Job amenities, health benefits and layoffs

A job amenity is a non-wage characteristic workers value, such as a flexible schedule, employer-paid health benefits or steady work; a disamenity is one they dislike, such as night shifts or predictable layoffs. Holding a worker’s productivity fixed, jobs with more amenities pay lower wages. Estimated differentials often have the opposite sign, because productivity and tastes are hard to hold fixed in data.

Amenities as part of compensation

A job offers a package: a wage plus amenities. Workers compare packages, and firms can pay in either currency. A firm with a pleasant workplace, a predictable schedule or generous benefits can attract workers at a lower wage; a job with disamenities, such as rotating night shifts, long commutes or weeks in a remote work camp, must pay more. The logic is the one behind compensating differentials for risk: the differential compensates the marginal worker.

Taxes tilt the mix. Wages are taxed; many amenities are not. In Canada an employer’s contributions to a private health services plan are not a taxable benefit for the employee.¹ A higher income tax rate lowers the after-tax value of a dollar of wages but not of an untaxed benefit, so workers shift toward amenity-rich packages.

Why estimated differentials often have the wrong sign

Apart from the risk of death, the evidence that bad jobs pay more is weak: wrong-signed or insignificant estimates are common.² Two reasons:

1. Unobserved ability. A more productive worker generates more value and receives more total compensation, which she takes partly as wages and partly as amenities. If the data measure productivity poorly, high wages and good amenities appear together, even though for any one worker more amenities cost wages.
2. Preferences of the marginal worker. Heavy outdoor work looks unpleasant to most people. But if enough workers enjoy it to fill those jobs, the marginal worker needs little or no premium. The premium depends on the tastes of the people who take the job, not of the average person.

The standard fix is to follow the same workers as they change jobs. A worker’s ability and tastes move with her, so the change in her wage when the job’s characteristics change isolates the differential. The method helps but is no cure: job changes can themselves be driven by things the data miss, and differencing magnifies measurement error. An early study using US panel data on young men found that the estimates did not become markedly more plausible, though the premium for fatal risk was positive and significant.²³

Health benefits

Put annual health benefits $H$ on the horizontal axis and the annual wage $w$ on the vertical axis. A dollar of benefits costs the firm a dollar, so an isoprofit line for a given worker has slope $-1$. Workers value both wages and benefits, so their indifference curves slope down and are convex.

Extra productivity of B^* ($ thousands per year)

$10.0k

Wage gap, B^* minus B

$8.0k

Benefit gap, B^* minus B

$2.0k

True trade-off on one isoprofit line (wage per $1 of benefits)

−$1.00

Open in Graph Lab

Figure 1. Equally productive workers A and B face the same isoprofit line, along which each $1 of benefits costs $1 of wages: A takes all pay as wages and B takes $5,000 of benefits and a $45,000 wage. B*, $10,000 more productive, takes $7,000 of benefits and a $53,000 wage, so comparing B with B* suggests that benefits come with higher pay.

• Worker A, for example young and healthy, values coverage little. His indifference curve is flatter than the isoprofit line at the no-benefits corner, so he takes the highest wage and no benefits.
• Worker B values coverage and chooses the tangency, with benefits $H_B$ and a lower wage $w_B$. A and B together reveal the true trade-off: one dollar of wages per dollar of benefits.
• Worker B* has B’s tastes but higher productivity, so firms can offer B* more of everything at zero profit. B* sits on a higher isoprofit line with both a higher wage and more benefits.

Comparing B with B* shows wages and benefits rising together. That positive correlation comes from productivity, not from the trade-off, and it is the empirical problem in miniature: unless productivity is well measured, the data can reject a theory that is true.

Key idea

Along one isoprofit line, amenities are paid for with lower wages. Across isoprofit lines, more productive workers get more of both, so raw data can show the opposite sign.

Layoffs and seasonal work

Steady work is also an amenity. Smith observed that a mason or bricklayer “can work neither in hard frost nor in foul weather”, so what he earns while employed “must not only maintain him while he is idle, but make him some compensation for those anxious and desponding moments”.⁴

Use the work-leisure model. A worker with total time $T$ who can choose her hours at wage $w_0$ picks $h_0$ hours, where her indifference curve $U_0$ is tangent to the budget line (point P). A seasonal job offers the same work but, because of layoffs known in advance, only $h_1 < h_0$ hours a year. At wage $w_0$ that bundle is on her budget line but was not chosen, so it is on a lower indifference curve. To attract her, the seasonal job must pay a higher wage $w_1$ that puts the point with $h_1$ hours back on $U_0$.

