New Keynesians Explain Cyclical Unemployment

Print/Save PDF

Last week’s post on unemployment provides context for today’s essay. In it, we return to a fundamental message of the GEM Project: The absence of microfounded meaningful wage rigidity deprives NK model-builders of crucial guidance in their attempts to be stabilization-relevant. It is a debilitating problem. Absent rational MWR many mainstream theorists are lost at sea, especially when they tackle real-world problems. It is a sad lesson of modern macroeconomics that usefulness  plays second-fiddle to fealty to the stubborn conviction that market-centric general equilibrium is settled theory.

Today’s example of macro modeling absent practicality is Jan Eeckhout and Ilse Lindenlaub’s (hereafter E&L) “Unemployment Cycles”, recently published in the American Economic Journal: Macroeconomics. The authors describe their paper: “The labor market by itself can create cyclical outcomes, even in the absence of exogenous shocks. We propose a theory in which the search behavior of the employed has profound aggregate implications for the unemployed. There is a strategic complementarity between active on-the-job search and vacancy posting by firms, which leads to multiple equilibria: in the presence of sorting, active on-the-job search improves the quality of the pool of searchers. This encourages vacancy posting, which in turn makes costly on-the-job search more attractive—a self-fulfilling equilibrium.”

The E&L innovation to labor-market search theory is allowing employed workers to join the unemployed in their search for satisfactory jobs. “Central to the model’s mechanism is the fact that the employed crowd out the unemployed when on-the-job search picks up during recovery.” In their model, unemployment cyclicality requires neither exogenous shocks nor propagation by nominal demand disturbances.

E&L Objectives

Cyclical unemployment. As indicated by their title, E&L’s objective is to explain the cyclical behavior of unemployment. The exercise is almost wholly self-referential, allowing them to simply ignore the most important evidence. “Business cycles have a wide variety of origins, ranging from financial crises, over oil price shocks, to productivity spurts and slowdowns. Often, of all economic agents, workers are those affected most dramatically, mainly through unemployment. For long, researchers—most notably Diamond (1982)—have asked whether frictional markets can generate cyclical outcomes, even in the absence of any exogenous shocks or changes in fundamentals. But so far, there has been no compelling mechanism where the labor market by itself can generate cycles and that fits the facts.” As will be shown, the promise to fit the facts must be some sort of sick inside joke.

The unspoken reason for the longstanding NK reliance on labor-market search to explain cyclical movement in unemployment is the mainstream need for an explanation of labor cyclicality that does not rely on meaningful wage rigidity. Delivering unemployment analysis from its dependence on MWR, which mainstream theorists have given up trying to microfound, has long been high on the EK research agenda. The huge, unproductive investment in labor-market search/match theory, illustrated by the content of the  AEJ:M since its founding in 2009, has of course been the result of NK editorial choices.

The GEM Project has microfounded MWR and, unlike E&L, does not have to ignore the elephant in the room. Recall E&L’s central claim: “Our model provides a simple rationale for large fluctuations in unemployment.” The elephant is the most crucial cyclical-unemployment fact, which is never even mentioned in E&L’s 60-page article: the overwhelming importance of involuntary job loss (IJL) in actual, especially large, fluctuations of unemployment. E&L must ignore IJL because its rational existence requires microfounded MWR, the search for which has been erased from the NK research agenda.

                         JOB-LOSS BEHAVIOR IN U.S. RECESSIONS

Peak-to-Trough Change in:
Unemployment Rate Job-Losers Incidence Job Losers (000)
1969-70 +2.4 points +8.2 points +1,230
1973-75 +3.8 points +16.0 points +2,599
1980 +1.5 points +7.4 points +1,315
1981-82 +3.6 points +11.2 points +3,433
1990-91 +1.3 points +6.8 points +1,373
2001 +1.2 points +6.0 points +1,423
2007-09 +4.8 points +13.1 points +5,807


The data are from the Current Population Survey, Bureau of Labor Statistics. Involuntary-job-loser data are seasonally adjusted and available only from 1967. Job-loser incidence is the ratio of job losers to total unemployed.

Adequate macroeconomists already know that forced job loss is always the largest, by far, component of rising unemployment in recessions. For example, in the perilous 2007-09 Great Recession, the nearly six-million increase in involuntarily lost jobs accounted for more than three-quarters of the increase in total unemployment, pushing the job-loss incidence up by 13 percentage points. Adequate macroeconomists know that, if you lack an explanation of involuntary job separation, you cannot hope to explain the cyclical behavior of unemployment. Moreover, if you lack microfounded MWR, which the GEM Project has shown uniquely enables the existence of forced layoffs, you cannot construct a micro-coherent stabilization-relevant macro model.

Lacking rational MWR, E&L focus wholly on familiar market-centric labor-search theory that can only produce voluntary (frictional) unemployment. Since quits are pro-cyclical, frictional joblessness interestingly has a small pro-cyclical tilt. But that is very small potatoes compared to forced layoffs. The evidence is incontrovertible. How can E&L not know their claim that their voluntary joblessness model “provides a simple rationale for large fluctuations in unemployment” is an obvious lie. If their analysis is any guide, E&L are thoroughly ignorant of the actual cyclical behavior of unemployment.

Other promises. This post has insufficient remaining space to look closely at the other, lesser E&L promises. Explaining jobless recoveries? E&L unconventionally define a jobless recovery as “the fact that it takes a long time for unemployment to drop even after vacancies and productivity have recovered”, an interesting characteristic of post-recession labor behavior more typically named “unemployment persistence”. Their explanation is again loosely motivated by already employed workers job crowding-out the unemployed in the contest for newly created jobs. (E&L overlook the need to identify who fills the jobs that the successful employed applicants previously held.) In the GEM Project, unemployment persistence is largely rooted in the rational response to the rational payment of wage rents on jobs that the newly unemployed have lost. After failing to be recalled, a significant period of price discovery is needed for job seekers to disentangle those rents from their true market-opportunity costs. (For elaboration, see Chapter 6 in the website’s e-book.)  Procyclical vacancy-rate behavior? In the generalized-exchange theory, job vacancies posted by firm rationally shrink when those firms are laying employees off. Wage dispersion? Generalized-exchange theory, with its rational time-varying labor rents paid by highly specialized bureaucratic firms, robustly explains the extraordinarily consequential behavior of the inter-industry wage structure over the postwar period. The E&L model has no capacity to model such behavior.

Two Summary Issues

First, one of the many drawbacks of macro modeling absent microfounded MWR – an unhappy task undertaken by E&L and other NK theorists – is that  the role of nominal demand disturbances, especially in the propagation of macro shocks, is inherently deemphasized. It is no secret that this outcome, if taken seriously, is dangerous. The absence of a powerful role for nominal demand is the major reason why stabilization authorities disdainfully rejected mainstream NK thinking in their successful efforts to avoid a depression in 2008-09.

Second, application of Samuelson’s  revealed preference helps us  understand the true import of the E&L article. Actual behavior of the AJE:M editors allows us to infer that they believe the publication of “Unemployment Cycles”, despite ignoring involuntary job loss that accounts for most of the rise in unemployment during recession,  better helps macroeconomists understand  the actual behavior of unemployment than would publication of the generalized-exchange microfounding of MWR. How can the implicit AJE:M  ban on nonmarket research microfounding the suppression of wage recontracting, saving space for market-centric research that is profoundly out-of-touch with actual behavior, be justified?

Blog Type: New Keynesians San Miguel de Allende, Mexico

Write a Comment

Your email address will not be published.