Lane Kenworthy, The Good Society
July 2017

Employment isn’t always a good thing. The need for a paycheck can trap people in careers that divert them from more productive or rewarding pursuits. Paid work can be physically or emotionally stressful. It can be monotonous, boring, alienating. Some jobs require a degree of indifference, meanness, or dishonesty toward customers or subordinates that eats away at one’s humanity. And work can interfere with family life.

Yet employment has significant virtues.1 It imposes regularity and discipline on people’s lives. It can be a source of mental stimulation. It helps to fulfill the widespread desire to contribute to, and be integrated in, the larger society. It shapes identity and can boost self-esteem. With neighborhood and family ties weakening, the office or factory can be a key site of social interaction. Lack of employment tends to be associated with feelings of social exclusion, discouragement, boredom, and unhappiness.

Societies also need a significant majority of people in paid work to help fund government programs. In countries that have made commitments to pensions for their elderly, health care for all, and assorted other services and transfers, there is a need for additional government revenue. Some can come from raising tax rates, but that has become a tall order in a world with mobile capital. Increasing the share of the population in paid work can help to ensure the fiscal viability of a generous welfare state. It provides an increase in tax revenues without requiring an increase in tax rates. High employment eases the fiscal crunch another way too, by reducing the number of people fully or heavily reliant on government benefits.

Employment is front and center in the American ethos. Self-sufficiency and self-realization via work are at the core of the “Protestant ethic” that shaped the country in its early years and the “American dream” that has animated it since the mid-1800s.

How successful have we been in getting Americans into paid work? Are too few men working now? Is it possible for countries to combine egalitarianism with high employment? Will robots soon take all of the jobs?


In the second half of the twentieth century, America’s employment rate — the share of working-age persons in a paying job — rose steadily. In the 1980s and 1990s the increase stood out among affluent democratic countries, a number of which were experiencing little job growth and persistently high levels of joblessness. Commentators took to referring to our economy as the “great American jobs machine.”2

But as figure 1 shows, around 2000 that ended. The employment rate as of 2016 was about 4 percentage points lower than in 2000 and nearly 9 percentage points lower than it would have been if the preceding half century’s trend had continued.

Figure 1. US employment rate
Employed persons aged 25-64 as a share of the population aged 25-64. The vertical axis doesn’t begin at zero. These data are from surveys of households. Surveys of establishments suggest a similar pattern over time. Data source: Current Population Survey, Bureau of Labor Statistics, series LNU02000060, LNU02000095, LNU00000060, LNU00000095.

Some would argue that a better measure is employment hours, calculated as the employment rate multiplied by average annual hours worked among those employed. That’s probably wrong if our concern is with individual well-being; people may derive meaning and satisfaction simply from being employed, even if they don’t work many hours. In terms of generating tax revenue, on the other hand, it’s of limited help to have lots of people in paid employment but working few hours. In any event, the long-run pattern for employment hours is broadly similar to that for the employment rate, as figure 2 shows.

Figure 2. US employment hours
Employment hours per working-aged person: average hours worked per employed person multiplied by the employment rate for persons age 25-64. “k” = thousand. The vertical axis doesn’t begin at zero. Data source: OECD.

Our recent employment record isn’t just disappointing compared to our past performance. It also is mediocre relative to how other affluent nations have done since 2000, as figure 3 indicates. The US employment rate was among the highest at the turn of the century, but it’s now toward the low end of the ranking.

Figure 3. Employment rate
Employed persons aged 25-64 as a share of all persons aged 25-64. The vertical axis doesn’t begin at zero. Data source: OECD. “Asl” is Australia; “Aus” is Austria.

The US is missing about 9 percentage points of employment. That’s the gap between the actual employment rate in 2016 and what the rate would have been if the previous half century’s trajectory had continued (figure 1). As a rough approximation, about 1 percentage point of that gap owes to a drop in government employment, 1 point to an acceleration in the decline of manufacturing employment, 3.5 points to the timing and severity of the great recession, and 3.5 to a slowdown in growth of private service employment during economic upturns.

As figure 4 shows, the share of the working-age population employed by federal, state, and local governments increased until 1980. It then flattened out in the 1980s and 1990s. Since 2000 it has dropped by about 1 percentage point.

