Age Discrimination in the U.S. Labor Market

By David Neumark

Population aging poses fundamental public policy challenges. Barring changes in the employment of older adults, population aging implies a shrinking number of workers to sup­-port a growing number of retirees.

Population Aging and Age Discrimination

The projected population aging of the United States is depicted in Figure 1 (on page 52). Most significantly, the low employment rate of older adults implies that the labor force will grow more slowly than the population, which also im­­plies a rising dependency ratio.

This creates an imperative to increase older adult employment, so as to lower the dependency ratio, to increase tax revenues, and—as programs are currently structured—to decrease public expenditures on health insurance, retirement benefits, and income support.

This imperative has led to numerous reforms intended to increase the employment (or num­ber of work hours) of those who would other­wise retire, including the following: lowering Social Security benefits at the early retirement age of 62; raising the retirement age at which full benefits are available from age 65 to age 67, beginning with the 1938 birth cohort (Ameri­can Academy of Actuaries, 2002; Munnell et al., 2004); and changing benefits taxation, to reduce the marginal tax rate on the earnings of Social Security recipients in excess of the earnings cap, to increase the exempt amount of earnings (the cap), and to broaden  the age range of those not subject to the earnings test (Friedberg, 2000). Additional changes are likely to be considered as part of efforts to shore up the solvency of Social Security or to reform the system.

However, if age discrimination creates sub­stantial barriers to employment of older work­ers, then the policy goal of increasing older adult employment may be frustrated. The clash between age discrimination and policies to encourage working longer can be particularly problematic. If age discrimination is a pervasive barrier to employment of older workers, then policy makers may find that efforts to induce older Americans to work longer fail to deliver sufficient employment increases. Instead, pol­icy makers may be compelled to enact even stronger or harsher incentives to delay retire­ment—such as lowering Social Security benefits paid before, say, age 70. These kinds of changes could have adverse consequences for some older workers for whom working beyond what were normal retirement ages in the past poses diffi­culties because of disability or declining health. In other words, supply-side incentives we have adopted—for example, cutting Social Security benefits at early retirement ages—may run up against demand-side barriers posed by age discrimination.

Importantly, even without this fiscally induced public policy imperative stemming from population aging, many older Americans want to keep working past ages at which they have typically retired and continue to retire. We, as a society, deem barriers to work attributable to discrimination as wrong and something to be rooted out. This is evidenced by the long history of passing laws barring discrimination in the U.S. labor market, including the Equal Pay Act (1964), Title VII of the Civil Rights Act (1964), the Age Discrimination in Employment Act (ADEA, 1967), and the Americans with Disabilities Act (1991).

These considerations raise a central question: Is age discrimination a signifi­cant problem for older workers? And if it is, what might we do about it, both to enable those who want to work longer to be able to do so, and also as a comple­ment to reforms intended to increase the labor supply of older workers?

Evidence of Age Discrimination

Economists and other social scientists typically are more cautious in drawing an inference of discrimination than are more casual observers of the labor market. In particular, an observed difference in outcomes—for example, the lon­ger unemployment durations for older workers searching for jobs than for younger workers—does not necessarily imply hiring discrimination against older workers.

Such evidence is consis­tent with age discrimination, but it may have another explanation. There is occasionally—but rarely, at least these days—blatant and direct evidence of discrimination. Usually the evidence documented is more indirect, making conclu­sions potentially more ambiguous. As a result, economists and others have pursued more rigor­ous ways of testing for discrimination, including experimental and other approaches.

Direct evidence of age discrimination

The most direct evidence of age discrimination comes from the period prior to the passage of the ADEA, when hiring ads frequently featured explicit age restrictions (U.S. Department of Labor, 1965). While this is unambiguous evi­dence of age discrimination, the other, more recent evidence we have is less direct, and sub­ject to alternative interpretations.

Indirect behavioral evidence of age discrimination

Indirect evidence of discrimination is based on observed behavior (i.e., from surveys or other means of collecting data) of older workers com­pared to other workers, or of the behavior of other agents regarding older versus younger workers.

