Understanding Unemployment — Chapter-by-Chapter Outline

Author: Lawrence H. Summers (with contributions by Kim B. Clark, James M. Poterba, N. Gregory Mankiw, Julio J. Rotemberg, and Olivier J. Blanchard) First published: 1990 (MIT Press, hardcover May 1990; paperback edition 1992) Edition covered: First edition (MIT Press, 1990; ISBN 0262192659). The 1992 paperback (ISBN 0262691574) reproduces the same content without additions. This volume collects eleven previously published papers by Summers and co-authors — most originally appearing in Brookings Papers on Economic Activity, the American Economic Review, the Journal of Labor Economics, the Review of Economic Studies, and NBER volumes — together with a new Introduction and Conclusion written specifically for the book.

Central thesis

Unemployment is not a residual statistical artifact or a temporary friction that markets automatically self-correct. It is a pervasive, measurable, and costly labor market condition whose causes — labor market turnover patterns, measurement distortions, supply-and-demand factors, wage rigidity, efficiency wages, and institutional forces including unions — can be empirically identified and theoretically explained. The standard economic models of the early 1980s, which emphasized brief unemployment spells and rapid market clearing, were empirically inadequate; the actual distribution of unemployment is heavily concentrated among workers experiencing long spells, and the actual level of unemployment is systematically understated by official measures.

The book assembles Summers' research program — carried out with Clark, Poterba, Mankiw, Rotemberg, and Blanchard over roughly a decade — to show that understanding unemployment requires both better measurement and richer theory. On the measurement side, the dominant picture of unemployment as a pool of short-spell job-seekers is misleading: a small fraction of workers account for the bulk of unemployment weeks. On the theory side, labor markets are not continuously clearing auctions; wages are rigid downward, insider workers defend their own employment above market-clearing wages, efficiency wage considerations give firms an incentive to pay above the market, and unemployment insurance affects workers' incentives in ways that alter both duration and measured flows.

Why do labor markets fail to clear, and why does unemployment persist even when jobs appear to be available at the going wage?

Chapter 1 — Introduction: The Facts Any Theory Must Confront

Central question

What are the empirical regularities about unemployment — its level, its distribution, its cyclicality, and its costs — that any serious theoretical account must accommodate?

Main argument

The concentration of unemployment among long-spell workers

The Introduction presents Summers' central empirical finding, developed with Clark across several papers: the popular image of unemployment as a pool of many workers each experiencing brief spells is wrong. At any given time, a disproportionate share of total unemployment weeks is accounted for by a relatively small group experiencing very long spells. This is shown by decomposing the stock of unemployment into flows of workers entering and exiting: while most spells are short, short-spell workers contribute little to the total stock. The finding has immediate theoretical implications — models built around brief frictional unemployment miss the essential phenomenon.

The measurement problem

Official unemployment statistics, based on Current Population Survey (CPS) monthly interviews, have two systematic biases. First, workers who leave the labor force entirely are not counted as unemployed even if they would take a job at the going wage. Second, reporting errors in month-to-month labor force status transitions create spurious flows that overstate gross labor market dynamism while potentially misclassifying workers between employment, unemployment, and non-participation.

The costs of unemployment

The Introduction sets out why unemployment matters: not merely the direct income loss to affected workers, but the long-term scarring effects on wages and career trajectories, the concentration of unemployment among vulnerable demographic groups (youth, minorities, low-skill workers), and the macroeconomic loss of output.

What theories must explain

The book's organizing agenda is stated explicitly: any adequate theory of unemployment must explain (1) why unemployment is concentrated in long spells; (2) why wages do not fall to clear labor markets; (3) the cyclical behavior of unemployment across demographic groups; (4) the role of unemployment insurance in extending durations; and (5) the persistence — hysteresis — of unemployment following aggregate demand shocks.

Key ideas

• The stock of unemployment at any moment is dominated by long-spell workers, not brief-spell job-changers.
• Official unemployment rates undercount true labor market distress by excluding discouraged workers and non-participants who are effectively involuntarily out of work.
• Unemployment is distributed unequally by age, race, sex, and education in ways that amplify welfare concerns beyond aggregate rates.
• Measurement distortions in the CPS — particularly classification errors — make gross flow statistics unreliable without correction.
• Wage rigidity, not merely friction, appears necessary to explain persistent unemployment in the face of apparent labor market slack.

Key takeaway

The empirical record of unemployment is more severe, more concentrated, and more persistent than the standard frictionalist model acknowledges, and explaining it requires both better data and new theory.

Chapter 2 — Labor Market Dynamics and Unemployment: A Reconsideration

(Originally: Kim B. Clark and Lawrence H. Summers, Brookings Papers on Economic Activity 1, 1979, pp. 13–60)

Central question

Is unemployment primarily a pool of many workers experiencing brief, frequent spells (the "turnover" view), or is it dominated by a smaller group with long, infrequent spells (the "long-term" view) — and what are the implications for welfare and theory?

Main argument

Decomposing the unemployment pool

Clark and Summers use longitudinal CPS data to decompose total unemployment weeks into contributions from workers with different spell lengths. The calculation is conceptually simple but decisive: even if the median unemployment spell is short (say, 5–6 weeks), workers experiencing spells of 26 weeks or more may account for the majority of the total unemployment stock at any moment, because their spells accumulate in the cross-section. This is an application of the length-biased sampling insight: a survey of currently unemployed workers will oversample those with long spells.

The turnover view challenged

The dominant view in the late 1970s, associated with Feldstein and Hall, emphasized high labor market turnover and brief unemployment spells, concluding that most unemployment was frictional and voluntary — the inevitable consequence of a dynamic matching process. Clark and Summers show this view is misleading. The relevant statistic for welfare analysis is not the frequency of unemployment spells but the fraction of total unemployment weeks contributed by different spell-length groups. By this measure, long-term unemployment is the heart of the unemployment problem.

Labor force withdrawal as disguised unemployment

A further finding complicates the picture: many unemployment spells officially "end" not in employment but in labor force withdrawal. Workers coded as leaving the labor force in one month frequently return the next month as unemployed. This cycling between unemployment and non-participation means that standard spell-duration measures understate the true duration of non-employment for many workers, especially youth.