Hours in the seasonal job, h₁ (thousands)

1.50

EI replacement rate during layoffs

0%

Wage the seasonal job must pay, w₁

$31.74

Compensating differential, w₁ − w₀

$1.74

EI benefits during layoffs

$0.0k

Open in Graph Lab

Figure 2. A worker who chooses 2,000 hours a year at $30 an hour (P) would be worse off in a seasonal job offering 1,500 hours at the same wage. Without employment insurance the seasonal job must pay $31.74 to keep her on $U_0$; with EI replacing 55 percent of lost earnings during layoffs it needs to pay only $26.82.

Two results follow. The seasonal wage must exceed $w_0$, because the hours constraint binds. It need not keep annual earnings constant, because the worker gets some value from the extra leisure. The argument assumes the layoffs are anticipated. Layoffs that come as a surprise impose risk rather than predictable idleness, and the model does not price them in the same way.

Employment insurance. Canada’s EI pays 55 percent of average insurable weekly earnings, up to $729 a week in 2026, for 14 to 45 weeks; qualifying takes 420 to 700 hours of insurable work, depending on the regional unemployment rate.⁵⁶ Self-employed fishers have their own EI fishing benefits.⁷ Benefits paid during an expected layoff do part of the work the higher wage would otherwise do, so the compensating differential for seasonal work shrinks. Premium rates are set nationally, in 2026 $1.63 per $100 of insurable earnings for employees ($1.30 in Quebec) and 1.4 times that for employers, and do not depend on an employer’s layoff record.⁸ EI therefore replaces part of a private arrangement between seasonal employers and their workers with a public one financed in part by stable employers and their workers. See job search and employment insurance.

Worked example

1. The wage-benefit trade-off. The isoprofit line for workers A and B is $w + H = 50{,}000$. A chooses $H = 0$ and $w = 50{,}000$; B chooses $H = 5{,}000$ and $w = 45{,}000$. Between them, $\Delta w / \Delta H = -5{,}000 / 5{,}000 = -1$.

2. The productivity trap. B* is $10,000 more productive: $w + H = 60{,}000$. B* chooses $H = 7{,}000$ and $w = 53{,}000$. A researcher comparing B and B* finds $\Delta w / \Delta H = 8{,}000 / 2{,}000 = +4$, the wrong sign.

3. Bounds on the seasonal wage. A worker chooses 2,000 hours at $30 an hour, earning $60,000. A seasonal job offers 1,500 hours. It must pay more than $30, or she is worse off, and less than $60{,}000 / 1{,}500 = 40$ dollars, at which she would have the same earnings and 500 more hours of leisure.

4. The exact wage. Let utility be $U = C^{0.4} L^{0.6}$ with $T = 5{,}000$ hours. With no other income she works 40 percent of her time, 2,000 hours, and $U_0 = 60{,}000^{0.4} \times 3{,}000^{0.6}$. In the seasonal job she has 3,500 hours of leisure and needs consumption $C_1$ with $C_1^{0.4} \times 3{,}500^{0.6} = U_0$, so $C_1 = 60{,}000 \times (3{,}000 / 3{,}500)^{0.6/0.4} = 60{,}000 \times 0.7936 = 47{,}614$ dollars. So $w_1 = 47{,}614 / 1{,}500 = 31.74$ dollars: a compensating differential of $1.74 an hour.

5. Adding EI. If EI replaces 55 percent of the earnings lost during the 500 hours of layoff, income is $1{,}500 w_1 + 0.55 \times 500 \times w_1 = 1{,}775 w_1$. Setting this equal to $47,614 gives $w_1 = 26.82$ dollars, below the steady job’s wage.

Exercise: a longer layoff

With only 1,000 hours of work and no EI, $C_1 = 60{,}000 \times (3{,}000/4{,}000)^{1.5} = 38{,}971$ dollars, so $w_1 = 38.97$ dollars: a differential of $8.97. Longer anticipated layoffs need larger differentials, rising more than in proportion.

Evidence

Valuing amenities directly. Mas and Pallais built the choice into the hiring process of a national call centre, offering applicants positions that differed in schedule and pay. Most applicants would pay nothing for flexibility, but a minority valued it highly. On average, applicants would give up 20 percent of wages to avoid a schedule set by the employer on short notice and 8 percent for the option to work from home.⁹ Because the researchers designed the choices, the estimates do not depend on comparing workers of different ability.

