments of how much downward wage pressure is exerted by those not in the labor force—indicate that workers out of the labor force are more substitutable for the unemployed than they were in the past. In turn, this means that compared with previous historical periods, any given unemployment rate today obscures the slack embodied in adults not in the labor force who are nevertheless quite likely to be tomorrow’s potential workers.

Be careful when comparing unemployment rates over time

It is by now well recognized that the relationship between the unemployment rate and wage growth is different in recent years than it has been in the past. Part of this divergence is likely due to genuine structural changes in the economy, particularly the degraded bargaining power and leverage of typical workers due to policy choices. But some of the divergence is likely simply because a given unemployment rate is not providing the same signal about labor market tightness over time due to compositional changes in the workforce and difficulties in deriving an aggregate unemployment rate from a household survey. It is worth noting that many of these issues that blur signals from the unemployment rate will apply to other quantity-side measures of labor market slack, like the prime-age employment rate. Finally, the line between the officially unemployed versus potential workers currently classified as not in the labor force has become ever muddier over time.

All of this implies that policymakers—particularly those at the Fed—must continue to be extremely modest in how well they think they can forecast coming wage and price pressure stemming from unemployment. Instead, they should let the data speak and worry about wage and price pressure when it comes—and not rely on historical relationships between unemployment, wages, and prices to make policy going forward.

Specifically, I address the widespread speculation about the possibility of “wage-led productivity growth”—the hypothesis that tight labor markets that push up labor costs lead firms to invest more in labor-saving equipment and practices. Some suggestive evidence of this wage-led productivity growth has been shown in patterns of macroeconomic data. This newsletter provides some more suggestive evidence from patterns of productivity growth broadly but also across a set of very detailed industries when the labor share of industry income rises and falls. Here are the key findings:

• At the aggregate level, a rise in the labor share of corporate-sector income is associated with a small but significant rise in the pace of average productivity growth over the subsequent two years.
• At detailed industrial levels (looking at 124 industries mostly in manufacturing), this relationship is even stronger: a 1 percentage-point rise in the labor share of industry income is associated with a 0.1 percentage-point increase in the average pace of productivity growth over the subsequent two years.
• These relationships between labor share of income and productivity growth are consistent with a scenario in which firms try harder to make labor-saving investments and organizational changes when labor becomes scarcer and the need to pay higher wages threatens to pinch profits. If the labor share of corporate-sector income rose from today’s 78% to the 82% that characterized the tight labor markets of 2000, this would imply a boost to productivity of roughly 0.4 percentage points—an amount that would cut almost in half the gap between the productivity growth in recent years and the productivity growth of the late 1990s.

These results are obviously suggestive, not dispositive. The key challenge to testing the causal link that runs from higher pressure labor markets to increased effort by firms to find and adopt labor-saving practices and investments is finding truly exogenous changes in labor market tightness. This search for exogenous changes in the cost pressures firms face should be a prime preoccupation for macroeconomic policymakers in the near future. In the rest of this newsletter, we’ll describe our findings in a bit more detail and discuss their potential implications.

The two-way causal links between labor costs and productivity
It is well-known that in competitive markets productivity growth and wage growth are supposed to rise in tandem across the entire labor market, with the source of causality running from higher productivity to higher wages or labor costs. It is equally well-known that this has not happened in the U.S. economy in recent decades, strongly suggesting that the aggregate U.S. labor market does not fit textbook descriptions of competitive.

Despite the pronounced failure of U.S. wage growth and productivity growth to track each other in recent decades, it remains true that the “all else equal” effect of productivity growth on wage growth over reasonable periods of time is likely positive: faster productivity growth may not translate one-for-one (or even close to one-for-one) into wage growth, but some of the productivity growth does serve to make wages grow faster.

A much-less-discussed strand of economic thought highlights a potential reverse causality whereby rapid growth in labor costs spurs productivity growth as firms become more aggressive in finding labor-saving investments and practices. While a number of research papers have highlighted this reverse causality as a theoretical possibility, there has been less headway in assessing its real-world empirical relevance (see this Twitter thread for an excellent list of papers on this topic).

Part of the reason for the lack of empirical findings on this issue is that it’s hard to isolate the causal effect of faster-growing labor costs on productivity growth. As noted before, some correlation between labor costs and productivity is to be expected, so identifying the one-way causal effect is quite hard. The obvious first-pass solution to this problem of causality is to use leads and lags, and this is the strategy we adopt in this newsletter.

In Figure A, we show the effect of a one-year change in the labor share of corporate-sector income on average productivity growth over the subsequent two years (with the change represented by a fixed dot and the confidence interval represented by the dotted line). At first blush, this regression setup could do a decent job of isolating the causal effect of labor cost pressure on productivity growth as it seems unlikely that average productivity growth between 1997 and 1999 could be the causal force behind a change in the corporate labor share of income between 1996 and 1997. We use two-year changes in productivity growth because this data series is so volatile, with any timespan significantly shorter than two years displaying lots of statistical noise.

Figure A

Using data from 1949 to 2019, the association between labor share changes and subsequent productivity growth is significant, as shown in the first data point reading from the left in Figure A. With no other controls, a 1 percentage-point increase in the labor share of income leads to a 0.125 percentage-point increase in average productivity growth over the next two years. We also run the same regression using controls for a trend, recessions, and interactions between recessionary quarters and the change in the labor share of income. These controls reduce the coefficient on the labor share in half, though it remains statistically significant.

The figure also shows coefficient estimates from the same regression using data from a panel of 124 industries (mostly manufacturing) between 1987 and 2016. Again, we regress the subsequent change in the two-year average of productivity growth against a change in the labor share of industry income. This bivariate relationship is shown in the figure. It indicates that a 1 percentage-point increase in the labor share is associated with a 0.02 percent increase in average productivity growth over the subsequent two years. One important thing to note about this regression is that the conventional view that causality runs from productivity growth to changes in labor costs does not hold at the industry level. Across industries, the predicted relationship between productivity and labor cost growth in conventional presentations of competitive labor markets is zero.

In this richer industry data, we can also include controls for both year and industry fixed effects, as well as industry-specific trends. In this case, the relationship between the change in labor’s share and the subsequent growth in productivity is actually strengthened by the inclusion of more controls. In the regression with a full set of controls, a 1 percentage-point increase in the labor share is associated with a 0.09 percentage-point change in average productivity growth in the subsequent two years.

While previous EPI research has focused on the role of labor cost pressure in boosting capital investments, we also tested whether a rising labor share is associated with faster multifactor productivity (MFP) growth at the industry level. MFP measures output produced with a fixed bundle of labor and capital. It is often thought to serve as a proxy for general technological or organization advances. But it has also been shown to capture the effects of economies of scale or just more-intensive utilization of factors of production.

A far-smaller number of these disaggregated industries reported MFP growth—just 11. Yet even with that much-smaller data set, an economically and statistically significant relationship between a rising labor share of industry income and faster subsequent MFP growth is found. This could result from firms feeling pressure to adopt more-efficient managerial practices and production techniques in response to labor cost pressure—or it could simply be the result of economies of scale that are detectable as output growth rises and pushes up wage growth.

Implications
An extensive, if not well known, strand of macroeconomic theory predicts that tight labor markets that begin to push labor cost growth higher might spur a countervailing reaction from firms looking to protect profit margins. One strategy these firms can take is to search more aggressively for labor-saving investments and practices. The incentive to aggressively search for such productivity improvements might be spurred as the labor share of income begins to rise (a development that means by definition that labor costs rise a bit faster in inflation-adjusted terms than productivity). If this theory is true, then it suggests that the weak productivity growth of the past decade should not be treated as a given that can’t be changed. Instead weak productivity growth should be seen as damage done—by the Great Recession and subsequent too-slow recovery—that has yet to be addressed.

Previously we have shown evidence that the labor share of income is indeed nudged up in late recoveries by tighter labor markets. This newsletter provides some suggestive evidence that as the labor share is nudged up, one margin of reaction from firms is searching for faster productivity improvements. The potential magnitude of this effect is important. If we take the coefficient on industry productivity with the full set of controls shown in Figure A, this implies that each 1 percentage-point change in the labor share of income boosts average productivity growth by roughly 0.09 percentage points over the subsequent two years. Currently the labor share of income in the corporate sector is 78%. In 2000, the labor share peaked at roughly 82%. If the U.S. economy saw the corporate labor share rise to 82% again in coming years, this would imply a boost to productivity growth of roughly 0.4%. This would move today’s pace of productivity growth almost halfway back to the level it averaged in the late 1990s, and would provide substantially more room for wages to grow.

Of course, these findings should be explored further, and particular attention should be paid to finding unambiguously exogenous sources of labor cost growth that are clearly not caused by contemporaneous productivity growth. Such exogenous changes are hard to find—minimum wage changes are the most obvious ones. But the minimum wage changes seen in the U.S. economy in recent decades have generally not been large enough to lead to labor cost increases large enough to spur a reaction from firms looking for productivity improvements.