Figure 4. US employment rate: government
Government employees as share of the population aged 16-64. Data sources: Federal Reserve Bank of St. Louis, FRED database, series, usgovt; Bureau of Labor Statistics, series LNU00024887, LNU00000060, LNU00000095. In figures 4, 5, and 6, the employment data come from a survey of establishments rather than of households. The establishment employment data aren’t broken down by age, which makes it impossible to isolate 25-to-64-year-olds in the numerator. I therefore use the population aged 16 to 64 as the denominator in calculating employment rates.

The share of working-age Americans employed in manufacturing has been falling since 1970, as figure 5 indicates. The decline accelerated after 2000. In the 30 years from 1970 to 2000, the share fell by 6 percentage points, or about 3 percentage points every 15 years. In the 15 years after 2000, it fell by 4 percentage points.

This owes largely to a surge in manufacturing imports from China, which cost Americans 2.4 to 3.2 million jobs according to recent analyses.3 In principle those workers could have shifted to other jobs, but in practice that can be difficult unless the displaced worker is willing to accept significantly lower pay or move to another city or state.4

Figure 5. US employment rate: manufacturing
Manufacturing employees as share of the population aged 16-64. Data sources: Federal Reserve Bank of St. Louis, FRED database, series manemp; Bureau of Labor Statistics, series LNU00024887, LNU00000060, LNU00000095.

Figure 6 shows the share of the working-age population in private-sector nonmanufacturing — mostly service — jobs. From 1970 to 2000, as the share of working-age Americans employed by government held steady and the share employed in manufacturing declined, private service employment grew rapidly, accounting for all of the rise in the overall employment rate. Since 2000, by contrast, we’ve gained no ground in private service employment.

Figure 6. US employment rate: private sector other than manufacturing
Private-sector employees not in manufacturing as share of the population aged 16-64. Data sources: Federal Reserve Bank of St. Louis, FRED database, series uspriv, manemp; Bureau of Labor Statistics, series LNU00024887, LNU00000060, LNU00000095.

The most obvious culprit, readily apparent in figure 6 (also figure 1), is the timing and severity of the “great recession.” The onset of the recession in 2008 meant that the employment growth phase of the 2000s business cycle lasted just four years, from 2003 to 2007, and the recession then knocked the employment rate down by far more than in any other downturn since the 1930s.

But this isn’t the entire story. In addition, service employment expansion during upturns has slowed. In the growth phase of the business cycles during the 1970s, 1980s, and 1990s, the employment rate in private-sector services increased, on average, by 1 percentage point per year. In the 2000s and the 2010s, it did so by just 0.67 percentage points per year.

What accounts for this? One hypothesis points to a slowdown in economic growth since 2000.5 This is quite plausible, but sorting out the causality is difficult, because employment is not only a consequence of economic growth but also a key determinant of it.

Second, a significant amount of net job creation comes from relatively young firms, and the rate of new business creation is lower today than in the past, perhaps due to an accumulation of regulatory impediments.6 The problem with this explanation is that the business startup rate has been falling steadily since the mid-to-late 1970s, so it isn’t clear why this would have hindered employment growth only after 2000.7

A third hypothesis points to shifts in corporate governance and in the leverage of employers relative to employees.8 Beginning in the 1980s, many US firms shifted to a “shareholder value” orientation, prioritizing the company’s stock price over the well-being of employees. Growing use of stock options as a key component of executive compensation accentuated this shift by creating a direct financial incentive for top management to try to drive up the firm’s share price. At the same time, financial advisors, believing many firms had gotten overly bloated with unnecessary employees, increasingly favored those that “downsized” their workforce. Meanwhile, declining unionization, rising imports, automation, and an increase in immigration weakened workers’ ability to resist employment cutbacks during hard times and to press for more jobs during good times.

A fourth hypothesis focuses on America’s lack of progress in family-friendly policy.9 Employment among prime-working-age men has decreased since 1970, but at a steady pace, with no acceleration after 2000 (see below). The big change since 2000 has been among women. The employment rate among women of prime working age (age 25-54) rose from 35% in 1950 to 74% in 2000. By 2016 it had fallen to 71%.

