Legislative activity will likely begin in earnest next week on the Biden administration’s $1.9 trillion relief and recovery package. Recently, the Biden package has attracted criticism for being too large, but that criticism assumes far too-mechanical a relationship between one-time fiscal support and a durable recovery.
This post highlights the following points:
- $1.9 trillion is the right scale of fiscal support that the U.S. economy needs over the next two years; it is in no way “too much.”
- Current estimates of the “output gap”—the difference between actual gross domestic product (GDP) and what GDP could be if unemployment was at its lowest sustainable level (potential GDP)—are understated.
- Most importantly, the relevant output gap to compare the Biden package to is the cumulative gap over the next couple of years. Measured correctly, this cumulative output gap signals the need for a package as large as the Biden proposal.
- Relatedly, it is not a problem if a large chunk of the $1.9 trillion in relief and recovery is distributed gradually (either by policy design or through households smoothing their consumption out of relief aid) over the next couple of years – in fact it would be a good thing.
Recently, concerns have been raised that the Biden relief and recovery package might be “too large” relative to current estimates of the output gap. The specific concern is that if too much fiscal support is poured all at once into the economy, then the result will not be more jobs and higher incomes, but accelerating inflation and higher interest rates (which could in turn “crowd out” productive private investment).
These concerns should be largely discounted. As background, we should simply note that after a generation of going too-small in providing fiscal support to aid the economy’s growth, a move to erring on the high side would be welcome. But even more importantly, the Biden package almost certainly does not err on the high side of what is needed to support economic recovery in coming years.
Those expressing concern that the Biden package is too-large often point to measures of the output gap estimated by the Congressional Budget Office (CBO). In the last quarter of 2020, this output gap was roughly 3% of potential GDP. The Biden package would be just under 7% of 2020 GDP. It is this difference between today’s output gap and the size of the relief and recovery package that has generated some criticism.
However, there are two problems with this view: today’s output gap is larger than what the CBO estimates, and, it is not only today’s output gap that needs filled in, but, the cumulative gap over the next couple of years signals that fiscal support will be needed for longer than a year.
Today’s output gap is larger than what CBO estimates
The clearest signal that CBO’s output gap forecasts are understated can be seen in what they estimated in the immediate pre-COVID era (the numbers in the following paragraph—plus some others—can be found in Table 1 below). In the last quarter of 2019, CBO estimated that the U.S. economy was operating above its potential capacity, by about 1% of potential GDP. But there is almost no reason to think that the economy was operating above capacity then. The main data signature for an economy operating above its productive potential is upward pressure on wage growth, price inflation, and labor’s share of income. But at the end of 2019, “core” price inflation (excluding food and energy prices) measures monitored by the Federal Reserve, for example, was both beneath the Fed’s long-run target of 2% and was barely accelerating at all—having moved up just 0.2% in the previous two years. Wage growth in 2019 continued to lag behind productivity growth, and was also slightly decelerating. The labor share of corporate sector income was far beneath its high point of recent decades.
Selected indicators of labor market tightness during 3 recent episodes
|ECI hourly pay, % change||1.7%||0.6%||1.0%|
|Productivity – pay||1.8%||0.8%||0.6%|
|Inflation, core PCE||1.5%||2.2%||1.8%|
|Labor share of corporate income||81.7%||76.7%||76.7%|
|2-year change labor share||2.7%||3.7%||1.0%|
|CBO output gap||1.0%||0.8%||1.0%|
Source: Author’s calculations. Unemployment, hourly pay, productivity, inflation and the labor share are all calculated as 2-year averages. Inflation change and the labor share change are calculated as a change between the last quarter in the period to the same quarter a year before. The measure of hourly pay is the Employment Cost Index (ECI), total compensation, all civilian employees. To better-align it with the measure of productivity (the real output per hour measure from the nonfarm business sector productivity series), nominal ECI is deflated with the gross domestic product (GDP) price deflator.
While it’s true that the unemployment rate was historically low in 2019—averaging 3.8% over the previous two years—this far from seals the case for the economy operating above potential. For example, before 2019, the late 1990s are often pointed to as the last time the U.S. economy seemed to be close to a high-pressure labor market, one in which unemployment was low enough to give workers the leverage they needed to realize decent wage growth. In 1999 and 2000, unemployment averaged 4.1% and saw wage growth of just under 2%. As they did for 2019, CBO estimates indicate that the U.S. economy was operating above potential in those years. But again, there is little reason to believe that. Price inflation in those years averaged less than 1.5%, and wage growth—while healthier than it had been for decades before or since— actually lagged significantly behind productivity growth (even when deflating wages by a product deflator). Further, the labor share of corporate sector income was roughly the same as it had been during the peak of the late 1980s business cycle.
