Reducing Working Hours: American Workers' Salvation?
by Terry J. Fitzgerald
September 1, 1996

There's no question that the long-term salvation of work lies in reducing working hours.
—Thomas R. Donahue
Former Secretary–Treasurer of the AFL–CIO1


In March 1991—the trough of the most recent recession—the civilian unemployment rate stood at 6.8 percent. Fifteen months into the recovery, that rate had increased to 7.8 percent, leading many in the media to decry the recovery as "jobless."

Over this same period, many observers noted that as the unemployment rate was rising, so too was Americans’ average workweek. Average weekly hours in the manufacturing sector rose by almost one hour, while average overtime increased by more than 30 minutes.

These facts suggest that rather than hiring additional workers, firms were choosing to have their current employees work longer hours. The Full Employment Act of Fiscal Year 1994, introduced in the House of Representatives but never passed into law, sought to counter this trend by modifying the provisions of the Fair Labor Standards Act of 1938 (FLSA). In particular, it would have reduced the standard workweek from 40 hours to 30 hours and raised the overtime premium from one and a half to two times the standard wage.

Proposals intended to cut working hours and thereby boost employment, sometimes referred to as work-sharing or work-spreading, are not uncommon. Work-sharing legislation was introduced in the U.S. Congress in 1978 and 1985, and similar proposals have been hotly debated over the last two decades in many European countries.

While none of these proposals has become law in the United States, the work-sharing movement is likely to gain momentum if (or more realistically, when) the jobless rate begins to rise again. This Economic Commentary describes the basic rationale behind policies intended to "spread the work," and looks at some of the reasons why they are likely to be less effective at boosting employment than proponents claim.

Return to Index

The Argument for Reducing Hours

Proposals to cut working hours are largely motivated by two observations. First, during any given week millions of people are unable to find work. Second, during the same week millions of others work more than the standard 40 hours. For example, while 9.7 million Americans were unemployed in March 1992, 33.6 million others reported working more than 40 hours per week.2 Why, then, don’t we change our labor market policies to discourage firms from relying on long working hours and thereby encourage them to hire additional workers?

Adding further momentum to the push for shorter hours is the fact that working hours in the United States have remained largely unchanged over the past 35 years, while many industrialized countries have experienced a substantial decline. Figure 1 shows that between 1960 and 1994, U.S. manufacturing workers went from putting in the fewest number of hours per year to the most hours relative to their counterparts in Japan, Germany, France, and the United Kingdom. This fact is sometimes interpreted as a "failure" of U.S. labor markets—that somehow America has fallen behind in the race for shorter hours. Proponents of cutting hours argue that such a policy would not only increase employment, but would also help us catch up with the trend in the rest of the industrialized world.

While specific proposals to reduce working hours vary, the basic intuition behind them is fairly simple: There is some total number of hours to be worked, and society would benefit by spreading these hours across more people. The resulting drop in the jobless rate would reduce government spending on unemployment insurance and on a variety of welfare programs that provide assistance to poor people unable to find work.

To illustrate this view more concretely, consider a policy that restricts the workweek to 40 hours. If we assume that the number of hours worked in the economy remains unchanged, then the hours left unworked by the 33.6 million Americans who put in more than 40 hours per week in March 1992 would have created enough new 40-hour jobs to put all of the 9.7 million unemployed to work.

Return to Index

Is an Hour Worked an Hour Worked?

While there are many reasons to be skeptical about the huge employment effects in the example cited above, I will highlight just a few of them. To begin with, an implicit assumption made by advocates of work-sharing is that an hour worked is an hour worked, regardless of who does the work. That is, the work currently being done by those putting in long hours could be performed just as well by the unemployed.

This hours substitution would require the skills of the unemployed to be similar to the skills of those who work long hours. For example, if those putting in long hours are plumbers and electricians, then reducing hours would create jobs for unemployed plumbers and electricians, but not for unemployed accountants.

Table 1 uses three broad measures—education, age (which is strongly correlated with work experience), and occupation—to illustrate how the skills of the unemployed compare with the skills of those working long hours (hereafter referred to as long-hour workers). The characteristics of the 9.7 million Americans who were jobless in March 1992 are compared with the roughly 34 million who put in more than 40 hours per week that same month.