This contrasts with developments in a number of other rich countries, as figure 7 shows. In 2000, the US had the sixth highest prime-age women’s employment rate among these countries. A decade and a half later, it ranked sixteenth.

Figure 7. Employment rate: women aged 25-54
Employed women aged 25-54 as a share of women aged 25-54. The vertical axis doesn’t begin at zero. Data source: OECD. “Asl” is Australia; “Aus” is Austria.

Many Americans with prekindergarten children want to combine family with paid work. But out-of-home care can be prohibitively expensive. Full-time out-of-home care for a child under age five costs, on average, about $9,500 per year.10 That’s a very large sum, particularly for those with low incomes. (The average income of the bottom fifth of US households is only $20,000.11) Faced with such unaffordable costs, some parents settle for care that is mediocre or poor, while others forgo employment.

Denmark and Sweden pioneered a different approach.12 Beginning in the 1960s, these countries introduced and then steadily expanded paid parental leave and publicly-funded childcare and preschool. Today, Swedish and Danish parents can take a paid year off work following the birth of a child. After that, parents can put the child in a public or licensed private early education center. The quality tends to be high, as early education teachers get training and pay comparable to elementary school teachers. Parents pay a fee, but the cost is capped at less than 10% of a household’s income.

We can see the impact in employment patterns. Among mothers whose youngest child is six to sixteen years old, and thus eligible for free K-12 schooling, the employment rate in the United States is just a few percentage points lower than in Denmark and Sweden. Among mothers with a child younger than six, it’s 15 percentage points lower.13

In recent decades many other rich nations have moved toward the types and generosity levels of the Nordic family-friendly policies, and a number of them have pulled ahead of the US in women’s employment. We’ve done far less: at the federal level, a requirement that medium-sized and large firms allow 12 weeks of unpaid family leave, enacted in 1993; among states, a few have paid parental leave policies and some have pre-kindergarten for 4-year-olds.


The employment rate among American men aged 25-54 has been falling for nearly half a century. As figure 8 shows, the rate was around 94% in the 1950s and 1960s. Since 1970 it has declined steadily, dropping to 85% as of 2016.

Figure 8. US employment rate: persons aged 25-54
Employed persons aged 25-54 as a share of the population aged 25-54. The vertical axis doesn’t begin at zero. The values on the vertical axis are the starting and ending levels for each of the three series. Data source: Bureau of Labor Statistics, series LNU02300060, LNU02300061, LNU02300062.

This is a product of multiple developments. In that same period, from 1970 to the present, the employment rate of prime-working-age women jumped from 48% to 71%. Job competition from women has thus played a role in the fall in men’s employment. So has complementarity: today, more men can forgo employment because they have a wife, partner, or ex-partner with a paycheck.

Because of automation, trade, the shareholder value orientation in corporate governance, union decline, and other developments, the labor market has become more difficult for and less attractive to less-educated Americans, with fewer decent-paying jobs, stagnant wages, and less job security.14 Men without college degrees, and even moreso those with no college at all, have seen the sharpest reduction in employment.

The shrinking of the manufacturing sector has also contributed. About 16% of all (male and female) working-age Americans had manufacturing jobs in the 1950s and 1960s, but since 1970 the share has fallen steadily, down to just 6% as of 2016. Men who used to work in manufacturing, or who grew up expecting to, have been less likely than women to take a service job instead, perhaps because they see service positions as inconsistent with traditional notions of masculinity.15 Metropolitan areas with the lowest prime-age male employment rates include a number of former manufacturing hubs.16

As a result of rising crime in the 1960s and 1970s and our incarceration-centered response to that rise, a large number of men have been saddled with the labor market obstacle of a prison record. According to one estimate, a third of nonemployed men aged 25 to 54 have a criminal record.17

Another contributor is the creation of new government benefits and expansion of some existing ones — disability benefits, unemployment insurance, food stamps, housing assistance, Medicaid, veterans benefits, and others. Working-age men who receive such benefits, or who live with someone who does, can make do without employment in a way their counterparts in the 1950s and the early 1960s couldn’t.