Josh Bivens, director of research at EPI

Earlier this month, Olivier Blanchard—the former chief economist at the International Monetary Fund and an influential figure in macroeconomics—suggested that the Federal Reserve should consider targeting inflation in nominal wages rather than (or in addition to) inflation in prices. I was predictably intrigued by this: I proposed a nominal wage inflation target for the Fed a few years back.

This edition of the newsletter quickly sketches out the logic of a nominal wage inflation target to guide Fed decision-making on interest rates, and it highlights one particularly important detail: the assumed rate of productivity growth used to specify the target. If the rate of productivity growth is endogenous to the degree of labor market slack (as some evidence indicates), then using real-time estimates of productivity growth as an input into the wage target could threaten to lock in the damage to wage growth done by too-slack labor markets.

The key takeaways are:

• A nominal wage growth target has many virtues over the Fed’s current approach of targeting price inflation.
  • Price inflation is affected by many factors besides the pace of wage growth, yet only wage growth is the proper target for the Fed’s aggregate demand management. A wage growth target hence carries more signal and less noise than the price inflation target.
  • One key determinant of price inflation—the size of markups of firms’ prices over labor costs—actually varies countercyclically over wide phases of the business cycle. This introduces unuseful noise into the target the Fed is aiming for.
• A relatively long-lagged measure of trend productivity growth is the best input to use for specifying a nominal wage target.
  • High-frequency (real-time) measures of productivity growth are notoriously volatile, making them a bad input for any policy target.
  • As the economy exits a steep recession, real-time measures of productivity growth can be dragged down by the wage-suppressing effects of lingering labor market slack. Using these real-time measures in the construction of a nominal wage target can lead to a too-conservative target and prompt the Fed to raise interest rates prematurely.
  • There is some preliminary evidence that the steady tightening of U.S. labor markets over the past decade seems to finally be pushing up productivity growth rates. This provides more evidence that accepting low productivity growth rates posted when there was substantial labor market slack as unchangeable could be a large policy error.

What is a nominal wage target and why does it makes sense?
It is relatively well known that the Fed has a target for price inflation—2%—and that this target guides its decisions on setting interest rates. Given this price inflation target, as long as one is willing to make an assumption about the underlying trend in productivity growth, it is possible to specify a nominal wage inflation target that is consistent with an overall price inflation target.

The short version of this nominal wage target is that it should equal the sum of the Fed’s price inflation target plus the estimated trend rate of productivity growth. Productivity is simply the amount of output (i.e., income) produced in an average hour of work in the economy. As long as nominal wages are growing at or beneath the rate of productivity growth, then labor costs are putting no upward pressure on prices. Say that both nominal wages and productivity rose 2% in a year. Hourly wages climb 2%, but the amount produced in each hour of work—the definition of productivity—also rises by 2%, so costs per unit of output (i.e., prices) would not rise.

Of course, the Fed isn’t committed to zero upward pressure on prices. Fed officials say they’re comfortable with 2% price inflation. (I’d argue that they should be comfortable with inflation well above that, but we’ll take their target for now.) This price inflation target means that nominal wage growth can be 2% higher than trend productivity growth before wages threaten to push price inflation over the Fed’s target. If trend productivity growth is estimated to be 1.5%, this would mean that nominal wage growth of 3.5% would be needed before labor costs stopped dragging on the Fed’s ability to push price inflation up to its 2% target.

However, while it’s possible to specify a nominal wage growth target that is consistent with the Fed’s 2% price inflation target, is it useful to do this? It is. Put simply, the only determinant of price inflation that the Fed should reliably try to control is wage growth. Wages (or labor costs generally) are by far the largest single component of costs, so the Fed having influence only over wages with its interest-rate setting is not a problem per se. But nonwage determinants of price inflation either cannot or should not be targeted by the Fed as it raises rates as the economy heats up or lowers rates as the economy cools down.

Take the most obvious case where nonwage determinants should not factor into the Fed’s decision-making on inflation: the distinction between core and noncore prices. The Fed tends to ignore price inflation changes stemming from the volatile prices of food and energy. It is right to do this: Food or energy prices can spike upward even in a depressed economy simply because the weather changes. It would be a disaster if the Fed responded to food and energy price spikes when the economy is depressed by raising rates. This is not an academic concern: In 2011 the European Central Bank (ECB) raised rates twice, hard on the heels of the Great Recession, because of a small transitory increase in inflation. This was a grievous policy mistake.

Take a less obvious case of when nonwage determinants of prices can send the wrong signal to the Fed: the rise and fall of profit margins. Given the large and prolonged rise in unemployment stemming from the Great Recession, one might have expected a collapse in price inflation, or maybe even deflation, in the years after 2009. Yet price inflation (after stripping out food and energy) held up reasonably well. In fact, looking only at the price data might have convinced some that aggregate demand (spending by households, businesses, and governments) was not really all that depressed relative to the economy’s potential. But even as price inflation held up reasonably well, unit labor costs saw zero growth for a number of years. The only thing that kept price inflation from stagnating was an enormous increase in profit margins. In this case, trends in the labor market (flat unit labor costs stemming from anemic wage growth) were providing a much better signal of continued slack in aggregate demand than trends in price inflation (which was subdued, but not terribly so). In the years following the Great Recession, the Fed did the right thing in putting much more weight on wage growth than price inflation and kept monetary policy strongly expansionary. It was right to do so. Why not just operationalize this more formally by specifying a wage target?

Why the right productivity assumption is crucial for specifying the right wage growth target
In his remarks calling for a wage growth target, Blanchard argued that the target had essentially already been hit and that “we are more or less at full employment.” More specifically, he indicated that 1% should be taken as the proper assumption regarding productivity growth: “So, when you see wage inflation at 3% and you see productivity [growth] of 1% then you’re home.”

But a 1% productivity growth assumption is too pessimistic. It’s true that a three-year rolling average of productivity growth since 2008 saw it sit below 1% between 2011 and 2018. However, it has been shown decisively that when forecasting future productivity growth, one should use a long lag of past trends—even longer than seven years. In this longer run, 1.5% is a conservative estimate of trend productivity growth.

The need to examine long-run trends when making productivity forecasts is especially true given that the weak productivity growth of the current decade is almost certainly a casualty of the slack demand caused by the Great Recession. Surely it’s not a coincidence that productivity growth collapsed right as the economy entered the worst recession in many generations. Further, there is compelling evidence that a tightening labor market can spur labor-saving investments and organizational change. The logic is simple: When workers are plentiful and cheap because the economy is depressed, containing future labor costs is not a huge priority for businesses—and they certainly may be loath to borrow or take on other risk to make uncertain investments to keep labor costs in check. But when labor markets tighten and wage growth solidifies, investments to contain future labor costs make a lot more sense.

I noted nearly two years ago that tightening labor markets appeared to be nudging up productivity growth. This nudge seems to have become a shove in recent quarters. Figure A shows the three-year average of productivity growth (measured as total economy productivity—the broadest measure). Given how volatile short-run measures of productivity are, three years is essentially the shortest timespan over which one can look at these trends to get any signal at all through the noise. As we noted earlier, for forecasting future trends, one would want to look over a much longer past period. The spike in productivity shown on the graph during the worst of the Great Recession is a statistical fluke indicating only that employment actually fell faster than gross domestic product (GDP) in these years. The key thing to focus on is the sharp recent upturn within the last two years: Productivity growth is indeed closing in rapidly on the longer-run trend of 1.5%.

Figure A

Given that the Fed should be looking to overshoot any long-run wage target for a long spell to undo damage done by years of undershooting, the choice of productivity assumption matters a lot here. Figure B compares cumulative growth in (real) wages and productivity with productivity at a 1.5% growth trend. If wage levels just need to regain the level implied by growth in actual productivity in recent years (i.e., rise fast enough to close the gap between the bottom two lines), then this should not take too much more time. If, however, we want to target a clawback of productivity lost due to demand weakness in the post–Great Recession period (i.e., close the gap between the top and bottom lines by accelerating actual productivity growth), then labor markets will need to be hot for quite a long time.

Figure B

Year-over-year nominal wage growth has flattened out well below 3.5% in 2019. Adopting a 1.5% productivity growth assumption for nominal wage targeting implies that this is below full employment levels—and that a long period of being above full employment is still needed to undo the wage-flattening effects of the many years of below-target growth. Olivier Blanchard’s reintroduction of nominal wage inflation as a target of monetary policy is most welcome, but we should be careful about what assumptions are being embedded in this target when we pin a number to it.

Josh Bivens, director of research at EPI

The Federal Reserve meets this week against a backdrop of mounting evidence of a slowing economy. Since the last Federal Open Market Committee (FOMC) meeting, revised data on gross domestic product (the widest measure of the nation’s economic activity) and job growth have shown that 2018 saw much slower growth than previously reported.

