Unemployment and the business cycle

Cyclical unemployment is unemployment above the natural rate, caused by a fall in labour demand in a recession. In a competitive market a fall in demand lowers the wage until everyone who wants work at the new wage has it, so the theories differ on why unemployment rises: very elastic labour supply (intertemporal substitution), reallocation across sectors (sectoral shifts), or wages held above the market-clearing level (efficiency wages and sticky wages). Inflation and unemployment trade off only in the short run; the long-run Phillips curve is vertical at the natural rate.

The puzzle

Recessions bring large swings in employment and small changes in wages. From 2008 to 2009 Canada’s unemployment rate rose from 6.3 to 8.5 percent (annual averages),¹ yet average hourly wages of full-time employees kept rising: from $25.03 to $25.85 for men and from $21.12 to $22.16 for women.² Part of that rise can be composition, since an average rises when lower-paid workers are the ones laid off, but wages plainly did not fall to clear the market. Figure 1 sets out the two ways to reconcile the competitive model with this pattern.

Elasticity of labour supply

0.5

Rigid wage (R): employment

40.0k workers

Rigid wage (R): involuntary unemployment

20.0k workers

Flexible wage (F): employment

53.8k workers

Flexible wage (F): wage

$15.86

Open in Graph Lab

Figure 1. Labour demand falls by $6 an hour. With the wage stuck at $20, employment falls to 40 thousand at R and 20 thousand people who want work at $20 cannot find it; with a flexible wage the market clears at F, where employment falls by less the less elastic labour supply is.

To get a large fall in employment with a small fall in the wage, either labour supply must be very elastic (point F with a high elasticity) or the wage must not fall (point R).

The intertemporal substitution hypothesis

In the life-cycle model a worker who expects the wage to be temporarily low works less now and more later, when the same hours buy more (life cycle and business cycle, intertemporal choice). The intertemporal substitution hypothesis applies this to recessions: a downturn lowers the real wage temporarily, workers choose more leisure now, and employment falls along a very elastic labour supply curve.³ On this view recession unemployment is voluntary, and a recession is an efficient response to a temporary fall in productivity.

Three pieces of evidence count against it as the main explanation:

• Small micro elasticities. Using micro data on individual workers, Altonji found the intertemporal substitution elasticity of married men’s hours positive but small.⁴ The response can be large where workers set their own hours: in a randomized field experiment, Fehr and Goette found that a temporary wage increase raised hours sharply, while effort per hour fell.⁵
• Layoffs, not quits. If workers were choosing leisure, unemployment would rise through quits. It rises through layoffs. Among Canada’s unemployed, the number who had lost their jobs rose from 538,500 in 2008 to 837,600 in 2009, while the number who had left their jobs rose only from 274,000 to 292,400. In 2020 job losers more than doubled, from 491,900 to 1,220,700, while job leavers fell slightly.⁶
• Wages. The hypothesis needs the wage to fall noticeably in recessions. In 2009, as shown above, average wages did not fall at all.

The sectoral shifts hypothesis

Many shocks reallocate labour demand across industries and regions rather than shrinking it everywhere. Workers whose skills and locations suited the shrinking sector need time to retrain or move, so unemployment rises even with total labour demand unchanged. The extra unemployment is structural. Lilien found that US unemployment was high when employment growth was unusually dispersed across industries, and read this as evidence that sectoral shifts drive much of cyclical unemployment.⁷

Abraham and Katz showed that a fall in aggregate demand produces the same correlation, and separated the two explanations with job vacancies, measured by help-wanted advertising. A sectoral shift raises vacancies in growing sectors while unemployment rises in shrinking ones; a fall in aggregate demand lowers vacancies everywhere. Dispersion moved with falling help-wanted advertising, and they concluded that pure sectoral shifts have not been an important source of cyclical swings in US unemployment.⁸

Unemployment rate (moved by aggregate demand)

6.0%

Mismatch (position of the curve)

18

Unemployment rate

6.0%

Vacancy rate

3.0%

Unemployed people per vacancy

2.0

Open in Graph Lab

Figure 2. The Beveridge curve. A change in aggregate demand moves the economy along the curve, so vacancies and unemployment move in opposite directions; a sectoral shift that worsens matching moves the curve out, raising both.