Another hypothesis says video games have become so enticing that they’ve encouraged more young employable men to stay home.18 Yet this seems unlikely to be a major driver, as the trend in prime-age male employment has been a steady downward one for nearly half a century, with no acceleration during the recent period of dramatic improvement in video game quality.19

According to some observers, the decline in male prime-working-age employment is a “catastrophic failure,” “a problem so urgent, so immense that it should demand immediate attention and action.”20 That’s probably an overreaction. During the period in which the decline has occurred, from 1970 to the present, the overall prime-working-age employment rate has risen from 70% to 78% (see figure 8), because the rise in women’s employment has more than compensated for the fall among men. Given these numbers, it isn’t clear that the fall for prime-age men — from 19 in 20 employed to 17 in 20 employed — should be high on our list of economic concerns.

In any case, we can’t turn back the clock. The rise in women’s participation in higher education and the workforce isn’t going to reverse itself. And as traditional gender norms fade, more men will want to spend more time at home caring for children or other family members. Manufacturing employment isn’t going to revitalize, even if the United States reduces imports from low-wage countries such as China and Mexico. The share of people in manufacturing jobs has been falling steadily not just here but in all affluent countries, and automation will ensure that this decline continues.21

We could and should take steps to improve pay levels in low-end jobs. We also need to do more to help former prisoners into paid work. But as figure 9 shows, the downward trend in employment among prime-working-age men has occurred in other rich nations too, including those with a higher wage floor and no incarceration problem. This suggests limits to how much America can achieve with these strategies.

Figure 9. Employment rate: men aged 25-54
Employed men aged 25-54 as a share of men aged 25-54. The vertical axis doesn’t begin at zero. “Asl” is Australia; “Aus” is Austria. Data source: OECD.


Since the early 1990s, policy makers have worried that there is a sharp tradeoff between high employment and low or modest inequality. A high wage floor and generous government benefits, in this view, reduce employer demand for workers at the low end of the labor market and reduce the incentive for benefit recipients to enter paid work.22 A decade ago the cross-country evidence suggested that egalitarian institutions and policies might indeed have some adverse impact on employment, albeit not a large one.23

Today there is greater cause for optimism. Figure 10 shows that countries with less income inequality between households at the bottom and those in the middle have tended to do as well or better than their more unequal counterparts in achieving a high and rising employment rate.

Figure 10. Employment by income inequality
1979 to 2015. In the first chart, the vertical axis is the employment rate for persons aged 25-64 in 2015. In the second chart, the vertical axis is change in the employment rate for persons aged 25-64 between 1979 and 2015, adjusted for starting level. Income inequality: ratio of 50th-percentile (median) household income to 10th-percentile household income, averaged over the period from 1979 to 2015. Data sources: OECD; Luxembourg Income Study. “Asl” is Australia; “Aus” is Austria.

It’s worth noting in particular the employment performance of the United States. The US has pursued a “market liberal” approach to employment growth: a low wage floor, very limited labor market regulations, relatively stingy government benefits, comparatively low taxes, steady deregulation of product markets, and limited support for retraining, job placement, and work-family balance. Up to the turn of the century the “great American jobs machine” was comparatively successful; the US was among the rich world’s leaders in raising its employment rate. But fifteen years later, America’s employment record looks quite mediocre.


At some point, artificial intelligence may get good enough that robots and other machines can do even nuanced analytical and in-person service tasks as well as (or better than) humans.24 At that point, we’ll almost certainly need a universal unconditional basic income at a level that ensures each person or household a reasonable level of consumption — perhaps $25,000 per adult and $10,000 per child.

But in all likelihood this is a generation or three away. Until then, there is good reason to presume that lots of us will have employment in various types of service positions — particularly in-person services.25

As we get richer, most of us are willing to outsource more tasks that we don’t have time or expertise or desire to do ourselves changing the oil in the car, mowing the lawn, cleaning, cooking, caring for children and other family members, advising, educating, organizing, managing, coaching, transporting. And improved productivity and lower costs abroad will reduce the price we pay for food, manufactured goods, and some services, leaving us with more disposable income. So we’ll want more people teaching preschool children, coaching and mentoring teenagers, helping adults find their way in the labor market or through a midlife career transition, caring for the elderly, and so on, and we’ll be better able to purchase and provide such services.