Between April 2018 and March 2019, for example, the economy created 500,000 fewer jobs than had originally been reported. Only 105,000 jobs were created in August if temporary Census positions are excluded: this is roughly half the pace of growth that characterized pre-revision estimates of average job growth in 2018.

These clear signs of an economic slowdown raise the obvious question, “Why has growth faltered?”

While many pundits and economists have blamed the escalating trade conflict between the Trump administration and China, there are much more obvious sources of this slowdown: the Fed’s own premature interest rate increases between December 2015 and 2018 and the utter waste of fiscal resources that was the Tax Cuts and Jobs Act (TCJA) passed at the end of 2017.

To be clear, the Trump administration’s trade conflict is stupid and destructive, and its attempt to pin the blame for the slowdown on the Fed is self-serving. And the Trump administration’s scapegoating others for the weak economy takes real hubris given that its signature economic policy initiative—the TCJA—has been such an obvious failure in terms of spurring growth.

But the evidence is growing that the Fed did indeed raise interest rates too soon in the recovery and that this premature liftoff has begun dragging on growth. Worse, because raising rates slows growth more powerfully than lowering rates spurs growth, the Fed likely lacks the ability to offset this earlier mistake by pulling down rates going forward. The FOMC should certainly reduce rates at this week’s meeting, but it will need help from other policy levers—particularly effective fiscal stimulus—in the coming year.

Evidence of premature interest rate hikes

As Figure A below shows, after seven years of holding the effective federal funds rate at essentially zero, in December 2015 the Federal Reserve raised this rate by a quarter point. While many argued that a quarter-point increase in interest rates would not snuff out the ongoing recovery, I noted that raising rates while unemployment remained elevated and there was no sign of inflation made little economic sense. Worse, by increasing interest rates before any sign of inflation appeared in the data, the Fed clearly signaled that it was not eager to aggressively plumb just how low unemployment could be allowed to fall. This was a troubling signal: the Fed’s failure to aggressively target as low an unemployment rate as possible in recent decades has been a major reason why wage growth over this time has been so anemic. It took a year before the Fed followed up the December 2015 rate increase with another in December 2016, but in 2017 and 2018, it undertook seven quarter-point increases in the federal funds rate.

Figure A

Given this short history of interest rates, it makes sense to look at the evidence on the effect of these rate hikes on growth. The most obvious place to look is at the performance of “interest-sensitive” components of gross domestic product since the rate hikes began in 2015. Generally, residential investment, business fixed investment, durable goods purchases, and net exports are thought to be the components of GDP that will be slowed by interest rate hikes.

Figure B charts how the average contribution of various components of GDP made to overall GDP growth changed between two time periods: the era of zero federal funds rates (from the second quarter of 2009 to the end of 2015) and the era of rising federal funds rates (from the first quarter of 2016 to the most recent quarter available, the second quarter of 2019). For three of these components (residential investment, business fixed investment, and net exports) their contributions to growth slowed notably as interest rates rose. For durable goods, the contribution to growth is roughly the same in both periods, but this is striking given that personal consumption expenditures besides durable goods saw a sharp upswing in the latter period. In short, there is ample evidence that rising interest rates have worked as expected in slowing interest-sensitive components of GDP growth.

Figure B

To its credit, the Fed seems to have recognized that past rate hikes have dragged too much on growth and reduced rates at the last FOMC meeting in July. But research has shown that rate cuts spur growth less powerfully than equivalent rate increases restrain growth. This problem of “pushing on a string” was a prime argument made by those arguing that the Fed should err on the side of letting growth continue and letting unemployment continue to fall. If the economy continues to slow, the Fed will need help from fiscal policymakers to avert a recession; rate cuts by themselves are unlikely to do that job.

Very little sign of ‘trade war’ fingerprints on growth slowdown

The slight deceleration of net exports’ contribution to growth shown in Figure B may make some think there is a trade war–based explanation. There may be some influence of trade conflict on slowing growth, but this influence is likely pretty weak. For one, tariffs do not reliably reduce net exports—they instead reduce both exports and imports, with their effect on the trade balance (which is what matters for short-run growth) largely ambiguous. What is not ambiguous is the effect of a strengthening dollar on net exports—it reliably slows them. And since 2014, the dollar has risen sharply, driven strongly by developments in monetary policy in both the United States and its trading partners. It is important to realize that interest rate cuts made by the European Central Bank (ECB) late last week will put further upward pressure on the dollar going forward.

Some have noted that aside from the direct effect of tariffs, policy-induced uncertainty stemming from the Trump administration’s trade conflict might be holding back other components of growth, such as business fixed investment. Perhaps. But “uncertainty” is an awfully hard influence to define. And some of the only attempts to empirically measure this uncertainty actually show it is lower today than at many points in the last decade or more. Memories are short, but as recently as 2011 a Republican-led Congress seriously threatened to drive the federal government into totally unnecessary default on its debt if the Republican Party’s policy preferences were not signed into law by the Obama administration. This episode, it hardly needs to be said, created plenty of policy uncertainty.

The squandered opportunity of 2018’s TCJA

So if the recent economic slowdown is mostly not the fault of the Trump administration’s trade conflict, and it is mostly the fault of a too-hawkish Federal Reserve, does the president have a leg to stand on in scapegoating of Fed chair Jerome Powell? Not really. The reason why is simple: if President Trump had wanted faster growth, he should not have championed the waste of fiscal resources that was the TCJA, and instead should have used those resources to do things that actually would have created jobs and growth.

The TCJA is a debt-financed tax cut that will cost $150 billion annually over the next 10 years (before interest costs are added). Because the lion’s share of these tax cuts are accruing to rich households (mostly because the TCJA is primarily a corporate tax cut and owners of corporations are rich households), they have done very little to spur growth in aggregate demand (spending by households, businesses, and governments) in the short run. Rich households’ spending is not constrained by too-low disposable income, so boosting this disposable income largely does not lead to more spending. Poorer and moderate-income households, on the other hand, are indeed income-constrained in their current spending, so tax cuts or direct transfers to them would have boosted demand growth significantly. Direct spending—say on infrastructure or providing needed public investments like high-quality early child care—would have stimulated demand even more.

For a time, many credited the slight acceleration in growth in 2018 to fiscal stimulus generally. But revisions to 2018 data show this acceleration was even more subdued than previously thought. Further, the growth in government spending brought forth by a budget deal in 2018 actually provided much more stimulus than the more-expensive TCJA (see Figure B for this increase in government spending’s contribution to growth). The major component of GDP that has accelerated in recent years is consumption spending. Even if the entirety of the pickup in consumer spending shown in Figure B was attributed to the TCJA (and much of this pickup was surely driven by other factors, like tightening labor markets finally pushing up wage growth modestly), it would indicate that the TCJA was deeply inefficient as stimulus. TCJA proponents long argued that the main benefit stemming from it would not be short-run stimulus but a long-run increase in investment. This is awfully hard to see in the data—and as shown in Figure B investment (both business and residential) has been a prime source of weakness, not strength in recent years.

Essentially, President Trump and the Republican-led Congress largely squandered $150 billion in potential fiscal stimulus by prioritizing tax cuts for the rich over anything that would have plausibly created faster growth and jobs (see Table 1 in this report for a list of more and less effective fiscal policies to spur near-term growth). They are hence really the only economic observers in the world who have no standing to criticize the Fed for its clearly too-aggressive path of interest rate increases in recent years.

Despite bad-faith jawboning from President Trump, the Fed should cut rates

But just because President Trump calls for something—an interest rate cut in this case—doesn’t always mean it’s the wrong thing to do. The economy really is weakening, and this weakness has been led by sectors adversely affected by the Fed’s interest rate increases in recent years. Unfortunately, we could well find, a year from now, that rate cuts alone were not sufficient to avoid a recession—or even just a prolonged slowdown that pushes up unemployment. In that case, the Fed will need help from fiscal policymakers. But in the meantime, the Fed should take its role as the early-warning system on economic slowdowns seriously by cutting rates this week. This rate cut would provide a clear alert to other policymakers.

Josh Bivens, director of research at EPI

Federal Reserve Chair Jerome Powell’s July 10 testimony before the House Financial Services Committee was unlike any hearing featuring his predecessors.

Despite the vital importance of Fed decisions for the day-to-day lives of working families, congressional hearings featuring the Fed chair speaking about the state of the economy historically have disappointed. Disinterested and poorly informed questions posed by members of Congress have elicited opaque answers from Fed chairs.

This hearing was different. The questions were probing and informed, and Powell answered them with clarity.

Perhaps the most illuminating exchange occurred when Representative Steve Stivers (R-Ohio) asked Powell if the Fed was worried that low interest rates would cause the job market to run “hot.”