Canada’s 2014 to 2016 oil price collapse was a regional and sectoral shock. West Texas Intermediate crude fell from $105.79 a barrel in June 2014 to $30.32 in February 2016. Alberta’s unemployment rate rose from 4.8 percent in 2014 to 8.2 percent in 2016, while Ontario’s fell from 7.4 to 6.6 percent and the national rate barely moved, from 7.0 to 7.1 percent.¹

Crude oil price: West Texas Intermediate

Monthly average spot price, Cushing, Oklahoma

Crude oil price, West Texas Intermediate, monthly. The price fell from $105.79 a barrel in June 2014 to $30.32 in February 2016.Source: FRED, Crude Oil Prices: West Texas Intermediate (WTI) - Cushing, Oklahoma (MCOILWTICO), U.S. Energy Information Administration. Retrieved 2026-10-23.

Show the data and notes

Monthly averages of daily spot prices, nominal dollars (not adjusted for inflation).

Period	WTI
Jan 2012	$100.27
Feb 2012	$102.20
Mar 2012	$106.16
Apr 2012	$103.32
May 2012	$94.66
Jun 2012	$82.30
Jul 2012	$87.90
Aug 2012	$94.13
Sep 2012	$94.51
Oct 2012	$89.49
Nov 2012	$86.53
Dec 2012	$87.86
Jan 2013	$94.76
Feb 2013	$95.31
Mar 2013	$92.94
Apr 2013	$92.02
May 2013	$94.51
Jun 2013	$95.77
Jul 2013	$104.67
Aug 2013	$106.57
Sep 2013	$106.29
Oct 2013	$100.54
Nov 2013	$93.86
Dec 2013	$97.63
Jan 2014	$94.62
Feb 2014	$100.82
Mar 2014	$100.80
Apr 2014	$102.07
May 2014	$102.18
Jun 2014	$105.79
Jul 2014	$103.59
Aug 2014	$96.54
Sep 2014	$93.21
Oct 2014	$84.40
Nov 2014	$75.79
Dec 2014	$59.29
Jan 2015	$47.22
Feb 2015	$50.58
Mar 2015	$47.82
Apr 2015	$54.45
May 2015	$59.27
Jun 2015	$59.82
Jul 2015	$50.90
Aug 2015	$42.87
Sep 2015	$45.48
Oct 2015	$46.22
Nov 2015	$42.44
Dec 2015	$37.19
Jan 2016	$31.68
Feb 2016	$30.32
Mar 2016	$37.55
Apr 2016	$40.75
May 2016	$46.71
Jun 2016	$48.76
Jul 2016	$44.65
Aug 2016	$44.72
Sep 2016	$45.18
Oct 2016	$49.78
Nov 2016	$45.66
Dec 2016	$51.97
Jan 2017	$52.50
Feb 2017	$53.47
Mar 2017	$49.33
Apr 2017	$51.06
May 2017	$48.48
Jun 2017	$45.18
Jul 2017	$46.63
Aug 2017	$48.04
Sep 2017	$49.82
Oct 2017	$51.58
Nov 2017	$56.64
Dec 2017	$57.88
Jan 2018	$63.70
Feb 2018	$62.23
Mar 2018	$62.73
Apr 2018	$66.25
May 2018	$69.98
Jun 2018	$67.87
Jul 2018	$70.98
Aug 2018	$68.06
Sep 2018	$70.23
Oct 2018	$70.75
Nov 2018	$56.96
Dec 2018	$49.52

Wages that do not clear the market

If the wage stays above the market-clearing level, the economy ends up at R in Figure 1 rather than F. The gap between labour supplied and labour demanded at that wage is involuntary unemployment: people willing to work at the going wage who cannot find a job.

Efficiency wages. A firm may pay above the market-clearing wage on purpose, because a higher wage raises productivity: workers shirk less when losing the job is costly, quit less, and the firm attracts better applicants. In the shirking model of Shapiro and Stiglitz, unemployment is what makes job loss costly, so it persists in equilibrium as a worker discipline device.⁹ The unemployed cannot bid the wage down, because a firm would not trust a worker paid less to work hard. Ford’s 1914 five-dollar day is the classic case: it drew long queues of job applicants, and productivity and profits rose.¹⁰ The model is developed in efficiency wages.