Some of these jobs are likely to pay a modest wage. But that need not mean a person’s income also is low. A subsidy such as America’s Earned Income Tax Credit (EITC) or the UK’s Universal Credit (formerly Working Tax Credit) can boost household incomes while simultaneously encouraging employment.26

For some, a low-end service job might be a career. Others will want it to be merely a stepping-stone. Government can help ensure that people have the capability to move up, via healthcare, early education, elementary and secondary schooling, lifelong learning opportunities, retraining, job placement assistance, special services for the mentally or physically disabled, language assistance for immigrants, targeted programs for the young and the elderly, assistance with transportation, and help in organizing formal job ladders.

Mobility between jobs need not be confined to upward moves. It’s very difficult to predict at age 18 or even 22 what kind of interests and capabilities you will have at age 35 or 50. Policy should facilitate people’s ability to change job, occupation, or entire line of work at various points in the life course, even if the switch is simply to something different, rather than something better. This calls for training, counseling, mentoring, and perhaps several sabbaticals (every adult, not just parents of newborn children, should have access to several one-year paid leaves).27 It also means eligibility for pensions, unemployment insurance, sickness insurance, parental leave, holidays, and other nonwage benefits should be contingent on employment, but not on the particular job or employer you have.

If most people are expected to be in employment, policy also ought to improve the quality of work life. Low-end service jobs may offer limited mental stimulation or opportunity to participate in decision making, and some are stressful. There is a limit to the amount of stimulation that some of these jobs will ever be able to provide, but most could do better, and we should try to figure out how and to push firms in that direction. Indeed, we should aim to improve working conditions in all jobs, rather than assuming that higher-skilled, better-paying positions automatically have decent work quality. One possibility is an auditing procedure whereby government sets outcome standards for work conditions, leaves it up to firms to decide how to meet the standards, and monitors their efforts to do so.28

Finally, policy ought to limit the degree to which job inequality spills over into social inequality and segregation. We want a society that is modestly rather than severely unequal. Jobs inevitably come with inequalities of status. If they also have profoundly unequal pay, this can easily spill over into social segregation and inequality of respect. Policy should push against this. Neighborhoods should be designed or redesigned to encourage class mixing. Parks, beaches, libraries, and public transport ought to be attractive to all. And we might do well to consider a mandatory year of national service to ensure that everyone gets an experience of genuine social mixing as they embark on adulthood.


The employment rate increased steadily in the United States from 1950 to 2000. Since then it has moved in the opposite direction, and the US has fallen from one of the top performers among the world’s rich countries to a relatively poor one. This turnaround is a product of multiple developments. Most notable, perhaps, is the fall in employment among women, which may owe chiefly to our failure to enact and expand family-friendly policies such as early education and paid parental leave.

The employment rate among prime-working-age men has fallen steadily since 1970, from 94% to 85%. Some view this as a crisis. But the US experience is similar to that of nearly all affluent democracies, and in any case there is no going back to the institutions and policies that contributed to the very high male employment rate of half a century ago.

A decade ago conventional wisdom held that policies and institutions conducive to modest inequality — generous government benefits, high wage floors, heavy taxation, and more — reduce employment. Today that apparent tradeoff seems to have evaporated.

At some point robots and computers will do many of the job tasks currently performed by humans. But that is likely a half century or more away, and in the meantime service positions, particularly in-person services, can provide employment for most working-age adults. The key challenge won’t be to create jobs, but rather to ensure that their employees have adequate income, decent working conditions, and opportunity for mobility.


The appendix has additional data.