Some quick background throws Powell’s remarkable answer into sharp relief. One of the Fed’s two mandates—and the mandate that the Fed has unfortunately prioritized in recent decades—is to keep inflation under control. Traditionally, the perceived threat to controlled inflation in an expanding economy has been thought to come from the improving labor market. As unemployment falls and workers feel more confident, they can demand faster wage growth from employers. If wage growth (adjusted for inflation) exceeds economywide productivity growth, it puts upward pressure on prices (since labor costs are by far the largest component of prices).

Often in the Fed’s history, the response to Stivers’s question would have been, “Yes, unemployment this low definitely has us worried about overheating leading to inflation.” But too often this answer would have had very little empirical backing. For example, for a short spell in 2011 the estimated “natural rate” of unemployment below which inflation was forecast to begin accelerating was 2 full percentage points above today’s 3.7% rate. Yet no acceleration of inflation happened between 2011 and today.

‘To call something hot, you need to see some heat’

Powell’s answer to Stivers’s loaded question admirably reflected the facts.
We don’t have any basis or any evidence for calling this a hot labor market. We have wages moving up at 3%, which is good because it was 2% a year ago, but 3% barely covers productivity increases and inflation. And it certainly isn’t fast enough to put upward pressure on inflation. [However], we haven’t seen wages moving up as sharply as they have in the past. … 3.7% is a low unemployment rate, but to call something hot, you need to see some heat. While we hear reports of companies finding it hard to find qualified labor, we don’t see wages responding.

In this newsletter, I look for some “heat” in widely watched economic variables. But instead of seeing heat in these variables, I find pretty consistent cooling.

Wage growth has actually flattened recently

The most important indicator of a fast-warming job market is wage growth. Powell’s reference point for wage growth—the sum of productivity growth and inflation—is exactly the one EPI has used for years. Specifically, according to our Nominal Wage Tracker, as long as nominal wage growth is less than or equal to the Federal Reserve’s 2% overall price inflation target plus the growth rate of potential productivity, the labor market is not getting too “hot.” Our tracker usually looks at year-over-year changes in wages as our measure of growth. However, this measure could in theory miss (or at least obscure) quite recent accelerations or decelerations in the data.

Figure A presents our usual nominal wage growth measure for all (nonfarm) workers and a supplemental measure using a more timely (but more volatile) series measuring quarterly wage growth, and highlights a more recent period. The dark blue line in the graph measures growth relative to the same month in the previous year, and the light blue line shows wage growth measured as the average of the most recent three months relative to the average of the three preceding months. So, the last data point in this line is average wages in April, May, and June of 2019 as compared with average wages in January, February, and March of 2019. This growth is then expressed as an annualized rate to make it comparable with the other line.

Our original wage tracker shows clearly that wage growth steadily—but very slowly—improved in the years following the Great Recession. For example, growth was below 2% for much of 2010, but by December 2016 had risen to 2.7%. Moreover, as the figure shows, wage growth moved into a notably higher gear in 2018. From January 2018 to December 2018, year-over-year wage growth rose from 2.8% to 3.3%. And our more timely three-month measure of wage growth rose even faster in this period, consistent with wage growth acceleration.

One might even be tempted to call this evidence of heat. But in 2019, this growth has not just moderated but has actually decelerated slightly. In June 2019 wage growth was just 3.1%.

Figure A

The quarterly growth numbers (expressed at annualized rates) make the deceleration even clearer, as the more timely three-month measure of wage growth has dipped sharply below the year-over-year measure throughout 2019. In June 2019, this three-month measure of wage growth was just 2.7%.

As we’ve argued before, a nominal wage target really is the most important real-time indicator the Fed should be watching to assess economic overheating. If there’s no ongoing heat in this variable, the case for tightening is much, much harder to make.

Any signs of ‘heat’ in GDP, residential investment, and other variables?

Recent reports have indicated that the Fed’s intentions extend beyond just holding pat on interest rates to cutting rates in the next Federal Open Markets Committee (FOMC) meeting. Can we use the “heat check” of quarterly changes to shed any light on the wisdom of this decision by focusing on some other variables as well?

Table 1 shows growth rates for a number of variables, including average hourly earnings. It shows the annual growth rate for 2016, 2017, and 2018; the growth rate over the past six months; and the last quarter’s growth (with these last two growth rates expressed in annualized terms to make them directly comparable to the others). Essentially, this table aims to show longer-run trends in these variables as well as what has happened to them in increasingly recent periods. An “overheating” economy would see growth rates that were higher the more recently they were measured. In almost all cases we see the opposite of this pattern: Growth accelerates from 2016 to 2018, but then begins to decelerate.

Table 1

First, we examine growth in gross domestic product (GDP). Between 2016 and 2018 GDP growth accelerated from 1.6% to 2.9%. But in the latest six months, it decelerated to a 2.6% growth rate, and further decelerated to a 2.1% growth rate in the most recent quarter.

Next, we examine growth in the average hourly earnings measure shown in Figure A. Earnings accelerated from 2.6% to 3.0% between 2016 and 2018. But over the last six months the growth rate was 2.9%, and for the last quarter the growth rate was just 2.7%.

Growth in the price deflator for personal consumption expenditures excluding food and energy (the price inflation measure the Fed watches most closely) accelerated from 1.3% to 1.6% between 2016 and 2018. But the last six months’ growth was steady at this 1.6% rate, and growth ticked down slightly to 1.5% in the last quarter.

The next indicator we examine is residential investment—probably the component of GDP most sensitive to interest rate changes. As the Fed raised interest rates steadily (if slowly) between 2015 and 2018, this sector should be where we see evidence that these rate hikes have constrained growth. This is borne out in the data, as growth in residential investment slowed from 6.5% in 2016 to −1.5% in 2018. In the last six months residential investment decelerated a bit more slowly, contracting at a 1.3% rate. The last quarter saw a 1.5% contraction.

The second most interest-sensitive component of GDP is nonresidential fixed investment (NRFI—or business investment). NRFI accelerated from 0.7% growth in 2016 to 6.4% growth in 2018. In those years, the NRFI trends were dominated by changes in the prices of energy goods and services, with faster price growth in the energy sector inducing more business investment. But more recent data show a pronounced slowdown. In the last six months, NRFI grew at just a 1.9% rate, while it contracted 0.6% in the latest quarter.

Cooling economy does provide some basis for a rate cut

By many measures, the U.S. economy seems to be cooling. A Fed decision this week to cut rates would have some real evidentiary basis behind it.

Josh Bivens

For the past six months the Federal Reserve has been soliciting input to guide a reassessment of its “monetary policy framework.” This reassessment has been pegged to the 10-year anniversaries surrounding the financial crisis of 2008–09 and the Great Recession. While the Fed’s policy framework deserves much scrutiny, focusing too narrowly on what it could have done differently during the crisis and its aftermath would be a bad mistake.

The Fed failure that inflicted real damage on low- and middle-wage workers over recent decades was generally not insufficient effort in fighting recessions. Instead, the mistake was cutting short recoveries before they had maximized opportunities for employment and wage growth. In short, the time to worry about Fed actions that do not protect the interests of low- and middle-wage workers is during economic booms, not during slumps.

This newsletter explains why the Fed should keep the following points in mind as it undertakes its reassessment:

• Monetary policy is deeply asymmetric in its effects on macroeconomic stabilization. While interest rate increases work effectively to restrain growth, interest rate cuts provide very little boost to growth. This failure of interest rate cuts to spur growth—often analogized to “pushing on a string”—means that the Fed will frequently need help in fighting recessions. The limited effectiveness of Fed actions in spurring recovery from the Great Recession was not due to lack of effort, but to the inherent weakness of interest rate cuts as a tool.

• While it needs help in fighting recessions, the Federal Reserve has been given essential veto power on the issue of whether an economic expansion is proceeding too fast and needs to be restrained in the name of guarding against inflationary outbreaks. Too often the Fed puts the breaks on prematurely. These premature monetary policy contractions have cut recoveries short and kept unemployment unnecessarily high. This excess unemployment has deprived millions of people of potential work, and has been one of the single-largest factors explaining the disastrous wage stagnation afflicting the bottom 80 percent of the U.S. workforce for most of the post-1979 period. In fact, excess unemployment may well explain more than a third of the rise in wage inequality since 1979, and likely contributed to the redistribution of income from labor to capital owners. In short, excess unemployment is an absolutely primary source of the rise in American inequality.

The Fed did the right things during the Great Recession—but its tools are weak

Much of the discussion surrounding the reassessment of the Fed’s monetary framework focuses on whether the Fed used its institutional tools aggressively enough during the Great Recession and whether it needs new tools for future recessions. These are certainly questions worth considering, but they threaten to derail the debate from much more important ways that Fed decisions have undercut wage and job growth.