Implications for theories of unemployment

If unemployment is concentrated in long spells, then theories based on brief search-and-matching processes are insufficient. The data point toward models in which some workers face persistent barriers — low wages relative to reservation wages, structural mismatches, or demand deficiencies — rather than merely optimal job search.

Key ideas

• The stock-flow distinction is fundamental: the stock of unemployed workers is dominated by those with long spells even if most new entrants experience short spells.
• Length-biased sampling means that surveys of currently unemployed workers overrepresent long-spell workers.
• Transitions from unemployment to non-participation are frequent and often reversible, blurring the line between "unemployed" and "out of the labor force."
• The welfare costs of unemployment are concentrated among long-spell workers who experience both income loss and labor market deterioration.
• The turnover/frictional view of unemployment systematically understates the severity of the unemployment problem.

Key takeaway

Most of the social burden of unemployment is borne by a minority of workers experiencing persistent joblessness, not by the majority who experience brief, transient spells — a finding that fundamentally reframes the policy question.

Chapter 3 — Labor Force Transitions and Unemployment

(Originally: Kim B. Clark and Lawrence H. Summers, NBER Working Paper No. 277, 1978; later published in Brookings Papers on Economic Activity)

Central question

How do workers actually move among the states of employment, unemployment, and labor force non-participation, and what do those transition patterns reveal about the nature of unemployment?

Main argument

The three-state model of labor market status

Standard analyses treat the labor market as two-state (employed vs. unemployed), but a three-state model that includes non-participation is essential. Clark and Summers document the transition rates among all three states using CPS month-to-month matched data. The transition matrix reveals that the E→U (employment to unemployment), U→E, U→N (unemployment to non-participation), and N→U transitions are all substantial and economically significant.

Non-participation and disguised unemployment

A central finding is that many workers recorded as leaving the labor force (N) are in practice marginally attached to the labor market. Their transition probabilities back into unemployment or employment in subsequent months are much higher than those of genuinely retired or non-participating workers. This "hidden unemployment" in the non-participation category means official unemployment rates systematically understate labor market distress.

Demographic heterogeneity in transition rates

Transition rates vary sharply by age, sex, race, and education. Youth have extraordinarily high U→N and N→U transition rates, consistent with the finding in Chapter 2 that many youth unemployment spells end not in employment but in labor force withdrawal. Adult prime-age workers have lower transition rates but longer spells conditional on being unemployed. The demographic heterogeneity means that aggregate unemployment rates mask fundamentally different labor market experiences across groups.

The instability of youth employment

For teenagers and young adults, the dominant labor market dynamic is high turnover across all three states rather than sustained unemployment. But this turnover is not costless frictionalist matching — a shortage of attractive jobs forces many young workers into cyclical non-participation rather than productive search.

Key ideas

• The labor market's true state space is three-dimensional (E, U, N), and collapsing it to two states misclassifies workers and misstates transition patterns.
• Non-participation is heterogeneous: some N-workers are genuinely out of the market; others are marginally attached and would work at a slightly higher wage or lower search cost.
• Youth unemployment statistics understate youth non-employment because the N-state absorbs many frustrated job-seekers.
• Transition rates are strongly procyclical — in recessions, both U→N and N→U transitions rise, indicating greater labor market distress among the marginally attached.

Key takeaway

The three-state transition framework reveals that the official unemployment rate is an incomplete guide to labor market health, because it ignores the large pool of marginally attached workers cycling in and out of non-participation.

Chapter 4 — The Dynamics of Youth Unemployment

(Originally: Kim B. Clark and Lawrence H. Summers, in R. Freeman and D. Wise, eds., The Youth Labor Market Problem: Its Nature, Causes, and Consequences, University of Chicago Press, 1982)

Central question

Why is youth unemployment so high, and is it primarily a problem of frequent short spells (high turnover) or persistent long-term joblessness among a subset of youth?

Main argument

The distribution of youth unemployment

Applying the same spell-decomposition method from Chapter 2 specifically to teenagers and young adults, Clark and Summers find that youth unemployment, like adult unemployment, is dominated in welfare terms by a concentrated group with long spells. A minority of youth account for the majority of total youth unemployment weeks.

The role of labor force withdrawal

A distinctive feature of youth unemployment is the high rate at which spells end in labor force withdrawal rather than employment. Over half of youth unemployment spells in the CPS data end in the youth leaving the labor force entirely. This has two implications: (1) measured youth unemployment spells are shorter than the underlying non-employment spells, because the N-state hides continued joblessness; (2) the distinction between "unemployed" and "out of the labor force" is especially blurry for youth.

Shortage of attractive jobs as the root cause

Clark and Summers argue that the evidence points to a shortage of attractive job opportunities — jobs offering adequate wages, stability, and advancement — as the principal source of long-term youth non-employment. Youth are not primarily failing to search or refusing jobs; rather, the available jobs at entry-level wages fail to clear the labor market in a way that fully employs youth who are willing to work at those wages. This is inconsistent with a pure frictional view and points toward structural demand deficiency or wage floors.

Policy implications

If youth unemployment is primarily a problem of insufficient job quality and quantity rather than search friction, the appropriate policy response involves job creation and wage subsidy programs rather than search-assistance programs alone. The chapter thus has direct policy implications that extend beyond the positive analysis.

Key ideas

• Youth unemployment, like adult unemployment, is concentrated among a subset with long non-employment spells once labor force withdrawal is accounted for.
• Official youth unemployment rates, by ignoring labor force withdrawal, understate the extent of youth non-employment.
• A shortage of attractive jobs — not primarily poor search technology or reservation wages — appears to drive long-term youth non-employment.
• Demographic heterogeneity within the youth category (teenage vs. young adult, in school vs. out) requires careful disaggregation.
• The policy implications favor demand-side interventions over pure search assistance.

Key takeaway

High youth unemployment is not a benign consequence of normal job-switching; for a substantial minority of youth, it reflects a persistent shortage of acceptable job opportunities that standard frictional models cannot explain.