  1. Gøsta Esping-Andersen et al, Why We Need a New Welfare State, Oxford University Press, 2002; Richard Layard, Happiness, Penguin, 2005; Lane Kenworthy, Jobs with Equality, Oxford University Press, 2008; Nathalie Morel, Bruno Palier, and Joakim Palme, eds., Towards a Social Investment Welfare State?, Policy Press, 2012; Alan B. Krueger, “Where Have All the Workers Gone?,” 2016; Anton Hemerijck, ed., The Uses of Social Investment, Oxford University Press, 2017. 
  2. The Economist, “The Great American Jobs Machine,” 2000. 
  3. Will Kimball and Robert E. Scott, “China Trade, Outsourcing, and Jobs,” Economic Policy Institute, 2014; Daron Acemoglu, David Autor, David Dorn, Gordon H. Hanson, and Brendan Price, “Import Competition and the Great US Employment Sag of the 2000s,” Journal of Labor Economics, 2016. 
  4. David H. Autor, David Dorn, and Gordon H. Hanson, “The China Shock: Learning from Labor-Market Adjustment to Large Changes in Trade,” Annual Review of Economics, 2016. 
  5. Jared Bernstein, The Reconnection Agenda Reuniting Growth and Prosperity, 2016. 
  6. Nicholas Eberstadt, Men Without Work: America’s Invisible Crisis, Templeton Press, 2016. 
  7. J.D. Harrison, “The Decline of American Entrepreneurship — in Five Charts,” Washington Post: Wonkblog, 2015. 
  8. Robert J. Gordon, “Okun’s Law and Productivity Innovations,” American Economic Review: Papers and Proceedings, 2010. 
  9. Francine D. Blau and Lawrence M. Kahn, “Female Labor Supply: Why Is the United States Falling Behind?,” American Economic Review, 2013; Olivier Thévenon, “Drivers of Female Labour Force Participation in the OECD,” Social, Economic, and Migration Working Papers, OECD. 
  10. Brigid Schulte and Alieza Durana, “The New America Care Report,” New America Foundation, 2016. 
  11. Lane Kenworthy, “A Decent and Rising Income Floor,” The Good Society. 
  12. OECD, Starting Strong II: Early Childhood Education and Care, 2006; Miriam Nordfors, “What Preschool Means in Sweden,” New York Times: Room for Debate, 2013. 
  13. OECD, Doing Better for Families, 2011. 
  14. Council of Economic Advisers, “The Long-Term Decline in Prime-Age Male Labor Force Participation,” The White House, 2016. 
  15. Betsey Stevenson, “Manly Men Need to Do More Girly Jobs,” Bloomberg View, 2016. 
  16. Alan Berube, “Where Are the Nonworking Prime-Age Men?,” Brookings Institution, 2016. 
  17. Binyamin Appelbaum, “Out of Trouble, but Criminal Records Keep Men Out of Work,” New York Times, 2015. 
  18. Mark Aguiar, Mark Bils, Derwin Kofi Charles, and Erik Hurst, “Leisure Luxuries and the Labor Supply of Young Men,” Working Paper 23552, National Bureau of Economic Research, 2017. 
  19. See also Christopher Ingraham, “It’s Not Just Young Men — Everyone’s Playing a Lot More Video Games,” Washington Post: Wonkblog, 2017. 
  20. Arthur Brooks, “The Dignity Deficit,” Foreign Affairs, 2017; Eberstadt, Men Without Work. 
  21. Lane Kenworthy, “Trade,” The Good Society. 
  22. OECD, The OECD Jobs Study, 1994. 
  23. Fritz W. Scharpf and Vivien A. Schmidt, eds., Welfare and Work in the Open Economy, two volumes, Oxford University Press, 2000; OECD, OECD Employment Outlook: Boosting Jobs and Wages, 2006; Kenworthy, Jobs with Equality. 
  24. Kevin Drum, “Welcome, Robot Overlords. Please Don’t Fire Us,” Mother Jones, 2013. 
  25. Andrew McAfee and Erik Brynjolfsson, “Human Work in the Robotic Future,” Foreign Affairs, 2016. 
  26. Lane Kenworthy, “Do Employment-Conditional Earnings Subsidies Work?,” ImPRovE Working Paper 15-10, Herman Deleeck Centre for Social Policy, University of Antwerp, 2015. 
  27. David Card, Jochen Kluve, and Andrea Weber, “What Works? A Meta Analysis of Recent Active Labor Market Program Evaluations,” Working Paper 21431, National Bureau of Economic Research, 2015; Claire Cain Miller and Quoctrung Bui, “Switching Careers Is Hard. It Doesn’t Have to Be,” New York Times, 2017. 
  28. Duncan Gallie, “The Quality of Working Life in Welfare Strategy,” in Why We Need a New Welfare State, Oxford University Press, 2002.