With regard to the Great Recession, the Fed responded earlier and more aggressively than any other major policymaking institution. It invented new tools on the fly, and the vast majority of these tools were used in the laudable aim of spurring faster recovery. These tools worked in the right direction, but their influence was weak due to the intrinsic weakness of interest rate cuts as levers to spur growth. Essentially, interest rate cuts make it cheaper for households and businesses and governments to take on debt to make purchases, but such cuts cannot force anybody to actually make these purchases. During recessions, when other influences often restrain economywide spending, the expansionary potential of lower interest rates can get swamped by these other factors. Importantly, the evidence documenting the weakness of interest rate cuts in spurring faster recovery is not driven entirely by episodes when the economy has hit the zero lower bound (ZLB) on interest rates. Instead, even when the Fed has room to cut interest rates further, these cuts deliver generally disappointing results in boosting growth. The ZLB just makes this weakness even more extreme.

Further, while the Fed took unprecedented policy actions to boost aggregate demand (spending by households, governments, and businesses) during and after the Great Recession, fiscal policy became historically contractionary over this period. In short, while there is some room for improvement in how the Fed approaches the next recession, there is little to suggest that the Fed’s fundamental framework for fighting recessions is flawed. But there is also little to suggest that Fed action alone will be enough to quickly end future recessions and restore full employment.

The Fed has too often cut recoveries short

The Fed’s actions throughout history have not always been supportive of economic recovery and growth in jobs and wages. Indeed, too often the Fed’s decisions have greatly exacerbated the economic struggles of low- and middle-wage workers. But the episodes when Fed action has damaged these workers have occurred when the economy is robustly expanding, not during recessions.

The Fed’s dual mandate instructs it to pursue the maximum level of employment consistent with stability in inflation. However, post-1979 its actions suggest that Fed policymakers have taken the inflation mandate more seriously than the full employment mandate. When an economic expansion pushes unemployment down, the Fed often fears that tighter labor markets will have workers demanding higher (nominal) wages. The intuition is that the best lever for workers to wring wage increases from employers is the threat (often implicit) that they’ll quit and find better-paid work elsewhere. This quit threat is far more credible during times of low unemployment.

Wage growth resulting from tight labor markets can potentially feed into price growth. In turn, workers may demand even higher nominal wages to make up for higher prices, allowing wage–price inflationary momentum to build. The policy recourse for the Fed to avoid this inflationary momentum is to raise interest rates to slow the expansion and stop the downward movement of unemployment.

But when is unemployment at the sweet spot of allowing all workers a chance at decent work and wage growth but not fostering unsustainable inflationary pressures? Nobody knows for sure beforehand. Efforts at empirically identifying the economy’s “natural rate of unemployment” are notoriously imprecise. Given this uncertainty, the Fed must exercise judgment in weighing the benefits of tighter labor markets against the risks of building up inflationary pressures. Far too often in the post-1979 period, Fed policymakers have been too worried about the inflation risks and not impressed enough by the full-employment benefits.

Figure A shows real-time estimates of the natural rate of unemployment and the actual unemployment rate. Set aside for a second the possibility that the estimated natural rate may itself be too conservative. The post-1979 Fed was far more likely to hold the economy above the natural rate than the Fed was in the 30 years before 1979. In the 30 years between 1949 and 1979, the actual unemployment rate was a cumulative 15 percentage points below the natural rate. But in the 28 years between 1979 and 2007, the actual unemployment rate was a cumulative 15 percentage points above the natural rate.

Figure A

This clear shift in the Fed’s monetary policy framework to a more heavy weighting of the inflation mandate coincides strikingly with the post-1979 slowdown in wage growth for the large majority of American workers. Figure B shows the relationship between a typical worker’s compensation per hour worked and economywide productivity (a measure of total income generated in the economy in an average hour of work). Between 1949 and the mid-1970s, these measures rose in lockstep. Since 1979, rising inequality has driven these two measures decisively apart.

Figure B

The relationship between these two shifts—toward a monetary policy framework more tolerant of high unemployment and toward a labor market that does not generate much wage growth for the vast majority—is confirmed in more detailed statistical tests of correlations between unemployment and wage growth. These tests show that lower unemployment boosts wage growth more for low- and moderate-wage workers than it does for higher-wage workers. To put it simply, low- and middle-wage workers need the leverage and bargaining power that accompanies tighter labor markets more than high-wage workers do. This is especially true in the post-1979 period, as institutional bulwarks of bargaining power for these workers—like unions, high federal minimum wages, and relatively low levels of imports from low-wage nations— were all intentionally eroded by policymakers.

Encouragingly, there is good evidence that tight labor markets can temporarily offset much of the decline in these structural determinants of wage growth. In the late 1990s the Fed admirably allowed unemployment to fall well beneath real-time estimates of the natural rate. The result was not inflation, but the first sustained period of across-the-board wage growth in a generation. Figure C shows average annual wage growth of workers at the 20th, 50th, and 95th percentiles of the wage distribution for two periods: 1996–2001 (the period of tight labor markets) and every other post-1979 year. It shows that essentially all wage growth that happened after 1979 for low- and middle-wage workers happened in the five-year burst when labor markets were tight.

Figure C

How different would the world have been if the Fed allowed substantially tighter labor markets post-1979?

In a recent-ish paper, Ben Zipperer and I found that each 1 percentage-point drop in the unemployment rate was associated with annualized wage growth for low- and middle-wage workers that was faster by 0.6 and 0.5 percent, respectively. We noted before that actual unemployment was cumulatively 15 percentage points above the estimated “natural rate” between 1979 and 2007. Thus, a very rough estimate is that wages for these low- and middle-wage workers could have been 9.0 percent and 7.5 percent higher (respectively) in 2007, but for the effect of excess unemployment (multiplying 15 percentage points of excess unemployment by the coefficients of 0.6 and 0.5). This would have resulted in wage growth from 1979 to 2007 that was 75 percent faster than what workers at the 50 percentile experienced. Workers at the 10th percentile saw outright declines in wages between these years—tighter labor markets could have swung their cumulative wage growth from -4 percent to (positive) 5 percent. If wages at the top of the distribution were not affected by tighter labor markets (often coefficients on wage growth for high-end workers are right on the edge of statistical significance), this would imply that roughly 30–40 percent of the rise in inequality in wages that occurred between 1979 and 2007 could have been avoided (if inequality is proxied by the ratio of wages at the 95th percentile to wages at the 10th and 50th percentiles).

Additionally, a previous newsletter noted that tight labor markets also lead to a redistribution away from capital income and toward labor income overall, a shift that also would have allowed faster wage growth for most workers.

Of course, these are crude estimate with lots of caveats. The most important one is that perhaps inflation would’ve been higher and/or accelerated, making the tighter labor markets impossible to sustain. But there is ample evidence that the Fed didn’t just engineer stable inflation by keeping labor markets intentionally soft in those years; it overshot and saw steadily declining inflation rates. One objection to these estimates might be that it’s unrealistic for the Fed to have managed to hit the natural rate target on average over this entire 28 year period. But this objection would be wrong. For one, the Fed did manage to hit the natural rate (or even undershoot it) over the previous 30 years. For another, estimates of the natural rate should not be seen as hard floors below which unemployment can never be allowed to reach. Instead, extended periods when actual unemployment exceeds the natural rate should be matched by equally extended periods when actual unemployment dips below the natural rate.

What does this mean for the reassessment of the monetary policy framework?

The lessons of this history for the monetary policy framework are clear: the Fed should certainly try to maximize its effectiveness in fighting recessions, but its real opportunity for helping low- and middle-wage workers lies in being slower in raising interest rates to restrain expansions that are already underway.

The Fed and other central banks have often viewed their main job as “taking away the punch bowl just as the party gets going,” meaning that they think they have to restrain expansions and stop downward movement in unemployment before an overheating economy sparks inflation. Modern economies do need policymakers to keep an eye on inflation, but central bank orthodoxy before the Great Recession had swung way too far towards sacrificing large and progressive benefits of genuine full employment on the altar of inflation control.

The most important change the Fed could make in its monetary policy framework may just be a simple commitment to wait until inflation durably exceeds the Fed’s target in the data before it begins raising interest rates. Currently, the Fed and other central bankers worry about being behind the curve in containing inflation, and hence raise rates in expectation of future accelerations of inflation. The benefits of this kind of excess vigilance in fighting forecast inflation are utterly swamped by the potential benefits that would accrue to low- and middle-wage workers if tight labor markets were allowed to persist and inflation did not turn up. This is absolutely a low-stakes, high-return risk the Fed should commit to taking on as it reassesses its policy framework.

Josh Bivens, Director of Research

In a previous edition of this newsletter, I highlighted the labor share of income as a target variable the Fed should be monitoring to assess whether or not the U.S. labor market has returned to full health. Specifically, I argued that a period of above-trend growth in wages should be allowed if it leads to steady clawbacks in the share of national income that labor lost during earlier phases of the economic recovery and expansion. Figure A shows the evolution of labor’s share of income in the corporate sector in recent decades; it clearly documents that the post–Great Recession collapse in labor’s share has still largely not been reversed.1

Figure A

That newsletter wasn’t the first time I’ve stressed that the Fed should allow ever-tighter labor markets until the labor share of income normalizes or until there is an extended period of above-target price inflation. The argument, put simply, is this: If price inflation accelerates well before the clawback of labor’s share of income, this would be some evidence that structural changes (perhaps growing industrial concentration of product markets) have contributed to the fall in labor’s share, and that tighter labor markets by themselves will not be enough to return this measure to pre–Great Recession levels. However, if we don’t see this outbreak of extended above-trend price inflation well before any clawback, it means the Fed can and should strive to restore labor’s share by keeping labor markets tight.

