Record US productivity slump in first half of 2022 risks higher inflation and unemployment

Over the last two quarters, US employment in the nonfarm business sector grew at a very strong 4.3 percent annual rate while output fell at a 2.3 percent annual rate. With average hours per worker decreasing, this means that output per hour, also called labor productivity or simply productivity, fell at a 6.0 percent annual rate. We also calculate an “optimistic” estimate of productivity growth assuming output rose in the first half of the year, but even this optimistic estimate fell at a 4.0 percent annual rate.

Although productivity growth is very volatile and poorly measured at high frequencies, these declines are larger than the largest two-quarter declines since the data began to be collected in 1947—with the official estimate nearly twice as large as the largest previous decline. The exact magnitude of this decline may reflect measurement error, but the broad story—that output growth is slow or even negative while job growth remains very strong—is clear and robust across a wide range of data.

These data suggest the economy will not sustain higher compensation growth without passing it into price growth, perhaps even higher price growth than would usually be associated with this pace of compensation growth. Based on a range of measures, our best guess is that the underlying growth of wages is consistent with personal consumption expenditure inflation, excluding food and energy (core PCE inflation), of 4.5 percent or perhaps higher. Labor costs appear to have risen more than price growth, suggesting that, if anything, additional price increases are more likely than additional wage increases.

Moreover, continued strong increases in employment while output is weak or falling are unsustainable. This raises the fear that businesses will downshift their hiring in the coming year unless there is a large increase in output growth.

Recent Productivity Data

Our analysis centers around productivity, specifically output per hour in the nonfarm business sector. Measuring productivity in the recent period is complicated by unusually large discrepancies in the data about just how much the economy has grown. Our analysis shows two measures of productivity and other statistics derived from the productivity data: (1) the official data published by the Bureau of Labor Statistics (BLS) and (2) an “optimistic” measure that is based on our adjustment using an alternative measure of the size of the economy, gross domestic income. Most of the conclusions are robust to the alternative measure of productivity but some are not.

The productivity data are based on output in the nonfarm business sector, which the BLS derives from the Bureau of Economic Analysis’ gross domestic product (GDP) data, which are based on adding up final expenditures by households, businesses, and government. Recently, however, an alternative measure of the size of the economy—gross domestic income (GDI)—has been running much higher than GDP. In the first quarter, for example, GDP shrank at a 1.6 percent annual rate, while GDI grew at a 1.8 percent annual rate (the government’s first estimate for GDI for the second quarter will not be published until the end of August). Overall, the gap between GDP and GDI, shown in figure 1, is the largest it has ever been. In theory, the two measures should be identical, and they differ only because of measurement error. Evidence suggests that, in general, the average of GDP and GDI produces a better measure of growth than either measure individually—and, in fact, the Philadelphia Fed’s GDP plus model suggests the recent truth is much closer to the GDI numbers.

Our “optimistic” case adjusts the BLS estimate of real value-added output in the nonfarm business sector by the ratio of real GDI to real GDP. For the second quarter of 2022, we assume that real GDI growth was zero, which is more optimistic than the -0.9 percent decline in real GDP, in part reflecting some anomalies in the GDP data, which suggest GDP might have understated economic growth again in the second quarter.

The resulting two productivity series are shown in figure 2, both normalized so that the level of productivity was 100 in 2019Q4, before the onset of the pandemic. By both measures productivity fell over the last two quarters. The official BLS measure fell at a 6.0 percent annual rate, nearly twice the previous record for the worst two quarters since the data started in 1947. The optimistic measure fell at a 4.0 percent annual rate, which would also be a record two-quarter decline.

The magnitude of the decline in productivity over the last two quarters is likely more uncertain than even the range between the official BLS series and the optimistic case. The fact that productivity declined—or grew at a very slow rate—is robust. Both GDP and nonfarm business output fell in the first half of the year. Other measures, like personal consumption expenditures and investment, were at best growing below their normal pace. GDI, while still positive in the first quarter of 2022, was slightly below normal as well. In contrast, employment increased well above the normal pace in the last two quarters, including nonfarm business employment (+4.3 percent annual rate) and private payroll employment (+4.6 percent annual rate)—more than twice the pace of pre-pandemic employment growth. Falling or slow growth of output combined with rapid growth of employment is consistent with low or negative productivity growth.