Chapter 5 — The Demographic Composition of Cyclical Employment Variations

(Originally: Kim B. Clark and Lawrence H. Summers, Journal of Human Resources 16, No. 1, Winter 1980)

Central question

How does the cyclical sensitivity of employment vary across demographic groups, and what does this pattern reveal about labor market segmentation and the nature of cyclical unemployment?

Main argument

Decomposing cyclical employment variation

Clark and Summers use aggregate time-series data to decompose cyclical employment variation into changes due to unemployment and changes due to labor force participation. They then estimate how these components vary by age, sex, race, and marital status. The decomposition reveals that employment fluctuations during recessions and expansions are not distributed uniformly across the workforce.

Youth and secondary workers bear disproportionate cyclical burden

Young workers, women (especially secondary earners), and minority workers experience disproportionately large employment losses during recessions. This is consistent with a model in which primary workers — adult males with stable, high-paying jobs — are insulated from cyclical fluctuations, while secondary workers in more volatile, peripheral employment bear the adjustment burden.

Participation effects versus unemployment effects

An important nuance is that cyclical variation in employment for secondary workers operates substantially through participation rates, not just unemployment rates. In recessions, many secondary workers leave the labor force rather than registering as unemployed, which means aggregate unemployment statistics understate the full cyclical deterioration in labor market conditions for these groups. This reinforces the finding from earlier chapters that the three-state transition framework is essential.

Implications for the added/discouraged worker debate

The analysis bears on the long-standing debate over "added worker" effects (recessions pull secondary workers into the labor force to replace lost primary-earner income) versus "discouraged worker" effects (recessions push secondary workers out of the labor force when jobs become scarce). Clark and Summers find that discouraged worker effects dominate, consistent with aggregate unemployment being understated during downturns.

Key ideas

• Cyclical employment variation is demographically concentrated: youth and secondary workers experience larger proportional employment swings than prime-age adult males.
• Participation rate changes, not just unemployment rate changes, are the primary mechanism of cyclical labor market adjustment for secondary workers.
• Discouraged worker effects dominate added worker effects in recessions, meaning official unemployment rates understate cyclical distress.
• Labor market segmentation between primary and secondary jobs creates fundamentally asymmetric cyclical exposure.

Key takeaway

The cyclical burden of recessions falls disproportionately on the most vulnerable workers — youth, minorities, and secondary earners — and is substantially hidden by the participation margin, making aggregate unemployment statistics incomplete welfare indicators.

Chapter 6 — Labor Force Participation: Timing and Persistence

(Originally: Kim B. Clark and Lawrence H. Summers, Review of Economic Studies 49, No. 5, 1982)

Central question

Does labor force participation respond primarily to current economic conditions (a contemporaneous response to wages and job availability), or are participation decisions persistent — shaped by past experience and creating lasting labor market attachment or detachment?

Main argument

Persistence in participation decisions

Clark and Summers document substantial persistence in individual labor force participation decisions. Workers who are out of the labor force in one period have a dramatically higher probability of being out in subsequent periods than would be implied by a purely contemporaneous model. This persistence is not merely a reflection of stable individual characteristics (fixed effects); it also reflects genuine state dependence — the experience of non-participation itself reduces the probability of future participation.

Mechanisms of persistence

Several mechanisms can generate persistence in participation: (1) human capital depreciation during spells of non-participation, as workers' skills atrophy and their labor market contacts erode; (2) discouragement effects, as workers who experience persistent joblessness reduce their search intensity and eventually withdraw from the labor force; (3) social network effects, as workers outside the labor force lose access to job-matching networks; (4) habit formation in participation behavior.

Timing and cyclical entry/exit

The chapter also examines the timing of participation decisions across the business cycle, finding that entry and exit are asymmetric. Workers tend to exit the labor force more rapidly during downturns than they re-enter during upturns, contributing to the persistence of labor market distress at the individual level even after aggregate conditions improve.

Implications for the natural rate and policy

Persistence in participation has macroeconomic implications: aggregate supply of labor is not a simple function of current wages and prices but depends on labor market history. This is an early presentation of a theme Blanchard and Summers would develop more formally in their hysteresis paper (Chapter 10): the history of unemployment shapes the current equilibrium.

Key ideas

• Labor force participation exhibits genuine state dependence: past participation status affects current participation beyond what fixed individual characteristics predict.
• Non-participation has lasting adverse effects on future employment prospects through human capital depreciation and discouragement.
• Participation entry is slower than exit, creating asymmetric cyclical dynamics.
• The effective labor supply is path-dependent, not a static function of current wage levels.
• Persistence in non-participation has policy implications for the long-term unemployed and discouraged workers.

Key takeaway

Labor force participation is not purely a spot-market decision; it is persistent and path-dependent, meaning that cyclical downturns can cause lasting reductions in effective labor supply through discouragement and human capital loss.

Chapter 7 — Unemployment Insurance and Labor Force Transitions

(Originally: Kim B. Clark and Lawrence H. Summers, in M. Baily, ed., Workers, Jobs, and Inflation, Brookings Institution, 1983)

Central question

How does the availability of unemployment insurance (UI) affect workers' labor force transitions — particularly the duration of unemployment spells and the probability of returning to employment?

Main argument

UI and the labor supply margin

The standard economic analysis of UI focuses on the moral hazard problem: UI reduces the cost of remaining unemployed, thereby increasing the duration of unemployment spells as workers search longer for better job matches. Clark and Summers confirm that UI does extend spell durations, but they complicate the standard story in important ways.

UI and the employment-non-participation margin

A key contribution of this chapter is its examination of UI effects not just on the U→E transition (return to employment) but also on the U→N transition (exit to non-participation). UI benefits, by supporting income during unemployment, may actually keep workers classified as unemployed who might otherwise exit to non-participation. This means UI can, counterintuitively, increase measured unemployment rates even while keeping workers more attached to the labor force than they would be without it.