In today’s newsletter, I follow up on those arguments, looking specifically at whether there really is a reliable positive effect of tight labor markets on labor income shares. If there is such a reliable effect, this provides a strong argument that the Fed should keep labor markets tight until labor’s share moves much closer to its pre-recession levels.

This analysis yields three key conclusions:

• Recent years have seen the relationship between low unemployment and faster price inflation greatly attenuated. Put simply, in the last decade tighter labor markets have seemed less and less likely to lead to rapid price acceleration. This means the risk to the Fed’s inflation mandate from allowing tight labor markets has substantially declined.

• There is indeed a reliable effect of sustained tight labor markets leading to increases in labor’s share of income. This finding convincingly shows that the conventional wisdom characterizing the labor share of income as simply “countercyclical” is far too simple and masks important dynamics of the labor share over business cycles.

• For each year that today’s state of labor market tightness persists, we should see a nearly 1.5 percentage-point increase in labor’s share of income. As of the fourth quarter of 2018, the U.S. labor share of income was roughly 4 percentage points below its pre–Great Recession peak. This means that we would need to maintain today’s state of labor market tightness for more than two years to see a return to pre-recession labor share.

Let’s go into a bit more detail on each of these points.

How do we know we’re at less risk of rapid price acceleration from sustained tight labor markets?

Traditionally, the primary argument against allowing “high-pressure” labor markets to persist for a sustained period was that this risked an outbreak of price inflation well above the Fed’s target. The fear was that even a brief spell of above-target inflation could set off a wage–price spiral that would require large interest rate hikes to quell, and that these rate hikes would substantially slow growth or even put us into a recession.

New analysis from Jared Bernstein of the Center on Budget and Policy Priorities demonstrates that the relationship between unemployment and price inflation has greatly attenuated in recent decades. Figure B adapts a figure from Bernstein’s forthcoming report showing a rolling regression of price inflation predicted by the “unemployment gap”—the difference between the actual unemployment rate and the Congressional Budget Office’s (CBO’s) 2010 estimate of the natural rate of unemployment.

A rolling regression is an analysis of the changes in the statistical relationship between variables over time. In this case, the regression’s dependent variable is price inflation, and it is being correlated with the unemployment gap. The entire data set includes quarterly data from 1959 to 2018. But for the rolling regression, instead of using the entire data set to estimate the relationship between the unemployment gap and inflation, the regression is instead first estimated over just the first 80 quarters of data. The regression coefficient representing the correlation between the unemployment gap and inflation is then plotted. Then the first observation is dropped and replaced with the 81st quarter of data, and then the first observation of this new data set is dropped and replaced with the 82nd quarter of data, and so on. Each subsequent regression provides a new measure of the correlation between the two variables, and each subsequent regression uses more recent data than the last.

The striking finding evident in Figure B is that correlation between the unemployment gap and price inflation (the middle line) steadily weakens over time, becoming economically and statistically insignificant in recent years. In the graph, this is seen as the middle (green) line plotting the regression coefficient in successive years rapidly approaching zero since the mid-1990s.

Figure B

In short, this analysis confirms what many have been arguing in recent years: There is little danger that tight labor markets will quickly accelerate inflation to levels above the Fed’s target zone. Given this, targeting other variables with high-pressure labor markets seems like it can be done at quite low risk.

Yes, tight labor markets do lead to a rising labor share of income

It is often asserted that the labor share of income is a countercyclical variable—rising during recessions and falling during recoveries.2 This is an incomplete description, as its dynamics are quite a bit more complicated than that.

Figure C shows the labor share of income since 1949. In this figure, we highlight when it reaches local maximum levels during recessions—justifying an insight that this variable is countercyclical with regard to its increase during recessions. Moreover, in early parts of recoveries, it tends to fall rapidly, again potentially justifying the countercyclical description. But in later periods of recoveries it almost always rises sharply, well before (sometimes years before) the subsequent recession. (In the figure, these later-in-recovery upswings are marked with red arrows.)

Figure C

While this empirical regularity is much less well known than the incomplete description that labor’s share is “countercyclical,” it hasn’t completely escaped notice. Richard Clarida noted this pattern in a 2016 journal article, before he became Fed vice chair:

Note that this pattern of a mid-cycle rebound in labor’s share is typical of past U.S. recoveries. In the three expansions during the “Great Moderation”—the quarter-century after then-Fed Chair Paul Volcker broke the back of inflation—labor’s share of income initially fell during the early days of recovery and then began to rebound during the expansion phase of the business cycle. Importantly, this rise in labor’s share occurred before, and usually well before, the business cycle peak and continued as the economy fell into recession. The rise in labor’s share that occurs during recessions is well known and is usually attributed to the desire of firms to “hoard” labor initially in downturns as sales decline—holding off on firing workers until the decline in demand is clearly expected to persist. What is less appreciated is the phenomenon of labor’s rising share of income well in advance of the peak in economic activity and for reasons unrelated to labor hoarding.

Figure C also shows lots of 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, indicating that low unemployment gaps (signifying tight labor markets) are weakly associated with an increased labor share. Once decade-specific dummy variables and decade-specific trends are controlled for, however, this relationship dramatically strengthens, as shown in Figure D. 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 matching or below estimates of the natural rate. An unemployment gap of 1 or below indicates an economy operating below full employment, but at least within shouting distance of it. An unemployment gap of above 1 indicates an unhealthy labor market. For reference, the CBO’s estimated natural rate of unemployment for this year is 4.6 percent, making today’s unemployment gap roughly -0.8 percent.

Figure D

Given today’s unemployment gap, the regression coefficients shown in Figure D would indicate that another year at this low rate of unemployment should see a rise in the labor share of income by just over 1.5 percentage points, all else equal.3 The decline from the nonrecessional peak of labor’s share of income in 2007 until today has been roughly 4 percentage points, meaning that more than two years of labor markets as tight as today’s would be needed for full recovery of labor’s share.

What will rise in an extended high-pressure labor market—price inflation or labor’s share?

So far, the tighter labor markets of recent years have not translated into a robust clawback of the labor share of income lost in the early part of this recovery. But the evidence from the past six decades-plus of data seems to clearly indicate that eventually a hot labor market will translate into such a clawback.

Data on more recent periods indicate that while the labor share of income should reliably rise in high-pressure labor markets, price inflation may not necessarily increase. Avoiding rapid price accelerations is the main justification for Fed rate increases to slow the pace of growth and avoid the risk of wage-driven price inflation. This risk seems to have abated substantially. Given this, the Fed should go full speed ahead in targeting a return to pre-crisis levels of labor’s share of income. The evidence seems clear that sustained tight labor markets will eventually deliver this.

1. In the January newsletter, we provided details on how our labor share measure is constructed. Briefly, this labor share is labor compensation as a share of the sum of labor compensation and net operating surplus in the corporate sector.

2. For example, the first sentence in the abstract of a 2018 paper by Vasco Botelho begins, “There is a consensus that the U.S. labor share is countercyclical….” The paper then goes on to argue that the cyclical dynamics of the labor share have changed substantially in recent years.

3. To see this, multiply the unemployment gap (-0.8 percent) by the regression coefficient (-0.47 percent) and then multiply this quarterly relationship by 4 to calculate the effect for a full year.

Josh Bivens, Director of Research

Why have wages grown so slowly in recent years despite relatively low unemployment rates? This puzzle has dominated economic commentary.

Figure A below, for example, shows a scatterplot of quarterly nominal wage growth (measured against the same quarter in the previous year) and unemployment rates since 2008. The trendline showing the relationship between these variables demonstrates it’s very weak—both statistically and economically insignificant.

Figure A

In this newsletter, I address a number of questions raised by this weak relationship between unemployment rates and wage growth since 2008. My key conclusions are:

• Since 2008, the share of adults between the ages of 25 and 54 who are employed (or the “prime-age EPOP”) has predicted wage growth better than the unemployment rate.
• But even the prime-age EPOP has done a poor job at predicting wage growth since 2008 compared with both its own predictive power pre-2008 and the predictive power of the unemployment rate in earlier periods.
• The prime-age EPOP’s advantage in predicting wage growth seems to have started even a bit before the Great Recession, around 2001.
• Because both the unemployment rate and the prime-age EPOP have seen a large reduction in their predictive power regarding wage growth since 2008, efforts to explain this decline in predictive power should involve looking to the unique features of the Great Recession: very high rates of unemployment combined with very low rates of inflation.
• While both the unemployment rate and the prime-age EPOP are likely to be fine statistical predictors of wage growth moving forward, there has been a steady decline in how responsive wage growth is to a given change in either. In short, workers have seemingly needed ever-tighter labor markets (measured by quantity-side variables like the unemployment rate and the prime-age EPOP) to generate a given amount of wage growth.