The prime example is evidence that work­ers in their 50s and early 60s have long had, and continue to have, lengthier unemployment dura­tions than many other age groups, which is con­sistent with employers’ greater reluctance to hire older workers (Neumark and Button, 2014). While older workers’ longer unemployment dura­tions may be attributable to discrimination, they also could be due to other factors. For example, older workers may search among a narrower sub­set of less physically demanding jobs, or maintain a higher reservation wage (the lowest wage at which they would accept a new job) because they earned higher wages in a previous job, or are less interested in working at some types of jobs.

Another type of indirect evidence of age discrimination is negative stereotypes regard­ing older workers. These are usually measured in “the lab,” where respondents are presented with hypothetical scenarios and decisions that explore connections between attitudes or stereo­types regarding older workers, sometimes tying them to possibly adverse decisions about older workers (Gordon and Arvey, 2004; Kite et al., 2005). The limitation of this evidence is that it is difficult to tie these stereotypes to adverse outcomes in the actual labor market.

Another issue with these stereotypes is that even if they adversely impact older workers as a group—for example, reducing hiring because employers expect older workers to stay on the job for less time, hence making investments in their training less worthwhile—the stereotypes could be true. That is, a stereotype is an average characteristic attributed to a group and, in some cases (e.g., health of older people will eventually decline), a stereotype can be true. Interestingly, however, discrimination based on these kinds of stereotypes—which economists call “statis­tical discrimination”—is illegal under U.S. law. The EEOC regulations state: “An employer may not base hiring decisions on stereotypes and assumptions about a person’s race, color, reli­gion, sex (including pregnancy), national origin, age (40 or older), disability or genetic informa­tion”.

Making it clear that employers must evalu­ate people as individuals, and not as members of a group, the EEOC regulations further state: “The principle of nondiscrimination requires that individuals be considered on the basis of individual capacities and not on the basis of any characteristics generally attributed to the group”. Thus, while statistical discrimination, or decisions based on group ste­reotypes, may sometimes be rational for employ­ers, it is defined as illegal discrimination under the law. Still, it is harder to determine whether negative stereotypes about older workers result in real-world adverse outcomes.

Yet a third type of indirect evidence is that workers frequently report experiencing age dis­crimination on the job, and these workers sub­sequently exhibit more separations from their current employers, lower employment rates, slower wage growth, and reduced expecta­tions of working past their mid-60s (Johnson and Neumark, 1997; Adams, 2002). Again, these results from observational data are not decisive evidence of age discrimination. Self-reports of age discrimination may reflect other negative decisions or outcomes workers experience that, when asked about these, they attribute to age discrimination. And we would expect these neg­ative outcomes or decisions would be more likely to be followed by a worker leaving the firm, or experiencing fewer promotions or raises, etc.

Indirect evidence derived from policy variation

A different kind of indirect evidence comes from studying whether labor market outcomes for older workers improve when laws protecting them from discrimination are implemented or strengthened. If there is age discrimination, and these laws lessen discrimination, then an improvement in labor market outcomes could mean that the laws helped root out discrimination.

Research suggests that age discrimination laws have been effective in improving labor market outcomes for older workers. For exam­ple, Adams (2004) found that the advent of state and then federal age discrimination laws boosted older worker employment, and Neu­mark and Stock (1999) showed that these laws strengthened the employment relationship between older workers and firms, which they interpreted as evidence that age discrimina­tion laws made it harder for employers to fire older workers who had worked for them for a long time.

In more recent work, Neumark and Song (2013) studied responses to the changes in Social Security, implemented from 2003 to 2008, which lowered benefits at the early retirement age of 62 and increased the full retirement age. The study found that older workers’ responses—working longer and claiming Social Security benefits later—were higher in states that had stronger age discrimination laws. It is important to note that in these states, larger damages were made avail­able to plaintiffs in age discrimination lawsuits, and these states also extended coverage of age discrimination protections to smaller firms.