The spurious flows problem

Building on the measurement concerns developed in earlier chapters, Clark and Summers note that some of the apparent transitions between labor force states in the CPS data are artifacts of measurement error. Workers who should be coded as continuously unemployed are sometimes miscoded as employed in one month. Correcting for these errors matters for estimating the true effect of UI on labor force dynamics.

Policy implications

The analysis suggests that the standard "UI increases unemployment duration" conclusion, while empirically valid, captures only part of the story. UI also affects labor force attachment, labor market search quality (by allowing longer search for better matches), and the composition of those measured as unemployed versus not participating. These multiple margins complicate simple welfare calculations about the costs and benefits of UI.

Key ideas

• UI increases unemployment spell durations through the moral hazard channel, consistent with standard economic predictions.
• UI affects multiple labor force transitions simultaneously: not just U→E but also U→N and N→U.
• The effect of UI on measured unemployment rates may differ from its effect on actual labor market distress.
• Measurement error in CPS transitions complicates the estimation of UI effects on labor force dynamics.
• UI's net welfare effect depends on the balance between moral hazard costs (extended spells) and insurance benefits (income smoothing, search quality).

Key takeaway

Unemployment insurance affects labor market behavior through multiple channels beyond the standard moral hazard story, and its measured effects on unemployment rates must be carefully distinguished from its effects on genuine labor market attachment and search behavior.

Chapter 8 — Reporting Errors and Labor Market Dynamics

(Originally: James M. Poterba and Lawrence H. Summers, Econometrica 54, 1986, pp. 1319–1338)

Central question

How large are measurement errors in CPS labor force status reports, and how do these errors distort observed gross flows between employment, unemployment, and non-participation?

Main argument

The measurement problem in gross flow data

The CPS measures labor force status by interviewing households monthly. Because the same household is interviewed in consecutive months, it is possible to construct "gross flows" — the number of workers observed to move from one labor force status to another in a month. These flows are widely used to study labor market dynamics. But if reported labor force status contains errors, the observed flows will overstate true transitions and understate persistence in true labor force states.

Estimating the incidence of classification error

Poterba and Summers develop a formal model of measurement error in categorical labor force status and use auxiliary data — specifically, the CPS's re-interview survey, in which interviewers re-contact a subset of respondents shortly after the initial interview — to estimate the incidence and pattern of classification errors. Their estimates show that errors are substantial: a significant fraction of reported U→E and U→N transitions in a month are in fact workers who were continuously in the same state, misclassified in one of the two interviews.

Implications for measured gross flows

Correcting for classification errors substantially reduces the apparent size of gross flows. The "true" labor market, once corrected, shows more persistence in labor force status than the raw data suggest. This has implications for the welfare analysis of unemployment: if workers classified as briefly unemployed are in fact continuously unemployed (misclassified as employed in one month's interview), the long-spell concentration of unemployment is even greater than the raw data indicate.

Implications for estimating behavioral parameters

The classification error also biases behavioral parameters estimated from gross flow data — including estimates of the effect of unemployment insurance on transition rates (Chapter 7) and estimates of job-finding rates. Poterba and Summers show how to correct for these biases using the multinomial logit framework, deriving corrected estimates that show higher persistence and lower gross flow rates than naive analyses.

Key ideas

• CPS labor force status classification contains substantial measurement error, as confirmed by re-interview data.
• Classification errors generate spurious gross flows, making the labor market appear more dynamic than it actually is.
• True persistence in labor force status — particularly in unemployment — is higher than raw gross flow data indicate.
• The long-spell concentration of unemployment is even more pronounced after correcting for measurement error.
• Behavioral parameters (UI effects, job-finding rates) must be estimated from error-corrected data to avoid systematic bias.

Key takeaway

A significant fraction of observed month-to-month labor market transitions are measurement artifacts rather than real changes in labor force status, and correcting for this reveals a labor market characterized by greater persistence and greater long-spell unemployment than standard data suggest.

Chapter 9 — A Theory of Dual Labor Markets with Application to Industrial Policy, Discrimination, and Keynesian Unemployment

(Originally: Jeremy I. Bulow and Lawrence H. Summers, Journal of Labor Economics 4, No. 3, Part 1, 1986, pp. 376–414)

Central question

Can the empirical patterns of labor market segmentation — the existence of high-wage "primary" jobs and low-wage "secondary" jobs — be explained within an efficiency wage framework, and what are the implications for unemployment, discrimination, and industrial policy?

Main argument

Efficiency wages and the dual labor market

Bulow and Summers build on the efficiency wage idea: some firms find it profit-maximizing to pay workers a wage above the market-clearing level, because higher wages induce greater effort (through reduced shirking), lower turnover, or improved morale. The key insight is that efficiency wage considerations are not uniformly applicable across all jobs. In jobs where output is easily monitored, where turnover costs are low, and where effort variability is small, there is little reason to pay above the market wage. In jobs where monitoring is costly, turnover is expensive, and effort variation is large, efficiency wages become important. This creates endogenous segmentation: a primary sector with efficiency wages, above-market pay, and rationed employment, and a secondary sector with market-clearing wages and high turnover.

Unemployment as the equilibrating mechanism

In the dual labor market model, unemployment performs a disciplinary function. Workers in the primary sector earn a rent (the wage premium above the secondary-sector wage plus the value of the job). For the efficiency wage to deter shirking, there must be a cost to being caught and losing the job — that cost is unemployment. If workers who are fired immediately find equivalent primary-sector jobs, the efficiency wage mechanism fails. Unemployment is therefore not an aberration to be eliminated but a structural feature required to maintain the incentive properties of efficiency wages.

Implications for discrimination

Bulow and Summers show that the model can generate and sustain occupational discrimination even in competitive product markets. If women or minorities face statistical discrimination that excludes them from primary sector jobs — for historical or social reasons — market forces will not automatically erode this discrimination, because the rents in primary employment create a wedge that insulates incumbents from competitive pressure. This provides a coherent theory of persistent wage discrimination that is immune to the standard competitive arbitrage argument.

Implications for industrial policy

The existence of primary sector rents means that policies that expand employment in high-wage, high-productivity industries generate wage gains that exceed simple labor reallocation effects. An industrial policy that shifts labor from the secondary to the primary sector can raise welfare above market outcomes. This is a direct challenge to the view that competitive labor markets allocate workers optimally.