Unemployed Worked more than 40 hours
No high school diploma 30.8 8.2
High school diploma 38.9 31.2
Some college 21.2 25.6
Bachelor’s degree 7.0 21.7
Postgraduate degree 2.1 13.2
24 or younger 28.5 7.8
25–34 29.7 31.0
35–44 21.3 31.2
45–54 12.6 20.0
55 or older 8.0 10.1
Executive, administrative, managerial 5.6 20.4
Professional specialty 4.1 17.0
Service, except protective 13.6 5.9
Precision, production, craft, repair 16.6 11.6
Handlers, equipment cleaners, helpers, laborers 9.4 2.6

SOURCES: Author’s calculations; and U.S. Department of Labor, Bureau of Labor Statistics, Current Population Survey, March 1992.
Although these categories are quite broad, it is apparent that the pool of unemployed have very different characteristics than the population of long-hour workers. For example, almost 70 percent of the unemployed had only a high school education or less, while for long-hour workers that figure was below 40 percent. In contrast, only about 9 percent of the unemployed had earned a a bachelor’s degree or done postgraduate work, versus 35 percent for long-hour workers.

Looking at age distributions, those under 25 made up 28.5 percent of the unemployed, but only 7.8 percent of the long-hour group. People in the prime of their working lives, ages 35 to 54, accounted for more than 50 percent of long-hour workers, but only 34 percent of the unemployed.

There are also notable differences in the occupational breakdown of the two groups. Most striking is that executives, administrators/managers, and professionals made up about 37 percent of long-hour workers, but only 9.7 percent of the unemployed. Alternatively, out-of-work handlers, equipment cleaners, helpers, and laborers accounted for 9.4 percent of the total unemployed, but only 2.6 percent of the long-hour workers.

Thus, when it comes to work-sharing policies, an hour worked by long-hour workers is not the same as an hour worked by the unemployed, who as a group possess notably different skills. Looking at more specific measures of skills would almost certainly magnify these differences. This mismatch is not only likely to limit the employment effects of reducing hours, but could also lead to reduced productivity and output.

Return to Index

Policies Can Have Surprising Effects

A second reason to be skeptical about work-sharing policies is that it is very difficult to predict how policy changes will affect the employment decisions of firms and workers. If, for example, abiding by a new policy is costly to firms or their employees, we would expect them to devise ways of evading that policy.3 Even if a policy results in a shorter workweek, this may simply lead to an increase in moonlighting as those workers whose hours have been cut find second jobs to maintain their income.

As an example of the potential difficulty in predicting a policy’s effects, consider one of the provisions of the Full Employment Act: amending the FLSA by decreasing the standard workweek (after which firms must pay an overtime premium) from 40 hours to 30 hours. The intent of this provision, in conjunction with other provisions of the bill, was to encourage firms to hire additional workers and decrease the number of hours per worker.

Let’s examine the possible effects of reducing standard hours using a simple example. Suppose a firm can hire workers for $10 an hour up to 40 hours, then must pay $15 an hour for any additional time on the job. Furthermore, assume that the firm must pay $190 in fixed costs per worker (these costs may include fringe benefits, and hiring and training expenses). For simplicity, the firm can choose among three options, each of which leads to the same total employee hours (400) and the same output:

  1.  Employ 12 workers for 3-31/3 hr.
  2.  Employ 10 workers for 40 hr.
  3.  Employ 8 workers for 50 hr.
The cost of each of these options (cost per worker x number of workers) is
  1.  $523.33 x 12 = $6,279.96
  2.  $590.00 x 10 = $5,900
  3.  $740.00 x 8    = $5,920.4
Clearly the firm will choose option 2 and employ 10 workers.

Now consider lowering standard hours from 40 to 30 and assume that the base wage rate and fixed costs remain unchanged. The cost of the three options becomes

  1.  $540.00 x 12  =  $6,480
  2.  $640.00 x 10  =  $6,400
  3.  $790.00 x 8    =  $6,320.5
The firm will now choose to employ eight workers for 50 hours rather than 10 workers for 40 hours. Thus, reducing standard hours leads to an increase in hours per worker and a decrease in employment, exactly the opposite effect of what was intended.

Obviously, forecasting the impact of reducing standard hours is much more complicated than this simple example suggests. Wages, output, and productivity are likely to change when a new policy is implemented. Furthermore, increases in the overtime premium, another provision of the Full Employment Act, could lead to shorter hours and higher employment. However, these calculations do show that well-intended policies can have surprising and undesirable results.

Return to Index

Shorter Hours and Employment Growth

So far, I have simply argued that work-sharing policies are unlikely to boost employment as much as some proponents suggest. But one might reasonably ask, have the countries that experienced sharp declines in annual hours since 1960 also experienced large increases in employment?