The 2013 study’s outcomes not only provided further evidence that stronger age discrimina­tion laws boost employment (and delay retire­ment) of older workers, but also pointed more directly to policy “complementarities” between supply-side incentives to induce older workers to work longer, and demand-side efforts to reduce age discrimination.

Why characterize this evidence on age dis­crimination laws as indirect? Because age dis­crimination laws do not work perfectly. In some cases, they could provide a basis for a worker to contest an outcome that they do not like—e.g., being terminated, being paid less, or being passed over in hiring—and the law could simply provide a basis for contesting the outcome. If so, employers might respond by reducing termina­tions of older workers, or increasing pay or hir­ing, even if there was no actual discrimination in the first place. Though this is unlikely (especially in light of the evidence discussed below), this possibility cannot be dismissed out of hand.

Experimental, or “direct” evidence

In light of these empirical challenges, research­ers have turned to experimental methods, called “audit” or “correspondence” (AC) stud­ies, which focus on hiring practices. Although hiring discrimination may not seem closely related to encouraging older people to work longer, it may be critical to significant length­ening of work lives, because many older adults transition to part-time or shorter-term “partial retirement” or “bridge jobs” toward the end of their careers (Cahill et al., 2006; Johnson, 2014), or return to work after a period of retire­ment (Maestas, 2010). 

AC studies of discrimination in hiring gen­erally are viewed as the most reliable means of inferring labor market discrimination (e.g., Fix and Struyk, 1993). Observational studies on dis­crimination typically start with some observed difference between groups—such as older work­ers’ longer unemployment durations—and then ask whether other factors that differ between older and younger workers can explain that observed difference. If such inquiry cannot find an explanation, an inference of discrimination often is drawn.

In contrast, AC studies create artificial job applicants in which there are no average dif­ferences by group; hence there is no reason for employers to treat the applicants differently unless employers have different tastes for hiring from different groups of workers, or make dif­ferent assumptions about those workers that are not justified based on any information employ­ers have about the individual applicants. Thus, if there are differences in outcomes, they likely reflect discrimination.

Audit studies use actual applicants coached to act alike, and this research measures dis­crimination as differences in job offer rates. Cor­respondence studies create fake applicants (on paper, or electronically) and capture “callbacks” for job interviews. Correspondence studies can collect far larger samples of job applications and outcomes, especially using the Internet; due to interview time costs, even expensive large-scale audit studies usually have sample sizes only in the hundreds. Correspondence studies also avoid “experimenter effects,” which can influence the behavior of the actual applicants used in audit studies (Heckman and Siegelman, 1993).

Though correspondence studies capture only callbacks and not actual job offers there is evi­dence that callbacks capture most of the rel­evant discrimination. For example, Riach and Rich (2002) discuss evidence from studies by the International Labour Organization (ILO) indicating that 90 percent of the discrimina­tion detected in these ILO studies occurs at the selection-for-interview stage; similar evidence is reported in Neumark (1996).

In a recent like study of age discrimination in hiring, which focused on ages near retirement at which policy makers are trying to strengthen incentives to work longer, we (Neumark, Burn, and Button, 2019) conducted a large-scale cor­respondence study that was carefully designed to provide the best possible evidence on age dis­crimination in hiring, in part by overcoming potential biases in past studies. For example, the study used specifically crafted variations on resumés that older workers present, including one type of resumé showing the common path of moving to a lower-skill job later in life (e.g., a store greeter at Walmart).

We created fictitious but highly realis­tic resumés for young (ages 29 to 31), middle-age (ages 49 to 51), and older (ages 64 to 66) job applicants. These then were submitted to ads for job categories that employ large numbers of fairly low-skilled workers of all ages, and in which employers hire some of both older and younger workers. The jobs included administra­tive assistants and secretaries (to which we sent female applicants’ resumés), janitors and secu­rity guards (male applicants), and retail sales (both genders). The experiment covered fairly low-skilled jobs, as labor economists using audit and correspondence study methods believe that realistic responses to fictitious job applications are less likely in more high-skilled labor markets where employers are more likely to be familiar with job applicants.