Key ideas

• Efficiency wages — wages above market clearing paid to elicit effort — are heterogeneous across job types, generating endogenous labor market segmentation.
• Unemployment equilibrates the primary sector labor market by making job loss costly; it is not a temporary disequilibrium but a structural feature.
• Primary sector employment is rationed: at the going wage, more workers would like primary jobs than firms create them.
• Discrimination can persist in this model because primary sector rents insulate incumbent workers from competitive pressure.
• Industrial policy can raise welfare by shifting employment toward high-wage, high-monitoring industries.

Key takeaway

Efficiency wages, rather than union power or government regulation, can generate persistent labor market segmentation, rationed unemployment, and sustained wage discrimination — making unemployment an equilibrium feature of competitive markets, not a symptom of market failure correctable by deregulation.

Chapter 10 — Relative Wages, Efficiency Wages, and Keynesian Unemployment

(Originally: Lawrence H. Summers, American Economic Review 78, No. 2, May 1988, pp. 383–388)

Central question

Can the observed patterns of wage rigidity, inter-industry wage differentials, and involuntary unemployment be explained by efficiency wage theories in which workers' productivity depends on their relative wage — and how does this challenge the view that nominal rigidities explain business cycle fluctuations?

Main argument

The case against purely nominal models

Summers opens by distinguishing his argument from Keynesian models in which nominal wage stickiness drives unemployment. If unemployment were caused purely by nominal wage rigidity, one would expect that, in the long run, real wages would clear labor markets. But the data show persistent real wage differentials across industries and persistent involuntary unemployment — both inconsistent with a purely nominal explanation.

Relative wage efficiency theories

Summers argues that the most promising framework is one in which workers' morale, effort, and productivity depend on their wage relative to wages elsewhere in the economy. Firms that pay above-average wages attract workers who value their jobs more highly and are therefore less likely to shirk. The resulting equilibrium is characterized by inter-industry wage differentials (because different industries have different optimal efficiency wage premia) and by unemployment (because above-clearing wages in the efficiency wage sector mean that some workers who want primary-sector jobs cannot get them).

Evidence from inter-industry wage differentials

The paper reviews evidence — much of it developed by Krueger and Summers (1987) in a companion paper — that inter-industry wage differentials are large (50 percent or more between high- and low-wage industries, even after controlling for worker characteristics), persistent over time and across countries, and correlated with industry characteristics consistent with efficiency wage theories (capital intensity, monitoring difficulty, union presence). These differentials are too large and too persistent to be explained purely by compensating differentials or by unobserved worker quality.

Keynesian unemployment as a real phenomenon

The chapter argues that the efficiency wage framework provides a foundation for genuinely Keynesian — that is, demand-driven — unemployment that does not rely on nominal rigidities. When aggregate demand falls, the equilibrium level of employment falls because the efficiency wage does not adjust downward: cutting wages would reduce worker effort and morale, making it unprofitable even in the absence of nominal contracts. The resulting unemployment is involuntary — workers would prefer to work at the going wage but cannot find jobs.

Key ideas

• Efficiency wages — wages set to maximize worker effort relative to their cost — can explain involuntary unemployment without relying on nominal wage stickiness.
• Inter-industry wage differentials are large, persistent, and correlated with efficiency-wage-relevant industry characteristics.
• Relative wage concerns (workers caring about their wage relative to others) provide a particularly clean motivation for efficiency wages.
• Efficiency wage unemployment is genuinely involuntary: workers cannot underbid the efficiency wage without reducing the firm's profits.
• The framework explains cyclical unemployment through demand-side mechanisms that operate on real wages, not merely nominal adjustment lags.

Key takeaway

Efficiency wage theories grounded in relative wage considerations provide a coherent, empirically supported explanation for persistent inter-industry wage differentials and involuntary unemployment that is independent of, and more fundamental than, nominal wage rigidity stories.

Chapter 11 — Hysteresis and the European Unemployment Problem

(Originally: Olivier J. Blanchard and Lawrence H. Summers, NBER Macroeconomics Annual 1, 1986, pp. 15–78)

Central question

Why did European unemployment rise sharply in the early 1980s following the oil shocks and contractionary monetary policy, and — unlike in earlier cycles — fail to return to its pre-shock level when demand recovered? Can the phenomenon of "hysteresis" — in which the natural rate of unemployment itself shifts in response to actual unemployment — explain this pattern?

Main argument

The European unemployment puzzle

From the late 1970s through the mid-1980s, unemployment in Germany, France, the United Kingdom, and other European economies rose dramatically — from roughly 3–5 percent to 8–11 percent — and remained elevated well after the negative demand shocks that caused the initial rise had dissipated. Standard economic models predicted that unemployment would return to a stable natural rate once demand recovered. It did not. Blanchard and Summers call this persistence — the tendency of unemployment to remain near its recently elevated level rather than reverting to a prior equilibrium — hysteresis.

The insider-outsider mechanism

The primary mechanism generating hysteresis in the Blanchard-Summers model is the insider-outsider theory of wage determination. In firms with seniority-based employment protection and collective bargaining, wages are set by currently employed insiders — workers with jobs — rather than by competitive market forces that would weight the preferences of unemployed outsiders. When a demand shock causes layoffs, the new smaller group of insiders sets wages to target full employment of themselves, with no weight on the interests of the newly unemployed outsiders. The result: real wages do not fall enough to re-absorb the unemployed, and the higher unemployment rate becomes self-perpetuating.

The membership model of hysteresis

Blanchard and Summers formalize this as a membership model: the wage is set to clear the labor market at the level of employment that satisfies current insiders. When employment falls — whether from demand shocks or supply shocks — the number of insiders falls, the wage target adjusts upward (relative to output prices), and the new, lower employment level is sustained. The natural rate is therefore not a fixed parameter but a function of the history of actual employment — genuine hysteresis in the technical sense.

