Labor Market Slack And Inflation by Zach Pandl, ColumbiaManagement
Structural weakness in labor force participation means there is less slack in the labor market than commonly believed. Limited spare capacity implies earlier rate hikes and more risk that inflation eventually overshoots the Federal Reserve’s target.
Congress tasked the Federal Reserve (the Fed) with promoting “maximum employment” but left it to the central bank to decide precisely what that means. In recent decades the Fed’s practical mandate was a goal of “minimum unemployment,” with the minimum defined as the rate consistent with stable inflation. Judging by this standard, the unemployment gap today is less than one percentage point, implying a small amount of remaining labor market slack and only modest downward pressure on inflation (Exhibit 1).
However, the most recent recession and slow recovery have raised questions about whether this definition of full employment remains valid. On the one hand, an unusually large share of the unemployed have been out of work for long periods, and it’s unclear whether the long-term unemployed — who may be more marginally attached to the workforce — exert the same downward pressure on inflation as the short-term unemployed (e.g., Krueger, Cramer and Cho 20141). On the other hand, many observers see cyclical weakness in labor markets beyond unemployment – including workers on part-time schedules who would prefer full-time work and others who have dropped out of the labor force but would prefer to work if the right jobs and wages were available.
Conceptually, the Fed’s working definition of full employment will be related to the risk of overshooting the inflation target. By focusing on narrow measures of labor market slack, Fed officials will run less risk of overshooting on inflation (and some risk of undershooting). By aiming for broader measures of slack, above-target inflation becomes more likely. Exactly how much risk the Fed is taking depends on the underlying state of the economy: in particular, how much of the weakness in broader measures of labor utilization is cyclical vs. structural.
Exhibit 2 visualizes this basic idea. Along the x-axis are possible working definitions for the full employment part of the Fed’s mandate, ranging from narrow to broad. Along the y-axis is average inflation over the next several years. In the optimistic case, spare capacity remains abundant, and virtually all measures of labor underutilization exert downward pressure on inflation. In this world, it makes sense to target the broadest measures of labor market slack, because this best achieves the inflation target and maximizes welfare in the labor market. In the pessimistic case, only the narrowest measure of slack — the short-term unemployment rate — puts downward pressure on inflation. To affect a recovery in the labor market more broadly, the Fed would need to tolerate above-target inflation for a time (e.g., Rudebusch and Williams 20142). The pessimistic example includes a fat right tail on inflation for the possibility that aggressive pursuit of full employment eventually affects inflation expectations.
Unfortunately, our reading of the evidence is that the pessimistic case is closer to reality. In this report we tackle the decline in the labor force participation rate and argue that most of the recent weakness should be considered structural. But we would note that this is just one aspect of the supply-side problem. Low growth in the nation’s capital stock and soft trend productivity growth (e.g., Fernald 20144) also suggest spare capacity may be more limited than commonly believed. As a result, by targeting broader measures of underemployment, the Fed will run a higher risk of above-target inflation.
It’s unclear exactly where Fed officials stand on these issues. In public comments policymakers remain optimistic about the economy’s supply potential, and they seem to have in mind a broader full employment objective than the standard unemployment rate. However, their tolerance for overshooting the inflation target is one of the big uncertainties in the policy outlook. At the very least, some Fed officials think it would be optimal to trade a bit higher inflation for a faster recovery in employment. For instance, Chicago Fed President Charles Evans has said,5 “The surest and quickest way to reach our objectives is to be aggressive. This means, too, that we must be willing to overshoot our targets in a manageable fashion.”
While overshooting may be optimal in an economic model, it is certainly not the best case scenario for fixed-income markets. We think investors should remain cautious about U.S. interest rate risk despite strong gains in high-quality fixed income in the first half of 2014. In addition, limited spare capacity means that inflation risks have become more balanced. At current valuations, inflation-protected government bonds are more attractive than their nominal counterparts.
Labor force participation to remain low
The long-running debate around U.S. labor force participation is now very familiar to investors. Since the economy bottomed in mid-2009, real GDP has increased at an annualized rate of just 2.1%. If GDP growth could have been known in advance, standard macroeconomic models would have predicted a flat or rising unemployment rate over the last five years. Instead, after peaking at 10% in October 2009, the unemployment rate has declined to just 6.1% today. The primary reason for the steady decline has been falling labor force participation — only a portion of which can be explained by the aging of the workforce (Exhibit 3). The remaining “participation gap” has been a puzzle for economists and investors.
There are two schools of thought on how to interpret this participation gap. One camp argues that it represents cyclical weakness and should be added to the unemployment gap when estimating labor market slack (e.g., Erceg and Levin 20136). The alternative argument is that most of the decline should be interpreted as structural, because workers left the labor force for reasons unrelated to the recession and/or because cyclical weakness turned structural over time.7 In our view, the available evidence strongly supports the latter conclusion.
First, persistent weakness in labor force participation is a standard feature of recoveries after financial crises and major recessions. Exhibit 4 shows changes in labor force participation following 12 major financial crises (defined as deviations from trend in the 10 years prior to crisis). The first few are the so-called OCED Big Five financial crises. The remaining seven are from the 2008 downturn. On average, five years after the crisis year, participation rates are around 4% points lower than the precrisis trend. In the United States today, the labor force participation rate is about 3% points lower than the level in 2007. In other countries with financial crises, participation rates took a very long time to stabilize. In the Big Five sample, they remained below precrisis trends for more than 10 years after the crisis year (consistent with hysteresis effects). Academic research has found similar patterns in participation rates after other large recessions (Duval, Eris and Furceri, 20108).