The study leveraged technology to conduct its process on a far larger scale than ever used before, sending triplets of otherwise identical young, middle-age, and older fictitious applica­tions to more than 13,000 positions in twelve cities spread across eleven states (more than 40,000 applicants).

Overall, across all five sets of job applications, the callback rate was higher for younger appli­cants and lower for older applicants, consistent with age discrimination in hiring; however, there are some important differences. In both cases, there was a distinct pattern of the callback rates being highest for the young applicants, lower for the middle-age applicants, and the lowest for the old applicants. Relative to young applicants, older female applicants for administrative jobs had a callback rate that was lower by 47 percent (7.58 percent versus 14.41 per­­cent). In sales, the differ­ence was a bit smaller (a 36 percent lower call­back rate—18.43 percent versus 28.68 percent).

For male job applicants in sales, security, and janitor jobs there generally is a lower call­back rate for older men; however, the age pattern was not as consistent or pronounced. For sales jobs, for which we have a direct comparison with women, the difference in callback rates between old and young applicants was more modest (14.70 percent versus 20.89 percent, or a 30 percent dif­ference). None of the three cases show a clear monotonic decline in callback rates from young to middle-age to old applicants, in contrast to the evidence shown for women.

The results depicted in Figure 2 indicate that women face worse age discrimination than men. In retail sales, where we could directly compare results for both genders, we found a sharper drop-off in callback rates with age for women than for men. And for the janitor and security jobs to which we submitted applications from males, the pattern of lower callback rates for older applicants was less clear than for the older versus younger female applicants to administra­tive or retail jobs.

Our 2019 study contains several additional analyses, but they support the same three conclu­sions. First, there is evidence of age discrimina­tion in hiring, for both women and men. Second, while both middle-age and older applicants expe­rience discrimination relative to younger applicants, older applicants near the age of retirement experience more age discrimination. Third, women experi­ence more age discrimination than do men. This general finding of more dis­crimination against the oldest applicants (notably, among people near traditional retirement ages), and more robust evi­dence of age discrimination against older women, is confirmed in a number of our more detailed (and, in some cases, more technical) analyses.

Two additional points are key. First, although the 2019 study improves on past research, its findings are consistent with ear­lier AC studies of age discrimination (Bendick, Jackson, and Romero, 1997; Bendick, Brown, and Wall, 1999; Riach and Rich, 2006, 2010; Lahey, 2008) and with a recent study by Farber, Sil­verman, and von Wachter (2017). Thus, there is sizable body of work indicating the same conclu­sion. Moreover, while not contrasting men and women, the studies conducted by Lahey (2008) and Farber and colleagues (2017) focused on women and found evidence of age discrimination against older women.

Second, the stronger evidence of age dis­crimination against older women leads us to ask why. Evidence suggests that physical appearance matters more for women (Jackson, 1992) and that age detracts more from physi­cal appearance for women than for men (e.g., Deutsch, Zalenski, and Clark, 1986). This is con­sistent with evidence in research by Kuhn and Shen (2013) and Helleseter, Kuhn, and Shen (2016): In job descriptions posted on Internet job boards in China and Mexico, employers often expressed preferences for workers based on age and sex. This research found a “twist” in rela­tive preference away from women with respect to age. There was greater preference toward women in those job descriptions seeking young workers, and for men in those job descriptions seeking older workers.

These findings hold potentially significant policy implications. If older women suffer from discrimination because of both age and sex, anti-discrimination laws may be less effective than thought. Because Title VII of the Civil Rights Act, which prohibits sex discrimination, is separate from the ADEA, “intersectional” claims of age dis­crimination against older women are difficult to bring before the courts (McLaughlin, 2018). 

Reducing age discrimination against older women may be particularly crucial to their finan­cial security; many women outlive their husbands and end up quite poor. Unless more is done to combat age discrimination and make it easier for older women to stay in or return to the workforce, policy incentives to retire later may reduce older women’s retirement benefits without doing much to increase their employment—ultimately caus­ing more harm than good. If women worked into older ages, their post-retirement financial straits might be eased. But our 2019 study indicates that age discrimination may play a key part in limiting the ability of older women to work longer.