Danger. A study of Swedish men that used workers’ own reports of whether their job was dangerous found a significant premium of about 2 percent of wages for dangerous work.³

Layoffs. Abowd and Ashenfelter estimated compensating differentials for anticipated unemployment ranging from less than 1 percent in industries with little anticipated unemployment to over 14 percent in industries with substantial anticipated unemployment and unemployment risk.¹⁰

Common mistakes

Traps on exams

• “Good jobs pay more, so the theory fails.” The theory holds productivity fixed. Comparing different workers mixes productivity with the trade-off.
• An upward-sloping isoprofit line for benefits. In wage-benefit space it slopes down: benefits are paid for with wages. (In wage-risk space isoprofit curves slope up, because risk is a bad.)
• The seasonal wage keeps annual earnings constant. It lies between the steady wage and the earnings-equivalent wage, because extra leisure has value.
• Forgetting the assumption. The layoff differential is for anticipated layoffs. EI benefits reduce it.

On the exam

For health benefits, label the axes (benefits, wage), draw isoprofit lines of slope minus one, show A at the corner and B at a tangency, then add B* on a higher line and explain why B and B* together give a positive slope. For layoffs, draw the work-leisure diagram, mark the hours cap at $T - h_1$, the lower indifference curve at the old wage and the steeper budget line with wage $w_1$ through $U_0$.

Check yourself

1.

Across a large sample of workers, jobs with generous benefits also tend to pay higher wages. Within the compensating-differentials framework, what is the likeliest explanation?

1More productive workers earn more total compensation and take some of it as wages and some as benefits2Firms pass the cost of benefits on to their workers as higher wages3Benefits are taxed, so firms raise wages to cover the tax4Compensating differentials only exist for the risk of death

Check answer

2.

Workers B and B* have the same preferences, but B* is more productive. B gets a $45,000 wage and $5,000 of health benefits; B* gets a $53,000 wage and $7,000 of benefits. What explains the positive relationship between wages and benefits in these two observations?

Extra productivity of B^* ($ thousands per year)

$10.0k

Wage gap, B^* minus B

$8.0k

Benefit gap, B^* minus B

$2.0k

True trade-off on one isoprofit line (wage per $1 of benefits)

−$1.00

1B* is on a higher isoprofit line, so the comparison mixes higher productivity with the wage-benefit trade-off2Benefits make B* more productive, which raises the wage3B* values benefits more than B does4The isoprofit line slopes upward, so firms pay more to workers who take more benefits

Check answer

3.

A worker who freely chooses 2,000 hours a year at $25 an hour is offered a seasonal job with predictable layoffs: same work, but only 1,600 hours a year. To attract her, what hourly wage must the seasonal job pay?

1More than $25 but less than $31.252Exactly $31.25, so that her annual earnings are unchanged3Exactly $25, because the extra leisure exactly makes up for the lost hours4Less than $25, because she gains 400 hours of leisure

Check answer

Practise more questions on this topic

Sources

1. Canada Revenue Agency, Employer’s guide: Taxable benefits and allowances (T4130), section on private health services plans. ↩

2. Brown, C. (1980). Equalizing differences in the labor market. Quarterly Journal of Economics, 94(1), 113 to 134. ↩ ↩²

3. Viscusi, W. K. and Aldy, J. E. (2003). The value of a statistical life: A critical review of market estimates throughout the world. Journal of Risk and Uncertainty, 27(1), 5 to 76, which summarizes Brown (1980) and Duncan, G. J. and Holmlund, B. (1983), Was Adam Smith right after all? Another test of the theory of compensating wage differentials, Journal of Labor Economics, 1(4), 366 to 379. NBER working paper 9487. ↩ ↩²

4. Smith, A. (1776). An Inquiry into the Nature and Causes of the Wealth of Nations, Book I, chapter 10. Project Gutenberg edition. ↩

5. Government of Canada, EI regular benefits: How much you could receive. ↩

6. Government of Canada, EI regular benefits: Eligibility. ↩

7. Government of Canada, EI fishing benefits. ↩

8. Canada Revenue Agency, EI premium rates and maximums. ↩

9. Mas, A. and Pallais, A. (2017). Valuing alternative work arrangements. American Economic Review, 107(12), 3722 to 3759. ↩

10. Abowd, J. M. and Ashenfelter, O. (1981). Anticipated unemployment, temporary layoffs, and compensating wage differentials. In S. Rosen (ed.), Studies in Labor Markets. University of Chicago Press. NBER chapter page. ↩