Why does this episode from 1999–2000 matter? Because even if a 4% unemployment rate in those years really had defined something close to full employment (and that’s far from totally obvious), today’s definition of full employment would need to account for the steady educational upgrading and aging of the U.S. workforce since then. If one adjusts the workforce by change in the structure of educational credentials and age-bins since 2000, the unemployment rate consistent with full employment should have fallen by a 0.5% (see actual unemployment versus what it would have been at constant 2000 education and age weights below in Figure 1). This would imply that the 3.7% unemployment rate prevailing before the COVID-19 shock really did not constitute full employment. To push the 3.7% unemployment rate down to 3.5% would require closing the output gap that prevailed at that time by roughly 0.4%.
Unemployment rate, actual and adjusted for age/education changes
|Unemployment, actual||Unemployment, 2000 weights|
Notes: The adjusted unemployment rate is created by calculating 15 age and education cells (5 education categories and 3 age bins), calculating cell-specific unemployment rates, and holding 2000 cell-shares constant over time. Note that a positive output gap implies that the economy is operating above its long-run potential.
Source: Author’s calculations using data from the EPI extracts from the Current Population Survey (CPS).
The current estimate of the output gap by CBO is roughly 3% Their estimate that the pre-COVID economy was operating 1% above potential can be discounted, and the 3.7% unemployment rate at the end of 2019 likely signifies a small output gap of 0.4%. Adding these together, this implies an output gap today of nearly 4.5%.
Fiscal support can’t just cover today’s output gap—it needs to overshoot and wind down slowly
A more reasonable measure of today’s output gap moves it closer to the $1.9 trillion of the Biden package, but the gap is still noticeably smaller. But for making macroeconomic policy decisions, the size of today’s output gap is not the proper measure to target. Instead, one must assess how much fiscal support is needed to restore a high-pressure labor market, and, crucially, how slowly this fiscal support must be wound down to avoid having growth falter again. In our research brief urging policymakers to target $3 trillion in fiscal support between November 2020 and November 2024, we highlighted that a third of this should be planned to support aggregate demand in 2022 and thereafter.
Over the next three years, once one corrects for the too-small measures of potential output used by the CBO, cumulative output gaps would total more than 11% of GDP if no other policy support is given. Imagine a scenario where the output gap in 2021 was entirely closed with front-loaded fiscal support that was immediately spent by households, and then no further fiscal support was forthcoming. This scenario would hence imply a fiscal contraction of more than 4.5% of GDP in the following year. Do we really believe that private sources of demand growth would be strong and self-sustaining enough that this wouldn’t cause a hiccup in growth? In both our research brief and in similar work done by Brookings economists Wendy Edelberg and Louise Sheiner, the winding down of fiscal aid clearly drags on growth in late 2022 and 2023. This fiscal drag is predictable and hence should be avoided.
To avoid it, risk-averse policymakers should plan on weaning the economy off fiscal support slowly, and should not plan on a once-and-for-all drop of fiscal support that is then abandoned. This need for sustained support means that the ultimate price tag of a relief and recovery measures needs to be larger than today’s output gap by itself.
A more convincing criticism of the Biden package is that it provides “too much” support in the near-term while not leaving enough left over for fiscal support in 2022 and 2023. I certainly worry about the state of fiscal support a year from now, but I’m not sure the Biden package ignores this. The most-famous component of it—direct checks—do disburse quite quickly and so are a bit subject to this worry (to be clear, this is judging components only by their contribution to macroeconomic management, not their contribution to alleviating households’ economic distress). But other portions of the package—unemployment insurance and aid to state and local governments and public health measures—will likely see their economic impact be spread over a year or more. In fact, some of the fiscal aid to state and local governments is already being criticized by some observers precisely for not being spent quickly enough. But this is a feature, not a bug of the package.
Finally, even the bits of the package—like the checks—which will be disbursed quickly may not be spent right away. The personal savings rate, for example, spiked when an earlier round of checks were sent out as part of the CARES Act in April. These checks boosted incomes at a time when many possibilities for consumption spending were largely foreclosed by the coronavirus. This didn’t mean that the checks were wasted—as parts of the economy reopened, the savings rate fell quickly and household spending rose.
Additionally, much of the kind of pent-up consumption demand spurred by the virus is in sectors—face-to-face services—that are not likely to see a huge, front-loaded surge of spending once normal economic activity can safely resume. Spending actually rose on big-ticket items like durable goods and autos over the crisis. Once it is safe to again buy face-to-face services, spending will probably fall for these durables but will be consistently strong – but spread out over a long time – for face-to-face services. To put it simply, once the all-clear is lifted, it seems unlikely that people will decide to eat out every single night in a restaurant for six months. But they will consistently and gradually spend noticeably more on restaurants then they did even pre-COVID. This will give a powerful, but long-lasting release of aggregate demand into the economy.
The Biden package is at the right scale of what relief and recovery demands. It is also likely to spread fiscal support over time in a way supportive of recovery. Further, to ensure fiscal support is sustained in the future, a second package of public investments (some financed by progressive revenue) would be most useful (as well as being valuable for lots of reasons besides macroeconomic management).