What Can Be Done to Bolster Older Workers’ Employment?

Stronger age discrimination laws can help to reduce demand-side barriers to older workers extending their work lives. This conclusion is reinforced by more recent evidence (Neumark et al., forthcoming) suggesting that the kind of “clean” experimental measures of age discrimina­tion discussed above are somewhat lower in states where age discrimination laws are strengthened by allowing for larger damages to be paid to plain­tiffs in age discrimination cases. It is possible for additional states, or the federal government, to amend age discrimination laws to make the penal­ties for discrimination more stringent. It may be that some aspects of these laws—those that have not been studied as systemically as were oth­ers—also can be strengthened; however, effective changes to anti-discrimination laws must be sup­ported by solid evidence.

Aside from stronger laws and more assiduous enforcement, we also need a sea change in how we think about older workers. It seems much more acceptable to make decisions based upon age rather than, say, race or sex. I have never heard anyone say “We should not hire that per­son because they are black.” I have, however, heard people say “We should not hire that per­son because they are old.”

To be clear, I am not just preaching. I, like others, have uttered derogatory comments or jokes about older people that would be far more taboo with respect to minorities or women. Why this is so is a fascinating question. I think, though, that it implies that reducing age discrim­ination requires more than stronger laws.

David Neumark, Ph.D., is Distinguished Professor of Economics at the University of California, Irvine. He also is a research associate at NBER, and a research fellow at IZA Institute of Labor Economics and CESifo in Munich.


Adams, S. J. 2002. “Passed Over for Promotion Because of Age: An Empirical Analysis of the Con­sequences.” Journal of Labor Research 23(3): 447–61.

Adams, S. J. 2004. “Age Discrimi­nation Legislation and the Employ­ment of Older Workers.” Labour Economics 11(2): 219–41.

American Academy of Actuar­ies. 2002. Raising the Retirement Age for Social Security. Issue Brief. Washington, DC: American Acad­emy of Actuaries.

Bendick, M., Jr., Brown, L. E., and Wall, K. 1999. “No Foot in the Door: An Experimental Study of Employment Discrimination Against Older Workers.” Journal of Aging & Social Policy 10(4): 5–23.

Bendick, M., Jr., Jackson, C. W., and Romero, J. H. 1997. “Employ­ment Discrimination Against Older Workers: An Experimental Study of Hiring Practices.” Journal of Aging & Social Policy 8(4): 25–46.

Cahill, K. E., et al. 2006. “Retire­ment Patterns from Career Em­­ployment.” The Gerontologist 
46(4): 514–23.

Deutsch, F. M., Zalenski, C. M., and Clark, M. E. 1986. “Is There a Dou­ble Standard of Aging?” Journal of Applied Social Psychology 16(9): 771–85.

Farber, H. S., Silverman, D., and von Wachter, T. 2017. “Factors Determining Callbacks to Job Applications by the Unemployed.” RSF: The Russell Sage Foundation Journal of the Social Sciences 3(3): 168–201.

Fix, M., and Struyk, R. 1993. Clear and Convincing Evidence: Measure­ment of Discrimination in America. Washington, DC: Urban Institute Press.

Friedberg, L. 2000. “The Labor Supply Effects of the Social Secu­rity Earnings Test.” Review of Eco­nomics and Statistics 81(1): 48–63.

Gordon, R. A., and Arvey, R. D. 2004. “Age Bias in Laboratory and Field Settings: A Meta-analytic Investigation.” Journal of Applied Social Psychology 34(3): 468–92.

Heckman, J., and Siegelman, P. 1993. “The Urban Institute Audit Studies: Their Methods and Find­ings.” In M. Fix and R. Struyk, eds., Clear and Convincing Evidence: Measurement of Discrimination in America. Washington, DC: Urban Institute Press.