If unemployment fails, does anything predict post-2008 wage growth?

The weak connection between unemployment rates and wage growth since 2008 has led to many analysts looking for signs of labor market slack in other measures. For example, to be classified as unemployed, someone must be first classified as in the labor force, which requires that person to be actively engaged in a job search. If an extended period of weak hiring leads significant numbers of potential workers to despair of finding a job, then they might cease to look for work and drop out of the labor force, which lowers the measured unemployment rate. In early parts of the recovery from the Great Recession (until roughly 2015), this reduction in labor force participation actually explained about a third of the overall decline in unemployment.

A measure of labor market health uninfected by sometimes-arbitrary labor force participation decisions is the employment-to-population ratio (EPOP). This simply measures what share of adults have jobs, regardless of whether the jobless are classified as “actively” searching for work or not. Because large waves of baby boomers are currently reaching retirement age, the overall EPOP has trended down in recent years for demographic reasons. A measure that circumvents these demographic effects is the “prime age” EPOP—the EPOP measuring only adults between the ages of 25 to 54. At EPI we have often highlighted the importance of using the prime-age EPOP to supplement other measures of labor market health. In The Upshot from The New York Times, Ernie Tedeschi showed that a related measure (an “adjusted” EPOP that accounted for the effect of demography) predicted nominal wage growth quite well between 1990 and 2017.

This backstory raises the question of whether the prime-age EPOP has done a significantly better job since 2008 of predicting wage growth. Figure B shows a scatterplot like Figure A but with the prime-age EPOP substituted for unemployment.

Figure B

As the figure shows (and statistical tests confirm), the prime-age EPOP has done better at predicting wage growth than the unemployment rate has since 2008. The prime-age EPOP’s effect on wage growth is both economically and statistically significant, as it explains roughly 30 percent of the variation in wage growth over this period (this statistic is reported later, in Table 1).

But that’s actually a pretty poor performance! In earlier periods, the unemployment rate predicted a far higher share of variation in wage growth than this.

Figure C shows a scatterplot of the relationship between real wage growth and unemployment from 1979 to 2007. Over this longer period price inflation varied a lot, szo we can no longer use nominal wages but rather must account for price changes. The reasoning is simple: When inflation is hovering around 2 percent (like it has since the early 1990s), nominal wage growth of 4 percent would be pretty healthy—generating inflation-adjusted wage increases well in excess of productivity. But when inflation is hovering around 8 percent (like it did in the early 1980s), nominal wage growth of 4 percent is a disaster, leading to steep inflation-adjusted wage cuts for workers. In short, against the backdrop of unstable inflation that characterizes a long period like 1979 to 2007, a given nominal wage growth figure can mean very different things regarding the health of the labor market.

To calculate real wages, I generated my own measure of expected inflation. The idea is that workers bargain over nominal wages with their employers with some idea of future expected price inflation in the backs of their minds. If workers assume prices will rise 5 percent over the next year, they know that nominal wage growth of less than 5 percent constitutes a cut in real living standards. Workers’ expectations of price growth can be reasonably represented by a rolling average of the recent inflation experience, with more-recent inflation rates weighted more heavily. My expected inflation measure incorporates data from the previous 12 quarters. Year-over-year inflation rates lagged one to four quarters are each given a weight of 12.5 percent. Inflation rates lagged five to eight quarters are given a weight of 10 percent, and inflation rates lagged nine to 12 quarters are given a weight of 2.5 percent. None of the results that follow are sensitive to precise weighting decisions.

Figure C

The relationship between unemployment and wage growth is striking in comparison with the similar scatterplot in Figure A. Before 2008, the unemployment rate predicted wage growth quite well; far more accurately, for example, than the prime-age EPOP has since 2008.

What was the better predictor of wage growth before the Great Recession?

The predictive power of unemployment before 2008 raises the question of whether the superior performance of the prime-age EPOP in predicting wage growth is only a recent phenomenon. The short answer is that the prime-age EPOP seems to be doing better than the unemployment rate as a wage growth predictor only since around 2001. Table 1 below simply reports the coefficient and r-squared values of regressions of nominal wage growth on the unemployment rate or the prime-age EPOP and our measure of expected inflation (where r-squared in this case is the share of total variation in wage growth that is explained by each regression model).

Table 1

Before 2000, the regression with the unemployment rate outperforms the prime-age EPOP, by a relatively clear margin, judged on the basis of the r-squared. Between 2001 and 2007, the performance is very similar, with the prime-age EPOP seemingly having a slight advantage. It is only in the post-2007 period that the prime-age EPOP is clearly superior. And its advantage is starting to erode post-2013.

The reason the prime-age EPOP seemed to be the inferior predictor of wage growth pre-2001 seems clear: it had a strong structural (upward) trend that was largely unrelated to the overall health of the labor market. This trend was driven by the rising share of women entering the workforce (even as there was a very mild downward trend in male labor force participation). But since 2001, there has been no upward trend in women’s prime-age EPOP.

Lessons going forward

Where does this leave us? First, while the unemployment rate failed to predict wage growth accurately after the Great Recession, this doesn’t necessarily mean that we should throw it out forever as a predictive tool. Table 1 shows that the really poor performance mostly occurred between 2008 and 2013. Figure D depicts this. It shows actual wage growth along with predicted wage growth based on the relationship between wage growth and unemployment that prevailed between 1979 and 2007. The unemployment rate (along with expected inflation) does a fine job predicting wages until 2007—meaning that the actual and predicted wage growth lines are very close together. After 2007, the lines diverge, and radically so between 2009 and 2013. Basically, in the first few years into the recovery, inflation plummeted and the unemployment rate rose to some of its highest levels in the postwar period. These two influences lead our model to predict steeply negative nominal wage growth for a spell. But, as much research has pointed out, nominal wage growth seems rigid at zero and rarely goes outright negative.

Figure D

This leads to one clear lesson from this examination. Because historical relationships can go haywire when the economy becomes deeply depressed for a spell, we must make sure that the portfolio of indicators used to measure labor market health and to predict future wage growth is large. Relying on any single measure can lead to bad policymaking choices.

Finally, just because a statistically detectable relationship between unemployment rates and wage growth seems to be reasserting itself post-2013, this doesn’t mean policymakers should be on a hair-trigger impulse to slow growth for fear of inflation. Modest wage growth will not necessarily feed directly into price inflation because much of the post-recession decline in labor share of income has yet to rebound, and because wage pressure might actually induce some faster productivity growth.

More importantly, the reemergence of any statistical relationship at all between lower unemployment and wage growth is interesting—but the magnitude of this relationship is important to note. This magnitude has been falling steadily over time, as shown in the regression coefficients on unemployment in Table 1. Essentially, a given rate of wage growth requires ever-lower unemployment, a finding we’ve highlighted before. This is what you’d expect if structural determinants of wage growth were pushing down on workers’ bargaining power and leverage in the labor market. But this downward structural pressure on wages is also something that macroeconomic policymakers, including Federal Reserve officials, should keep in mind as they decide whether inflationary pressure is building in the economy.

Three percent growth for a quarter, however, should not constitute “mission accomplished” in the minds of macroeconomic policymakers like the Federal Reserve. In the long run, nominal wage growth should run at a rate equal to the Fed’s inflation target (2 percent) plus the long-trend growth in potential productivity (let’s call this 1.5 percent).1 This indicates that even the recent accelerations in wage growth leave us failing to meet these long-run goals.

Even more importantly, wage growth should run substantially above these long-run targets for a spell of time after long periods of labor market slack. The arithmetic reasoning for this is straightforward: any time wage growth runs slower than current rates of inflation plus productivity, the result will be labor compensation shrinking as a share of the economy. The economic intuition is simply that extended periods of labor market slack sap workers’ ability to secure wage increases from employers.

This undermining of labor’s leverage shows up clearly in the data. EPI’s nominal wage tracker, besides charting wage growth over time, also tracks a measure of labor’s share of income, precisely to highlight the accumulated shortfall of labor income that policymakers should aim to restore.

Figure A

Measuring the labor share correctly

We think tracking labor’s share of income is crucial for policymakers, and that this figure contains lots of interesting insights about macroeconomic dynamics. Before we delve into these insights, however, we should first note that our measure above is not the only way to measure labor’s share of income. In particular, a measure obtained directly from the Bureau of Labor Statistics (BLS) productivity series is often referenced. This measure—reproduced below and shown side-by-side with our preferred measure—shows some non-trivial differences in the labor share over time. Most strikingly, the BLS measure has a much stronger long-run downward trend than the measure we use.