Sticky wages. Wages set in contracts, union agreements or by custom adjust slowly, and employers are reluctant to cut nominal pay. When inflation is low, a fall in the real wage requires a nominal cut. Card and Hyslop estimated that about 6 to 10 percent of US workers have nominally rigid wages when inflation is 10 percent, and over 15 percent when it is 5 percent, although they found only a weak link between inflation and how fast relative wages adjusted across states.¹¹

Key idea

Employment falls a lot and wages a little in recessions. The intertemporal substitution hypothesis explains this with very elastic labour supply and voluntary unemployment; efficiency wages and sticky wages explain it with a wage above the market-clearing level and involuntary unemployment.

The Phillips curve and the natural rate

Phillips found a stable negative relationship between unemployment and the rate of change of money wages in the United Kingdom from 1861 to 1957.¹² Read as a menu, the curve seemed to let policy buy lower unemployment with higher inflation. Friedman and Phelps argued that the trade-off lasts only while inflation differs from what workers and firms expect.¹³ The expectations-augmented Phillips curve is

$$\pi = \pi^e - \beta\,(u - u^*), \qquad \beta > 0,$$

where $\pi$ is inflation, $\pi^e$ expected inflation and $u^*$ the natural rate of unemployment. Unemployment falls below the natural rate only when inflation exceeds what people expected, for example because wages were set expecting lower prices. Once expectations catch up, $\pi = \pi^e$ and unemployment returns to $u^*$. The long-run Phillips curve is vertical at the natural rate.

Unemployment rate, u

6.0%

Expected inflation, π^e

2.0%

Inflation, π

2.0%

Expected inflation, π^e

2.0%

Inflation surprise, π − π^e

0.0 points

Unemployment gap, u − u^*

0.0 points

Open in Graph Lab

Figure 3. Short-run Phillips curves $\pi = \pi^e - 0.5(u - 6)$ and the vertical long-run curve at the natural rate of 6 percent. Unemployment held at 4 percent puts inflation a point above expectations, and each time expectations catch up the short-run curve shifts up, so inflation keeps rising.

The natural rate is the frictional and structural unemployment the labour market generates when expectations are met, so it depends on how quickly workers and jobs are matched, on mismatch, and on institutions such as unemployment insurance. The same idea is often called the NAIRU, the non-accelerating-inflation rate of unemployment. The Bank of Canada has targeted inflation since 1991; the target is the 2 percent midpoint of a 1 to 3 percent range, renewed in December 2021 for the five years to the end of 2026.¹⁴

Worked example

The short-run Phillips curve is $\pi = \pi^e - 0.5(u - 6)$, and expected inflation each year equals last year’s inflation. The economy starts at $u = 6$ with $\pi = \pi^e = 2$ percent.

1. Year 1: policy holds unemployment at 4 percent. $\pi = 2 - 0.5(4 - 6) = 2 + 1 = 3$ percent.

2. Year 2: expectations catch up. $\pi^e = 3$, so at $u = 4$, $\pi = 3 + 1 = 4$ percent.

3. Year 3. $\pi^e = 4$, so $\pi = 5$ percent. Each year below the natural rate adds a point of inflation: the trade-off is between unemployment and the change in inflation.

4. Disinflation. To bring inflation from 5 back to 2 percent, policy holds $u = 8$: inflation falls by $0.5 \times 2 = 1$ point a year, so it takes three years and $3 \times 2 = 6$ point-years of unemployment above the natural rate, 2 point-years for each point of inflation removed.

5. Fully anticipated policy. If expectations adjusted at once to any announced inflation, $\pi = \pi^e$ always and $u = 6$: policy could not hold unemployment below the natural rate even for a year.

Policy implications

The right response depends on the source of the unemployment:

• Frictional and structural unemployment (the natural rate) responds to labour market policy: job matching, retraining, help with moving, and the design of unemployment insurance. Stimulating demand pushes unemployment below the natural rate only temporarily, at the cost of rising inflation.
• Cyclical unemployment from sticky wages can be reduced by monetary and fiscal policy that restores demand.
• Intertemporal substitution implies fluctuations are efficient choices, leaving no case for stabilization.
• Efficiency wages imply unemployment that persists even when the economy is at full strength: it is part of the natural rate.
• Sectoral shifts call for retraining and mobility rather than demand stimulus.