Helleseter, M. D., Kuhn, P., and Shen, K. 2016. “Age and Gender Profiling in the Chinese and Mexi­can Labor Markets: Evidence from Four Job Boards.” IZA Discus­sion Paper No. 9891. Retrieved June 4, 2019.

Jackson, L. A. 1992. Physical Appearance and Gender: Sociologi­cal and Sociocultural Perspectives. Albany, NY: SUNY Press.

Johnson, R. W. 2014. Later Life Job Changes Before and After the Great Recession. Draft final report to AARP. Retrieved June 4, 2019.

Johnson, R. W., and Neumark, D. 1997. “Age Discrimination, Job Separations, and Employment Sta­tus of Older Workers.” Journal of Human Resources 32(4): 779–811.

Kite, M. E., et al. 2005. “Atti­tudes Toward Younger and Older Adults: An Updated Meta-analytic Review.” Journal of Social Issues 61(2): 241–66.

Kuhn, P., and Shen, K. 2013. “Gen­der Discrimination in Job Ads: Evi­dence from China.” Quarterly Jour­nal of Economics 128(1): 287–336.

Lahey, J. 2008. “Age, Women, and Hiring: An Experimental Study.” Journal of Human Resources 43(Winter): 30–56.

Maestas, N. 2010. “Back to Work: Expectations and Realizations of Work after Retirement.” Journal of Human Resources 45(3): 718–48.

McLaughlin, J. S. 2018. “Limited Legal Recourse for Older Wom­en’s Intersectional Discrimination Under the Age Discrimination in Employment Act.” The Elder Law Journal. Retrieved June 4, 2019.

Munnell, A. H., et al. 2004. Should We Raise Social Security’s Earli­est Eligibility Age? Issue in Brief, No. 18. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Neumark, D. 1996. “Sex Discrimi­nation in Restaurant Hiring: An Audit Study.” Quarterly Journal of Economics 111(3): 915–41.

Neumark, D., and Button, P. 2014. “Did Age Discrimination Protec­tions Help Older Workers Weather the Great Recession?” Journal of Policy Analysis and Management 33(Summer): 566–601.

Neumark, D., and Song, J. 2013. “Do Stronger Age Discrimination Laws Make Social Security Reforms More Effective?” Journal of Public Economics 108(December): 1–16.

Neumark, D., and Stock, W. A. 1999. “Age Discrimination Laws and Labor Market Efficiency.” Journal of Political Economy 107(5): 1081–125.

Neumark, D., Burn, I., and But­ton, P. 2019. “Is It Harder for Older Workers to Find Jobs? New and Improved Evidence from a Field Experiment.” Journal of Political Economy 127(2): 922–70.

Neumark, D., et al. (forthcom­ing). “Do State Laws Protecting Older Workers from Discrimina­tion Reduce Age Discrimination in Hiring? Evidence from a Field Experiment.” Journal of Law and Economics.

Riach, P. A., and Rich, J. 2002. “Field Experiments of Discrimi­nation in the Market Place.” The Economic Journal 112(November): F480–518.

Riach, P. A., and Rich, J. 2006. “An Experimental Investigation of Age Discrimination in the French Labour Market.” Discussion Paper No. 2522. Bonn, Germany: IZA.

Riach, P. A., and Rich, J. 2010. “An Experimental Investigation of Age Discrimination in the Eng­lish Labor Market.” Annals of Eco­nomics and Statistics 99/100 (July/December): 169–85.

U.S. Census Bureau. 2014. “Projec­tions of the Population by Sex and Age for the United States: 2015 to 2060 (NP2014–T3).” Washington, DC: U.S. Census Bureau, Popula­tion Division. s5kb. Retrieved May 28, 2019.

U.S. Department of Labor. 1965. The Older American Worker. Washington, DC: U.S. Government Printing Office. 

This article is taken from the Fall 2019 issue of ASA’s quarterly journal, Generations. ASA members receive Generations as a membership benefit; non-members may purchase subscriptions or single copies of issues at our online store.