Duration theories as a complementary mechanism

A second mechanism, which Blanchard and Summers term the duration theory, is that long-term unemployed workers become progressively less effective as job-searchers. They may lose skills, become demoralized, or be stigmatized by employers who treat duration of unemployment as a negative signal. As the long-term unemployed effectively exit the effective labor supply (though they remain officially unemployed), the effective labor supply contracts and wages rise, reinforcing the elevated unemployment equilibrium.

Contrasting the United States and Europe

The paper compares the United States, where unemployment rose sharply in the early 1980s recession but then recovered relatively quickly, with Europe, where it did not. The contrast is attributed to institutional differences: U.S. labor markets have less employment protection, weaker insider effects, and more rapid adjustment of wages to excess supply. The European institutional framework — employment protection legislation, strong unions, centralized wage bargaining — creates the conditions for hysteresis by amplifying insider power.

Key ideas

• Hysteresis occurs when the equilibrium unemployment rate (NAIRU) shifts in response to actual unemployment, making shocks permanent rather than temporary.
• The insider-outsider mechanism generates hysteresis: insiders set wages without weighting the interests of unemployed outsiders, so job losses produce persistent wage pressure rather than correction.
• The membership model formalizes hysteresis: the target employment level tracks the current number of insiders, not some stable full-employment level.
• Long-duration unemployment compounds hysteresis through duration effects: long-term unemployed workers become less effective searchers, effectively shrinking the efficient labor supply.
• European institutional features — employment protection, union bargaining — amplify insider effects and explain the contrast with the United States.

Key takeaway

Hysteresis — the persistence of high unemployment after demand recovers — arises from the insider-outsider structure of European wage bargaining, which means that any shock that reduces employment creates a new, self-sustaining equilibrium at higher unemployment, transforming transitory shocks into permanent changes in the natural rate.

Chapter 12 — Conclusion: Theories of Unemployment for the 1990s

Central question

What are the theoretical frameworks best suited to explain the empirical facts about unemployment assembled across the preceding papers, and what do those frameworks imply for policy?

Main argument

The inadequacy of standard models

Summers' Conclusion evaluates the major theoretical paradigms against the book's empirical findings. The standard competitive model — in which unemployment is frictional, brief, and voluntary — fails on several dimensions: it predicts rapid wage adjustment that the data do not support, implies unemployment concentrated in short spells when it is concentrated in long spells, and cannot explain persistent inter-industry wage differentials.

The efficiency wage synthesis

The most promising class of models, in Summers' assessment, is efficiency wage theory — particularly relative wage versions and the dual labor market formalization — because it explains wage rigidity without relying on nominal contracts, generates rationed primary sector employment and involuntary unemployment, can account for inter-industry wage differentials, and is consistent with the data on long-spell unemployment concentration.

Hysteresis and the NAIRU

The Conclusion integrates the hysteresis findings of Chapter 11 with the domestic U.S. context. Summers argues that insider-outsider effects, while weaker in the United States than in Europe, are nonetheless present and contribute to the persistence of unemployment following recessions. The NAIRU is not a fixed structural parameter; it responds to the history of unemployment, with important implications for macro policy.

Policy implications

Drawing together the book's empirical and theoretical strands, Summers argues for:

• Active labor market policies targeting long-spell unemployed workers, who bear the bulk of unemployment's welfare costs.
• Recognition that unemployment insurance has multiple effects — some beneficial (insurance value, search quality) and some costly (moral hazard) — that must be weighed together.
• Caution about relying on wage moderation as the primary cure for unemployment, given the efficiency wage argument that wage cuts may reduce rather than increase employment in primary sector jobs.
• Greater attention to the distributional incidence of unemployment, which falls disproportionately on youth and secondary workers.

Rethinking the natural rate

Perhaps the Conclusion's most fundamental claim is that the natural rate of unemployment is not a fixed, policy-neutral benchmark but a malleable variable influenced by the history of actual unemployment (through hysteresis), by institutional features of wage setting (insider-outsider dynamics), and by the distributional structure of the labor market (efficiency wages, dual labor markets). This view has implications for how monetary and fiscal authorities should respond to unemployment — more aggressively combating demand shortfalls to prevent hysteresis, rather than accepting any given unemployment level as "natural."

Key ideas

• Efficiency wage theories, especially relative wage and dual labor market variants, provide the most empirically adequate theoretical framework for the book's findings.
• The natural rate of unemployment is endogenous to labor market history, institutions, and policies — not a fixed structural parameter.
• Hysteresis means that demand-side policies matter for the long-run equilibrium unemployment rate, not just for short-run cyclical fluctuations.
• Unemployment's welfare costs are highly concentrated: policy should focus on long-spell unemployed workers and disadvantaged demographic groups.
• No single theory adequately explains all unemployment phenomena; the correct framework is pluralist, combining efficiency wage, insider-outsider, and search elements.

Key takeaway

An adequate theory of unemployment for the 1990s must incorporate wage rigidity from efficiency wage considerations, hysteresis from insider-outsider dynamics, and careful measurement of unemployment's true incidence — a richer and more empirically disciplined framework than the natural-rate/frictional synthesis it replaces.