Table 2 shows the changes in the employment/population ratios that accompanied the changes in average hours worked shown in figure 1.6 Of the five countries listed, the United States is the only one whose hours increased between 1960 and 1993. We are also the only nation whose employment/population ratio rose over this period. While these data do not necessarily reflect the employment effect of any work-sharing-type policies, they do show that declining hours do not necessarily lead to increased employment.

(Percent change, 1960–1993)

Average annual hours in manufacturing Employment/population ratio
United States 1.7 9.9
Japan –20.9 –7.5
Germany –26.9 –13.7
France –17.9 –16.2
United Kingdom –15.0 –7.3

SOURCE: U.S. Department of Labor, Bureau of Labor Statistics, Monthly Labor Review (various issues), and International Comparisons of Manufacturing Productivity and Unit Labor Cost Trends, 1994.


SOURCE: U.S. Department of Commerce, Bureau of Labor Statistics, International Comparisons of Manufacturing Productivity and Unit Labor Cost Trends, 1994.

Return to Index

Working Hours: Are We "Falling Behind"?

One of the motivations for reducing American workers’ hours is to catch up with the rest of the industrialized world. But in what sense has a country with longer work hours fallen behind?

In every country, workers face a trade-off between how many hours they spend earning income, working at home (for example, cleaning and doing yard work), and pursuing leisure activities. Labor market regulations, tax rates, relative wage rates, along with many other factors, influence how workers decide to allocate their time.

If after-tax wage rates in one country are high relative to the cost of buying household services, workers may decide to spend a few additional hours on the job and hire someone to clean their house and cut their lawn. Does that make these people worse off?7

In another country, the marginal tax rate on labor income may be very high, discouraging workers from spending many hours earning income. Are these people necessarily better off than workers in a low-tax-rate country where people choose to work longer hours and consume more goods?

If people in two different countries were consuming the same amounts of all goods and services, the country whose workers put in fewer hours would obviously be the better off of the two. However, if people in one of the countries consume larger amounts of goods and services but work more hours, it is not obvious which nation’s people are better off. What is clear is that judging whether people in one country are better off than those in another based solely on their working hours is silly.

Return to Index


Although the impetus for work-sharing policies lost steam when employment levels began to pick up in 1993, one can expect the issue to rise again when the economy next begins to slow and the unemployment rate begins to climb. Furthermore, given the trends in working hours both in the United States and abroad, the motivation for policies that would shorten working hours is likely to persist. While cutting hours to boost employment may have some intuitive appeal, work-sharing policies are unlikely to provide the large employment gains that their advocates promise, and may result in lower output and productivity. Rather than being the "long-term salvation" of work, shorter working hours may have an effect more closely captured by another union official:
[Shorter workweeks] just spread the level of misery.
—Vernon Watkins
Official of the American Federation
of State, County, and Municipal Employees?8

Return to Index


1. See "Labor Wants Shorter Hours to Make Up for Job Losses," The New York Times, October 11, 1993.
Back to commentary

2. Data are from the April 1992 issue of Employment and Earnings, U.S. Department of Labor, Bureau of Labor Statistics, and apply to a particular week during that month.
Back to commentary

3. For example, noncompliance with FLSA has been estimated to be around 10 percent. See Stephen J. Trejo, "The Effects of Overtime Pay Regulation on Worker Compensation," American Economic Review, vol. 81, no. 4 (September 1991), pp. 719–40.
Back to commentary

4. $523.33 = ($10 x 3-31/3 + $190);
$590.00 = ($10 x 40 + $190);
$740.00 = ($10 x 40 + $15 x 10 + $190).
Back to commentary

5. $540.00 = ($10 x 30 + $15 x 3-1/3 + $190);
$640.00 = ($10 x 30 + $15 x 10 + $190);
$790.00 = ($10 x 30 + $15 x 20 + $190).
Back to commentary

6. While table 2 contains hours data only for manufacturing workers, data on hours worked by all workers also show a much smaller decline in the United States than in the other countries listed.
Back to commentary

7. See Kristin Roberts and Peter Rupert, "The Myth of the Overworked American," Federal Reserve Bank of Cleveland, Economic Commentary, January 15, 1995.
Back to commentary

8. See "Labor Wants Shorter Hours to Make Up for Job Losses" (footnote 1).
Back to commentary

Return to Index

Terry J. Fitzgerald is an economist at the Federal Reserve Bank of Cleveland.

The views stated herein are those of the author and not necessarily those of the Federal Reserve Bank of Cleveland or of the Board of Governors of the Federal Reserve System.