Productivity initially rose well above trend in 2021 as output recovered more rapidly than employment and short-handed businesses ended up with workers working much more intensively than usual. It has since fallen. Over the entire pandemic period it appears that productivity growth has been slower than it was in the previous business cycle, although this is less certain. The official BLS series for productivity rose at a 0.6 percent annual rate since the fourth quarter of 2019, well below the 1.3 percent annual rate in the peak-to-peak of the previous business cycle or the 1.7 percent annual rate in the two years leading up to the pandemic. The optimistic case shows productivity above the pace of the last business cycle and even above the pace immediately preceding the pandemic.

Interpreting pandemic productivity

Over longer periods of time, measured productivity is a good estimate of the progress of innovation, capital deepening, and management practices, what we call “genuine productivity.” For example, if, over a decade, restaurants adopt more online reservation systems and point-of-sale systems for ordering, they will be able to produce more for each hour of work. Over shorter periods of time, however, the independent dynamics of employment and output, what we call “residual productivity,” can matter a lot. For example, if the weather is unusually bad, people may not go to restaurants. The same number of restaurant staff will serve fewer patrons, producing less output per hour. Conversely, if a restaurant gets a good review and patrons flock to it, its output per hour will go up. This “residual productivity” does not reflect a change in the economy’s ability to produce, so changes will generally be transitory. Finally, there are also “compositional productivity” changes—for example, if a low-productivity sector in the economy shuts down, then overall productivity will rise—even if no individual business actually became more productive.

Most of the changes in productivity in the pandemic period have been driven by residual and compositional productivity, as emphasized in a recent paper by Robert Gordon and Hassan Sayed. For example, output per hour rose at a 10.3 percent annual rate in the second quarter of 2020. This was not a genuine change in the underlying productivity of the economy but a combination of residual productivity, as businesses were temporarily able to maintain operations with a much smaller workforce, and compositional productivity, as output was disproportionately reduced in lower-productivity sectors like leisure and hospitality.

Demand returned rapidly in 2021 as the economy was supported by enormous fiscal and monetary stimulus. Employment also grew, but the pace of job growth was restrained by the reluctance or inability of many workers to return to jobs, a shortage of immigrants, premature deaths due to COVID, and the time-consuming process of matching workers to jobs. As a result, people in jobs temporarily had to work with greater intensity, temporarily raising productivity. For example, this could be a restaurant where the customers have returned but staff levels remain low, so the existing staff ends up working especially intensively—raising measured productivity well above its normal levels.

In 2022, it appears that the opposite has been happening. Demand has fallen, or at least weakened, abruptly. Businesses have continued to add jobs, perhaps due to a combination of residual nervousness about the difficulty of getting employees amidst severe labor shortages and the belief that the output weakness is temporary, driven by factors like inventory cycles, not underlying demand. In addition, it is possible that employees who were temporarily willing to work with additional intensity earlier in the pandemic period have become less willing to do so and employers are compensating by getting back to more normal levels of staffing. Whatever the exact story, the result is the same: the decline in output per hour the economy has witnessed over the last two quarters.

These short-run gyrations in residual and compositional productivity make it harder to assess what has happened to genuine productivity. It appears to be below trend, which would not be surprising given that during 2020 and 2021 businesses invested much less than usual and most of their innovation efforts were focused on managing the pandemic, not improving their longer-term operations. In addition, workers suffering from long COVID and increased absences from the workplace due to illness may affect the productivity of other employees engaged in joint work. Supply chain issues have also likely affected productivity as businesses have workers who are not working while waiting for the inputs they need.

Finally, there is the open question of whether remote or hybrid work has increased or reduced productivity. Some research has found it has increased productivity, for example, because it provides more flexibility for employees and enables employers to focus on better productivity metrics than face time in the office. Others, including many businesses, have raised concerns that it has lowered productivity, either by reducing the creative interactions that happen in a face-to-face environment or by enabling employees to reduce their work intensity.

What This Means for Inflation

Over the last six months, nominal compensation or wages have been running about 2 to 3 percentage points above their pace in the run-up to the pandemic. Wages are one of the main costs for businesses. And just like airlines raise prices when input costs like jet fuel rise or restaurants raise prices when the cost of chicken increases, businesses raise prices when their wage costs increase. In fact, the link between nominal wage growth and price growth has generally been very strong as shown in figure 3.