The book's overall argument

1. Chapter 1 (Introduction) — Establishes the empirical agenda: the data show unemployment concentrated in long spells, systematically understated by official measures, and disproportionately costly to vulnerable groups — facts that standard frictional models cannot accommodate.
2. Chapter 2 (Labor Market Dynamics and Unemployment: A Reconsideration) — Demonstrates that the stock-flow decomposition of unemployment reveals long-spell workers dominate the welfare burden, directly refuting the turnover view that unemployment is primarily brief and frictional.
3. Chapter 3 (Labor Force Transitions and Unemployment) — Extends the analysis to a three-state framework (E, U, N), showing that the employment-to-non-participation margin is economically significant and that official unemployment rates substantially undercount labor market distress.
4. Chapter 4 (The Dynamics of Youth Unemployment) — Applies the long-spell and labor-force-withdrawal findings specifically to youth, showing that inadequate job supply — not merely search frictions — drives persistent youth non-employment.
5. Chapter 5 (The Demographic Composition of Cyclical Employment Variations) — Documents that cyclical downturns impose disproportionate costs on youth, minorities, and secondary earners, primarily through participation effects rather than unemployment rates alone.
6. Chapter 6 (Labor Force Participation: Timing and Persistence) — Establishes that labor force participation is path-dependent and persistent, meaning that downturns cause lasting reductions in effective labor supply through discouragement and human capital depreciation.
7. Chapter 7 (Unemployment Insurance and Labor Force Transitions) — Complicates the standard UI moral hazard story by showing that UI affects multiple labor force transitions simultaneously, with ambiguous net effects on measured unemployment and genuine labor market attachment.
8. Chapter 8 (Reporting Errors and Labor Market Dynamics) — Demonstrates that CPS measurement errors create spurious gross flows that overstate labor market dynamism and understate true persistence in unemployment, strengthening the case for long-spell concentration.
9. Chapter 9 (A Theory of Dual Labor Markets) — Provides the theoretical framework: efficiency wages generate endogenous segmentation between primary and secondary labor markets, with unemployment as the equilibrating mechanism that makes efficiency wages incentive-compatible.
10. Chapter 10 (Relative Wages, Efficiency Wages, and Keynesian Unemployment) — Develops the efficiency wage argument into a theory of involuntary unemployment independent of nominal rigidities, supported by evidence on large and persistent inter-industry wage differentials.
11. Chapter 11 (Hysteresis and the European Unemployment Problem) — Extends the insider-outsider framework to macroeconomic dynamics, showing that unemployment shocks can permanently shift the natural rate through the insider-outsider mechanism, explaining European unemployment persistence.
12. Chapter 12 (Conclusion) — Synthesizes the empirical and theoretical findings into an agenda for unemployment research and policy: the NAIRU is endogenous, efficiency wages explain involuntary unemployment, and policy should target long-spell workers.

Common misunderstandings

Misunderstanding: The book argues that unemployment is always high because labor markets never clear.

The book does not claim that labor markets never clear or that unemployment is always excessive. It argues that specific mechanisms — efficiency wages, insider-outsider dynamics, measurement distortions — explain why unemployment can be persistently elevated and involuntary in particular circumstances. Frictional and search-related unemployment are real and acknowledged; the argument is that they are insufficient to explain the observed distribution and persistence of unemployment.

Misunderstanding: Summers argues that unemployment insurance should be eliminated to reduce moral hazard.

The book reaches no such conclusion. The analysis of UI (Chapter 7) shows that UI has multiple effects — some that extend spells (moral hazard) and some that improve labor market attachment and search quality. The net welfare effect is treated as an empirical question, not predetermined. Summers' approach is to provide better measurement and modeling, not to advocate for UI reduction.

Misunderstanding: The hysteresis argument implies that any unemployment rate can be sustained permanently.

Hysteresis in the Blanchard-Summers model is not a claim that any unemployment rate is an equilibrium. It is a claim that the natural rate shifts in the direction of actual unemployment, so that transitory shocks can produce lasting effects on equilibrium unemployment. The mechanism is specific — insider-outsider wage setting — and has limits; extreme unemployment rates would eventually erode insider power or trigger institutional change.

Misunderstanding: Because the book collects papers from multiple co-authors, it lacks a unified argument.

The book is edited with a unifying Introduction and Conclusion written by Summers specifically to integrate the papers. The empirical papers (Chapters 2–8, primarily with Clark and Poterba) establish the stylized facts; the theoretical papers (Chapters 9–11, with Bulow, Blanchard) provide the explanations. The arc from measurement through theory to policy synthesis is deliberate and explicit.

Misunderstanding: Efficiency wages are the book's only theoretical framework.

While efficiency wage and dual labor market theories occupy a central position, the book explicitly treats them as partial explanations. The hysteresis chapter offers an entirely distinct mechanism — insider-outsider dynamics in collective bargaining — and the empirical chapters are deliberately theory-neutral, allowing the facts to constrain theory rather than the reverse.

Central paradox / key insight

The central paradox of the book is that unemployment, in standard competitive theory, should be brief and largely voluntary — a friction incurred during the search for better job matches. The data assembled across Summers' papers tell a fundamentally different story: the workers who account for most of the welfare cost of unemployment are those experiencing long, persistent spells of joblessness, not those briefly between jobs by choice. This long-spell concentration is not anomalous or cyclical; it is a structural feature of labor markets.

The resolution of this paradox lies in the book's theoretical contributions: wages do not fall to clear labor markets because reducing wages would reduce effort (efficiency wages), because incumbent workers set wages without regard to the unemployed (insider-outsider dynamics), and because the history of unemployment shapes the equilibrium unemployment rate (hysteresis). The market for labor is not a spot auction in which all willing workers find employment at market-clearing wages; it is a set of ongoing relationships governed by incentive, bargaining, and institutional forces that systematically exclude some workers from employment at going wages.

The core insight is that unemployment is concentrated, involuntary, and equilibrium-sustaining — not the brief, frictional, and self-correcting phenomenon that standard models predict.

Important concepts

Unemployment pool decomposition

The analytical distinction between the stock of unemployment (the number unemployed at a point in time) and the flow of unemployment (workers entering and exiting). The pool is dominated by long-spell workers even when flows are dominated by short-spell workers, because long-spell workers accumulate in the cross-section. This distinction is fundamental to the book's empirical claim about the concentration of unemployment's welfare burden.

Length-biased sampling

A statistical phenomenon in which sampling from a stock (currently unemployed workers) over-represents items with long durations. A randomly selected currently unemployed worker is more likely to be in the middle of a long spell than a short one, biasing perceptions of typical spell length toward long spells. The same phenomenon is the reason why a randomly chosen commuter on a subway car is more likely to be on a crowded train — long trains carry more passengers.

Labor force non-participation as disguised unemployment

The recognition that workers who leave the labor force (move from U to N in the CPS) often remain in a state of weak labor market attachment, cycling back into U or E in subsequent months. True non-employment — including discouraged workers and marginally attached workers — exceeds measured unemployment, making official rates underestimate labor market distress.