Figure B

What explains these differences? Three things: First, our measure examines labor’s share in the corporate sector only, while the labor share from the BLS productivity series looks at the broader non-farm business (NFB) sector. Second, our measure uses net, not gross, measures of income. So, our measure strips out depreciation from the income in the denominator of the labor share. Third, our measure strips out the profits of Federal Reserve banks. We think each of these decisions yields a better measure if you want a labor share indicator that sheds light on the relative bargaining strength of workers vis-a-vis employers—below we explain why.

First, we focus on the corporate sector rather than the total economy or even the NFB sector. We make this choice because in the corporate sector all income is either labor income or capital income. Since we are interested in using changes in the labor share as measures of relative bargaining strength between workers and employers, this clean distinction between incomes is useful. In the total economy, there are many income flows that do not necessarily tell us much about this relative bargaining strength. Rental incomes accruing to property-owners, for example, can come at the expense of both workers and their employers, and can in theory reflect land scarcity rather than anything profound about the current state of macroeconomic slack. Proprietor’s incomes are a mix of labor and capital incomes, and correctly apportioning which share of these should be chalked up to returns to labor versus capital is near-impossible. The corporate sector divides income cleanly into labor versus capital, and yet remains large enough (accounting for over 70 percent of the U.S. private sector) to provide generalizable findings.

Second, we focus on net measures of income that account for depreciation. Depreciation is the reduction in the value of assets that occurs through wear and tear or technological obsolescence. If income from current production is not used to reinvest and keep the value of the capital stock whole in the face of constant depreciation, this would lead to a smaller capital stock and reduced productivity over time. Hence, depreciation represents a claim on income that cannot boost the living standards of either workers or capital owners. Further, the share of depreciation in corporate sector value-added (as well as its share in total economy gross domestic product) has been rising steadily over time. The reason for this rise is well understood: information and communications technology equipment has been steadily rising as a share of the overall capital stock, and this type of capital depreciates far more rapidly than other types (think of how often you need to replace your laptop or cellphone these days). This steady rise in depreciation is a significant reason why the decline in the labor share shown in the BLS NFB measures is larger than in our measure. Yet the decline in the labor share that’s due to the rise of depreciation really doesn’t tell us much about labor market dynamics or macroeconomic slack.

Finally, we choose to strip out the profits of Federal Reserve banks for our measure.2 In the course of their open market operations (buying and selling bonds to move interest rates), the Fed makes profits. These profits are used to finance the Fed’s operations and they remit anything left over to the U.S. Treasury. These profits have very little to do with underlying labor market dynamics or macroeconomic slack. Before 2008, because these profits were generally quite small and relatively stable as a share of corporate sector value-added, they could largely be ignored in examinations of the labor or profit share. Post-2008, however, the Federal Reserve vastly expanded the scale of assets it purchased as part of its effort to hold down long-term interest rates (an effort commonly referred to as “quantitative easing”). This larger balance sheet led to significantly larger Federal Reserve profits during the recovery from the Great Recession. All else equal, this would have led to a further decline in the measured labor share of income. But since these Federal Reserve profits were not the result of increased leverage vis-a-vis workers over implicit wage bargaining, it doesn’t make much sense to allow them to depress the labor share measures we’re examining for evidence about this relative bargaining strength. In recent years, the Federal Reserve profits have been declining as their balance sheet has shrunk modestly. This would, all else equal, boost the labor share of income, but this boost would not represent increased leverage by workers. Given these considerations, our measure of corporate sector income and profits removes the profits earned by Federal Reserve banks. This again leads to a smaller decline in the labor share when compared to the BLS measures over the whole post-Great Recession period.

What does a correctly-measured labor share tell us about policy?

Assume for the moment that our way of tracking the labor share is better for assessing underlying labor market dynamics and the state of macroeconomic slack. This measure contains a number of interesting findings. One finding, which most might find counterintuitive, is that during recessions (the shaded areas on the chart) the labor share actually tends to rise. This mostly tells us that corporate profits are very cyclical, and fall even faster during recessions than labor income.3 But once the official recession is over, these profits tend to recover much faster than labor income and this leads to a rapid decline in labor’s share in the early phases of business cycle recoveries. As recoveries mature and turn into expansions, the reduction in labor market slack eventually allows the labor share to begin making up some of its early-recovery losses.

Another striking feature of our measure of the labor share is how depressed it remains even as we approach 10 years from the end of the Great Recession. This highlights that even wage growth significantly faster than we’ve seen in recent months can persist for a long time before it necessarily translates into inflation breaching the Fed’s 2 percent target. If inflation-adjusted wages start rising faster than economy-wide productivity, this does not necessarily lead to an inflationary wage-price spiral; instead some of this faster wage growth can be absorbed by firms in coming years through a slight reduction in their still quite-fat profit margins. In an earlier paper, we calculated for how long nominal wage growth could outpace the sum of inflation and productivity growth before previous labor share peaks were broached.4 The figure below reproduces these calculations, using the Fed’s 2 percent inflation target and productivity growth averages since the beginning of 2017 (1.1 percent), and various scenarios for nominal wage growth.

Figure C

The results are striking. The 3.2 percent wage growth that characterized the last quarter of 2018 would take literally decades and decades to yield a return to the labor share that characterized the last quarter before the Great Recession hit (the last quarter of 2007, or 2007Q4). 3.5 percent growth—the long-run target we referenced before—would only return the economy to the 2007Q4 labor share level by 2031. Even 4 percent nominal wage growth would only see the labor share return to 2007Q4 levels by the end of 2024. We should note here how fundamentally conservative these projections are, in that they assume productivity growth only runs at 1.1 percent in coming years, even as labor markets tighten and wages rise. If productivity growth—spurred by firms looking to make labor-saving investments as wages rise—instead moves closer to its long-run average of 1.5 percent, then it would take even longer at any given pace of wage growth to return to historical labor share levels.

Is it possible that structural changes, not just prolonged labor market slack, have contributed to the fall in the labor share and macroeconomic reflation alone won’t be able to claw back this post Great Recession decline? It’s possible—and the most-sensible theory for what precise structural change might have led to this is a rise in monopoly power. Empirical research clearly shows upticks in industry concentration ratios, and a number of large and crucial economic sectors (health care and finance, in particular) are characterized by immense pricing power. Other sectors (mostly technology) have also seen waves of consolidation that have not led (yet?) to rapid price growth, but have raised a whole host of concerns about the intersection of economic and political power.

And yet the first priority in setting the conditions for better wage growth remains pushing down the unemployment rate and wringing out any last sign of slack from the labor market. In fact, our examination of labor share indicates that we can’t even diagnose the extent of the monopoly problem’s effect on wages and incomes until we wring out this labor market slack and spark some wage-driven price inflation.

Consider the recovery following the recession in the early 1990s. Between the third quarters of 1990 and 1997, labor’s share of income in the corporate sector fell by almost 5 percentage points. This a big change—a change like that today would translate into about $600 billion being claimed by capital-owners rather than workers. So, this fall in labor’s share was large, but it was also quite persistent; seven years is a long time to see the labor share not recover from a recession-induced fall. Because most economists in 1997 thought the economy was clearly at full employment even as sluggish wage growth kept labor share flat, new theories were pushed forward (just like today) about how sluggish wage growth must be being driven by something else besides labor market slack. Skill-biased technological change was the fashionable explanation at the time.

But, three years later, in the third quarter of 2000, labor share was just 0.3 percentage points below its 1990 peak and wages had risen sharply across-the-board. What changed in those three years between 1997 and 2000? Unemployment continued to fall and the share of employed adults (and prime-age adults) continued to rise, tightening up labor markets. The real innovation of this period was a Fed that did not preemptively raise interest rates to choke off falling unemployment; instead they waited for actual wage-driven price inflation to appear in the data. When it didn’t, they let the recovery continue.

Today’s Fed should follow this advice and assume that the U.S. economy can restore the labor share of income lost during the recovery from the Great Recession; at least until the data tells us that it cannot by showing a large jump in inflation before this share is restored.

This is a preview edition of a new macroeconomic policy briefing that will be published by the Economic Policy Institute before each FOMC meeting. These analyses will take a deep dive into one aspect of monetary or other macroeconomic policies and data. To receive the briefing in your inbox, sign up below.

1. See Bivens (2015) and Bivens (2017) on the rationale for this nominal wage target and for evidence that recent years’ productivity malaise should not lead to large downward revisions in estimates of the economy’s potential productivity growth rate.

2. We call these “profits” of Federal Reserve banks because that’s what the Bureau of Economic Analysis (BEA) labels them in the data. Given that the Federal Reserve system is not a private business, it’s a bit of a strange label for lots of reasons, but we’ll follow the convention here.

3. This recession-induced rise in the labor share is most striking in the last half of 2008, as the write-off of the value of bad loans in the financial sector was reflected in large, one-time declines in corporate profits, sending labor’s share soaring for these quarters.

4. See Figure 7 in Bivens (2015).