Two factors can break the link between nominal wage growth and price growth. The first is productivity. If people are paid 3.5 percent more but they can produce 1.5 percent more per hour, then unit labor costs go up only 2.0 percent and the business only needs to raise prices at this pace to maintain constant profits. This is why the price growth shown in figure 3 was generally below compensation growth. The second is profits themselves. In this same example, if businesses raise prices only by 1.5 percent, they will be absorbing some of the wage costs in lower profits. Conversely, profits will increase if they raise prices by more than 2.0 percent.

Assuming underlying productivity and profit trends are the same as they were from 2000 through 2019, table 1 shows the core PCE inflation rates (the Federal Reserve’s de facto target) that would be consistent with different measures of the recent pace of compensation or wage growth. The Employment Cost Index (ECI) is the best measure of these and it is suggesting that wage growth is consistent with 4.5 to 5.2 percent annualized inflation. The other measures are broadly consistent with the ECI data, ranging from 3.8 to 6.0 percent over the past year. (Note the differences in the pace of wages and compensation growth across the various measures are because they are different concepts with different historical relationships to price growth.)

Table 1 assumes that wage increases will translate directly into price increases in the same manner they did in the decades leading up to the pandemic. Some analysts have argued that because real wages have been falling recently, it should be possible to have a period of higher nominal wage growth without a commensurate increase in inflation. While this is a possibility, the productivity data suggest that the opposite is more likely to be the case. As shown in figure 4, unit labor costs—the increase in compensation above productivity—have risen faster than prices over the pandemic period. Overall unit labor costs have risen at a 6.4 percent annual rate (or 4.9 percent using the optimistic series based on GDI), which exceeds 4.3 percent annual increase in prices over that period.

Why is this so different from what many analysts have argued? In part those analysts are mixing and matching different series for wages and prices that are measured in different ways and cover different segments of the economy. The price deflator for the nonfarm business sector—the relevant measure for businesses in this context—increased at a 4.3 percent annual rate since the fourth quarter of 2019, less than the 5.1 percent annual growth of the consumer price index but more in line with the 4.1 percent annual growth of the PCE price index.

In part, those analysts are neglecting to factor in the entire post-pandemic period. Prices grew very slowly in 2020 while labor costs rose well above them. Although the gap may have narrowed over the last year and a half (it is not clear from the data), even if there was a narrowing, it was not enough to undo the earlier disconnect. Overall, to the degree that productivity has been weak over the pandemic period as a whole, that greatly increases price pressures for a given increase in compensation.

Popular beliefs about the rise in profits and the decline in the labor share are much less robust than many people think. The BLS data itself show that the labor share has risen by 2.8 percentage points since the onset of the pandemic (or 0.8 percentage point in the optimistic case where output is higher) as shown in figure 5 (note other measures of the labor share, like the share of compensation in national income, have also risen but by less than the official BLS series).

In longer historical perspective, the labor share is still low, and it is possible and desirable that it will continue to rise—enabling compensation increases that do not fully translate into price increases. But it is more likely that the labor share stays where it is now or even falls back to where it was prior to the pandemic, in which case the pace of price growth would be even higher than shown in table 1.

Finally, the underlying trend of productivity also matters for inflation. If the recent performance of productivity is a harbinger of a slower pace of productivity growth going forward—something that is highly uncertain—that would further boost inflation rates relative to those reported in table 1 unless there is a significant slowdown in compensation growth.

Outlook for Productivity and the Economy

Looking forward, the evolution of productivity, both genuine and residual, will be one of the most important factors in the economic recovery. Productivity will not continue to decline like it did in the first half of 2022. One possibility is that the numerator, output, starts to rise more quickly—perhaps because the economy’s weaknesses were all in volatile components like inventories and net exports. Another possibility is that the denominator, employment, worsens, perhaps as employers rapidly shed labor that they determine they no longer need in an economy with much lower output.

The most important questions are about the long run. Will businesses be able to deploy pandemic-era innovations, including work from home, to operate at a sustained higher level or growth rate of productivity? Will other advances, like artificial intelligence, finally start to sustainably boost productivity growth? Or will the economy return to the relatively weak productivity growth of the pre-pandemic period or perhaps even something worse? The answer to this question will determine how fast the economy grows and how quickly living standards rise.