Second, the composition of the recent decline in labor force participation suggests a quick rebound is unlikely. Analysis by Federal Reserve Bank of Philadelphia demonstrated that 72% of the decline in participation since 2007 is accounted for by retirements and new disability recipients (Fujita 20149). Only a small portion of the decline (12%) represents persons who say they currently “want a job” (Exhibit 5). The distinction is important because reentry rates — the odds of coming back to the labor force from nonparticipation — are extremely low for both retirees and those on disability. Once these workers leave the labor force they are very unlikely to return (e.g., Livermore and Stapleton 201010). The increase in disability recipients is interesting because it points to a possible mechanism causing persistent weakness in participation — namely the interaction of joblessness and social insurance programs.
Third, although the decline in the participation rate has been large, some research says that it can be fully explained by long-term structural factors. In particular, a well-known paper from Federal Reserve Board economists argued in 2006 that demographics and other structural factors — such as an increased tendency for young people to remain in school and delay entering the labor force — meant the participation rate would fall to about 63% at the end of 2013 and slightly below that in 2014–2015 (Exhibits 6 and 7). The predictions of this model have been remarkably accurate. While one research paper is hardly the final word on the matter, it underscores the many long-term negatives for U.S. labor supply that were in place before the recession.
Fourth, recent changes in the unemployment rate look broadly consistent with other measures of slack in the economy (i.e., other observable proxies for the output gap). Exhibit 8 shows normalized scores for eight measures of slack: (1) the unemployment rate, (2) the industrial capacity utilization rate, (3) the labor differential from the Consumer Confidence Survey, (4) the business conditions index from the Consumer Confidence Survey, (5) the quality of labor index from the NFIB Small Business Survey, (6) the poor sales index from the NFIB Small Business Survey, (7) the open positions index from the NFIB Small Business Survey and (8) a measure of the employment-to population ratio for the 24–54 age group. The key point is that the unemployment rate is nearly indistinguishable in the chart — we are seeing a similar cyclical recovery in virtually all measures of spare capacity. This again hints that weakness in labor force participation is likely structural; if the broader employment-to-population ratio were an appropriate proxy for the output gap, we would expect to see it validated in spare capacity measures from other sources (e.g., household surveys). Instead, these measures of economic slack validate the message from falling unemployment.
Lastly, we would note that the U.S. participation rate remains well-within normal ranges for developed market economies — indeed, it is still slightly above average (Exhibit 9). Thus, while the participation rate declined substantially in recent years, it is still not obviously low in an absolute sense (though of course the drivers of participation rates across countries are varied).
Thus, available evidence is unfortunately pessimistic about the prospects of a rebound in labor force participation. Contrary to conventional wisdom, the traditional unemployment rate appears to be giving a broadly accurate signal about spare capacity in the U.S. economy.
Remain cautious on rates despite solid first half
As far as we can tell, most Fed officials do not share our views on labor market slack. For instance, the minutes of the March Federal Open Market Committee (FOMC) meeting noted that “several participants” stressed there might be “considerably more labor market slack than indicated by the unemployment rate alone.” At the same time, only “a couple” of officials said that the unemployment rate remains a sufficient summary indicator of slack in the labor market.
Interestingly, a recent research paper11 by economist Andrew Levin (a former advisor to Janet Yellen) hints that the Fed’s estimates of broad labor slack may be substantial. Levin puts the total shortfall at 5.5 million full-time equivalent workers, and he implies that current labor market conditions are worse than only 10%–15% of observations in postwar history (Exhibit 9; these are our estimates using Levin’s methodology with small modifications). Although FOMC members frequently talk about cyclical weakness in labor force participation and slack in hours worked, they rarely provide quantitative estimates like this. The Levin slack estimates are large, and if these views were shared by Fed officials, it would mean their opinions on labor market slack are indeed very far from our own.
As discussed in the introduction, there’s a connection between (1) the Fed’s full employment objective, (2) the risk of overshooting the inflation target and (3) the degree to which changes in the labor market are cyclical versus structural. By focusing on broad measures of underutilization, policymakers are signaling optimism about the economy’s productive potential and/or a high tolerance for overshooting. We think the former is inconsistent with the data and the latter is inconsistent with the normal conservative tendencies of central banks (e.g., Ball 200912). Perhaps they are simply hoping for the best.
We think Fed officials will ultimately change their minds on the degree of labor market slack in light of compelling evidence and then fall back to a focus on the unemployment rate. Indeed, we may have seen early signs of this already, with the committee’s funds rate forecasts beginning to react to the falling unemployment rate this year (Exhibit 11). Any hints of a faster recovery in wage or price inflation would likely accelerate this process. Although Chicago Fed President Evans and others have advocated a willingness to overshoot, we wonder if their message will change when above-target inflation becomes a clear and present danger.
Regardless of how the Fed’s views evolve, this is not a terribly attractive environment for high-quality fixed income — overshooting is not a term bond investors generally like. Despite healthy returns to interest-sensitive assets in the first half of 2014, we remain cautious on the outlook for the balance of this year. Limited spare capacity means one of two outcomes is likely: (1) earlier-than-expected rate hikes or (2) higher than-expected inflation. The distinction matters for the shape of the yield curve, but both outcomes are unfriendly for Treasury duration.
Separately, limited spare capacity means that inflation risks have become more balanced. We are not concerned about the size of the Fed’s balance sheet and its ability to tighten policy when the time comes. Rather, traditional cyclical upswings in wage and price inflation could begin well before the Fed is expected to normalize policy. Thus, at current valuations we believe inflation-protected government bonds are more attractive than their nominal counterparts.