Common mistakes

Traps on exams

• Treating the Phillips curve as a permanent menu. The trade-off holds only while inflation differs from expected inflation; in the long run the curve is vertical at the natural rate.
• Calling all recession unemployment involuntary. Under the intertemporal substitution hypothesis it is voluntary; under efficiency or sticky wages it is involuntary. Say which theory you are using.
• Confusing a movement along the Beveridge curve with a shift. Aggregate demand moves the economy along the curve; mismatch shifts it.
• Thinking efficiency wages mean firms overpay irrationally. Paying above the market-clearing wage maximizes profit when pay raises effort or lowers turnover.

On the exam

For the Phillips curve, draw a downward-sloping short-run curve labelled with its expected inflation, a vertical long-run curve at $u^*$, and the point where they cross at $\pi = \pi^e$. Show a second short-run curve shifted up after expectations rise. For involuntary unemployment, draw supply and demand, a wage line above the crossing, and mark the gap between labour supplied and demanded at that wage.

Check yourself

1.

From 2008 to 2009 the number of unemployed Canadians who had lost their jobs rose from 538,500 to 837,600, while the number who had left their jobs rose only from 274,000 to 292,400. Why is this a problem for the intertemporal substitution hypothesis?

1If workers were choosing leisure, unemployment would rise mostly through quits, not layoffs2It shows that wages fell sharply in the recession3It shows that labour supply is very elastic4It shows that job losers were discouraged searchers

Check answer

2.

True or false: in the expectations-augmented model, the long-run Phillips curve is vertical at the natural rate of unemployment.

1True2False

Check answer

3.

Labour demand falls in a recession. At the old wage of $20 an hour, firms now want 40 thousand workers, while 60 thousand people want to work. If the wage stays at $20, what happens?

1Employment is 40 thousand and 20 thousand people are involuntarily unemployed2Employment is 60 thousand, because labour supply has not changed3Employment is 50 thousand, halfway between demand and supply4The wage falls automatically until there is no unemployment

Check answer

Practise more questions on this topic

Sources

1. Statistics Canada, Table 14-10-0327-01, Labour force characteristics by gender and detailed age group, annual. ↩ ↩²

2. Statistics Canada, Table 14-10-0064-01, Employee wages by industry, annual. Average hourly wages of full-time employees aged 15 and over, current dollars. ↩

3. Lucas, R. E. and Rapping, L. A. (1969). Real wages, employment, and inflation. Journal of Political Economy, 77(5), 721 to 754. ↩

4. Altonji, J. G. (1986). Intertemporal substitution in labor supply: Evidence from micro data. Journal of Political Economy, 94(3, part 2), S176 to S215. ↩

5. Fehr, E. and Goette, L. (2007). Do workers work more if wages are high? Evidence from a randomized field experiment. American Economic Review, 97(1), 298 to 317. ↩

6. Statistics Canada, Table 14-10-0126-01, Reason for leaving job during previous year, annual. Unemployed people who left a job in the previous year, annual averages. ↩

7. Lilien, D. M. (1982). Sectoral shifts and cyclical unemployment. Journal of Political Economy, 90(4), 777 to 793. ↩

8. Abraham, K. G. and Katz, L. F. (1986). Cyclical unemployment: Sectoral shifts or aggregate disturbances? Journal of Political Economy, 94(3), 507 to 522. NBER working paper 1410. ↩

9. Shapiro, C. and Stiglitz, J. E. (1984). Equilibrium unemployment as a worker discipline device. American Economic Review, 74(3), 433 to 444. ↩

10. Raff, D. M. G. and Summers, L. H. (1987). Did Henry Ford pay efficiency wages? Journal of Labor Economics, 5(4, part 2), S57 to S86. NBER working paper 2101. ↩

11. Card, D. and Hyslop, D. (1997). Does inflation “grease the wheels of the labor market”? In C. D. Romer and D. H. Romer (eds.), Reducing Inflation: Motivation and Strategy, University of Chicago Press. NBER working paper 5538. ↩

12. Phillips, A. W. (1958). The relation between unemployment and the rate of change of money wage rates in the United Kingdom, 1861 to 1957. Economica, 25(100), 283 to 299. ↩

13. Friedman, M. (1968). The role of monetary policy. American Economic Review, 58(1), 1 to 17. Phelps, E. S. (1967). Phillips curves, expectations of inflation and optimal unemployment over time. Economica, 34(135). Phelps, E. S. (1968). Money-wage dynamics and labor-market equilibrium. Journal of Political Economy, 76(4, part 2), 678 to 711. ↩

14. Bank of Canada, Inflation-control target. ↩