Classification error / reporting error

Measurement errors in CPS labor force status coding that generate spurious month-to-month transitions between employment, unemployment, and non-participation. Correcting for these errors (as Poterba and Summers do in Chapter 8) reveals a more persistent labor market, with greater long-spell unemployment and less genuine dynamism than the raw data suggest.

Efficiency wages

Wages set above the market-clearing level by profit-maximizing firms, because higher wages induce greater worker effort (by making job loss more costly), lower turnover, or higher morale. In efficiency wage models, workers who want jobs at the going wage cannot find them — unemployment is genuinely involuntary, not merely frictional.

Dual labor market

An economy-wide labor market segmented into a primary sector — characterized by high wages, employment stability, advancement opportunities, and efficiency wage premia — and a secondary sector — characterized by low wages, high turnover, and absence of efficiency wage effects. Access to primary sector jobs is rationed, meaning some workers willing to accept primary sector wages cannot find such jobs and are consigned to secondary employment or unemployment.

Insider-outsider theory

A theory of wage determination in which wages are set by currently employed insiders — who have bargaining power through union membership, employment protection legislation, or firm-specific human capital — without regard to the unemployed outsiders who would like jobs at the going wage or lower. Insiders set wages that protect their own employment at the expense of the unemployed, preventing the wage adjustment that would absorb outsiders.

Hysteresis

In macroeconomics, the property that the equilibrium unemployment rate (the NAIRU or natural rate) shifts in the direction of actual unemployment following a shock, rather than reverting to a pre-shock level. Named by analogy with the physical phenomenon of path-dependent systems (iron retaining magnetization). In Blanchard and Summers' model, hysteresis arises from the insider-outsider mechanism: when employment falls, the insider group shrinks and the new smaller group targets wages to protect its own employment, sustaining the elevated unemployment rate.

NAIRU (Non-Accelerating Inflation Rate of Unemployment)

The rate of unemployment at which inflation neither accelerates nor decelerates, often used as a policy benchmark for "full employment." The book challenges the view of the NAIRU as a fixed structural parameter, arguing that it is endogenous to the history of actual unemployment (through hysteresis) and to institutional features of labor markets (insider-outsider dynamics, efficiency wages).

Inter-industry wage differentials

Persistent, large differences in wages paid to observationally similar workers across industries that cannot be explained by compensating differentials or measured differences in worker quality. Documented by Krueger and Summers and used by Summers in Chapter 10 as evidence for efficiency wages, because competitive markets with mobile workers should eliminate such differentials.

Discouraged workers

Workers who have given up actively searching for employment because they believe no suitable jobs are available. Because discouraged workers are not actively searching, they are classified as non-participants (N) rather than unemployed (U) in official statistics, even though they represent labor market distress. Their exclusion from official unemployment rates is one source of the systematic undercount of unemployment's true extent.

References and Web Links

Primary book and edition information

• Summers, Lawrence H. Understanding Unemployment. MIT Press, Cambridge, MA, 1990. ISBN 0262192659 (hardcover); 0262691574 (paperback, 1992).
  • MIT Press book page
  • Internet Archive — borrowable digital copy
  • Amazon listing — hardcover
  • Goodreads entry

Background and overview

• Lawrence Summers — AEA John Bates Clark Medal citation 1993, summarizing his unemployment research
• Lawrence Summers — Research page at larrysummers.com
• IDEAS/RePec entry for Lawrence H. Summers
• EconPapers — 1991 journal review of the book

Key chapter papers — original publications

• Clark, Kim B. and Lawrence H. Summers. "Labor Market Dynamics and Unemployment: A Reconsideration." Brookings Papers on Economic Activity 1 (1979): 13–60.

  • Brookings article page
  • BYU Scholars Archive

• Clark, Kim B. and Lawrence H. Summers. "The Dynamics of Youth Unemployment." In R. Freeman and D. Wise, eds., The Youth Labor Market Problem. University of Chicago Press, 1982.

  • NBER Working Paper No. 274
  • BYU Scholars Archive

• Clark, Kim B. and Lawrence H. Summers. "The Demographic Composition of Cyclical Variations in Employment." Journal of Human Resources 16, No. 1 (Winter 1980).

  • BYU Scholars Archive

• Clark, Kim B. and Lawrence H. Summers. "Labor Force Participation: Timing and Persistence." Review of Economic Studies 49, No. 5 (1982).

  • NBER Working Paper No. 977

• Poterba, James M. and Lawrence H. Summers. "Reporting Errors and Labor Market Dynamics." Econometrica 54 (1986): 1319–1338.

  • IDEAS/RePec entry

• Bulow, Jeremy I. and Lawrence H. Summers. "A Theory of Dual Labor Markets with Application to Industrial Policy, Discrimination, and Keynesian Unemployment." Journal of Labor Economics 4, No. 3, Part 1 (1986): 376–414.

  • NBER Working Paper No. 1666
  • Journal of Labor Economics abstract

• Summers, Lawrence H. "Relative Wages, Efficiency Wages, and Keynesian Unemployment." American Economic Review 78, No. 2 (May 1988): 383–388.

  • NBER Working Paper No. 2590
  • IDEAS/RePec entry

• Blanchard, Olivier J. and Lawrence H. Summers. "Hysteresis and the European Unemployment Problem." NBER Macroeconomics Annual 1 (1986): 15–78.

  • IDEAS/RePec chapter entry
  • ResearchGate PDF

Related Summers papers on unemployment (not necessarily in this volume)

• Clark, Kim B. and Lawrence H. Summers. "Labor Force Transitions and Unemployment." NBER Working Paper No. 277, 1978.

  • NBER page

• Poterba, James M. and Lawrence H. Summers. "Unemployment Benefits and Labor Market Transitions: A Multinomial Logit Model with Errors in Classification." Review of Economics and Statistics 77, No. 2 (1995): 207–216.

  • NBER Working Paper No. 4434

Additional study resources

These are secondary summaries and should be used alongside, rather than instead of, the original book.

• Econlib Encyclopedia entry on Unemployment — background on unemployment measurement and theory
• Efficiency wage — Wikipedia article providing background on the theory