AdobeStock_222534568.jpeg
VoxEU Column Labour Markets

Economic recessions and age discrimination

Around the globe, countries are experiencing a surge in the size and proportion of older individuals in their populations, leading to a ‘greying’ of the workforce. These older workers often encounter discrimination in the workplace. This column uses two unique data sources and two novel statistical approaches to show that economic downturns increase the incidence of illegal age discrimination. Whatever power disparities exist between workers and employers appear to grow during recessions, allowing employers to engage in higher levels of discrimination.

Around the globe, countries are experiencing a surge in the size and proportion of older individuals in their populations. A downstream effect has been the ‘greying’ of the workforce. In the US, for example, the employment of workers aged 65 or older has more than doubled over the past 20 years. This demographic shift in the workforce is projected to continue in the coming decades, with almost one fourth of all workers being age 55 or older by the year 2024 (Toossi and Torpey 2017). These older workers often encounter age discrimination in the workplace, both on the hiring and firing margins.

In light of the global recession caused by the COVID-19 pandemic, an important related question is whether economic downturns increase the incidence of illegal age discrimination. Older workers had a particularly difficult time becoming re-employed in the aftermath of the Great Recession (Neumark and Button 2014). However, whether this reflects increased discrimination or other factors is difficult to assess. While older workers experienced longer unemployment spells after the Great Recession (Johnson and Butrica 2012), attributing this to illegal discrimination is called into question by the early claiming of Social Security and early retirement (Bosworth 2012).

In recent work (Dahl and Knepper 2020), we take advantage of two unique data sources and two novel statistical approaches to test whether age discrimination increases during recessions. 

The impact of recessions on age discrimination 

Economists have long argued that taste-based discrimination – defined as an employer’s preference for worker characteristics which are unrelated to productivity – cannot survive in a competitive product market (Becker 1957). A similar argument has been made for the labour market based on search costs (Manning 2003, Biddle and Hamermesh 2013). The insight is that it hurts profits to pass on qualified, but disliked, workers and instead continue a costly search for favoured worker types. Therefore, a recession that increases the pool of unemployed workers creates opportunities for employers to indulge their discriminatory preferences at lower cost. Relatedly, if downwardly rigid wages necessitate layoffs, a discriminatory employer can fire less preferred workers without harming profits.

Testing this theory, however, has proven difficult. One of the challenges of assessing whether age discrimination varies with the business cycle is that discrimination is difficult to measure. Differences in employment or wages could well be due to discrimination, but they could also be due to productivity or cost differences across age groups. In our research, we tackle this measurement challenge using two complementary approaches. First, we use direct measures of hiring and firing discrimination for older workers based on charges filed by workers with the Equal Opportunity Employment Commission (EEOC). Our second approach uses fictitious resumes sent out to firms, where the age of the applicant responding to a job posting was randomly assigned to be older or younger.

Evidence from EEOC discrimination charges

The Age Discrimination in Employment Act of 1967 bans workplace discrimination based on age in hiring, firing, and other terms of employment. The EEOC is tasked with enforcing this law. We use EEOC microdata to assess whether the volume of discrimination charges filed by older individuals rises when the economy weakens. We do this by leveraging variation in the level of exposure to, and recovery from, the Great Recession across states, industries, and months.

We find that when the unemployment rate increases by one percentage point, the number of ADEA hiring charges rises by 1.6%; likewise, the volume of firing charges rises by 3.3%. The latter increase is not due simply to the mechanical increase that would be anticipated from the rise in discharged workers – in other words, a larger share of displaced workers file age discrimination cases during economic downturns.

While these findings indicate that the reported level of discrimination rises during a recession, they do not necessarily reflect higher levels of employer misconduct. Instead, the increase in volume could be due to an elevated incentive for employees to file. The reason is that when outside labour market opportunities decline, the opportunity cost of filing a claim shrinks, and hence more marginal cases will be filed even if actual discrimination has remained constant.

To address this concern, we take advantage of the fact that the EEOC investigates each case and determines whether a case has ‘merit’.  Using this as a proxy for the quality of a claim, we estimate that the fraction of cases with merit rises as the labour market deteriorates. A one percentage point increase in the unemployment rate increases the chance a case is determined to be meritorious by 0.7%. Under relatively mild assumptions, this leads to the conclusion that actual (as opposed to reported) age discrimination increases when unemployment rises. The intuition is that the rise in actual discrimination more than offsets the increased filing of weaker cases.

To help interpret the magnitude of our estimates, note that unemployment rose by 5.5 percentage points from the trough to the peak of the Great Recession. Our results imply that age-related firing and hiring discrimination charges rose by 18% and 9%, respectively. Similarly, the quality of firing and hiring discrimination charges increased by 4%.

Supplementary analyses suggest that the rise in age discrimination during slack labour markets is not driven by productivity considerations, changes in the composition of laid-off workers, employee efforts to win a case, industries with a large fraction of high-tech jobs, or the level of resources the EEOC has to pursue claims.

Evidence from correspondence studies

Our second approach uses what economists call a correspondence study. This is a common design used to study hiring discrimination, where fictitious resumes are sent to employers. We repurpose data from two correspondence studies originally conducted by Farber et al. (2017, 2019). These researchers sent out fake resumes of women, varying whether an applicant was older or younger.

Fortuitously, the resumes were sent out over multiple years in the aftermath of the Great Recession, and across eight different cities with different unemployment levels. We exploit this rich across-time and across-city variation in unemployment to see how potential employers treat the otherwise equivalent resumes of older and younger workers when economic conditions vary.

Figure 1 plots how relative callback rates by age group vary as a function of the unemployment rate in a city and time period. The graph illustrates that the age penalty grows substantially as the labour market weakens. As indicated by the downward sloping line, each one percentage point increase in the local unemployment rate lowers the relative callback rate of older versus younger women by 1.5 percentage points on average. Given a baseline callback rate of 10.2 percent, this translates to a 15% increase. This finding suggests that firms discriminate against older workers more when they have more workers to choose from. 

Figure 1 Age callback penalty as a function of the local unemployment rate in a city and time period

 

Discussion and interpretation

Our two analyses complement each other well, as each has its own strengths. The EEOC results cover the entire US and reflect real age discrimination borne by workers during a recession. We also examine the often difficult-to-study firing margin, which constitutes the majority of these types of age discrimination cases (85% versus 15% for hiring). The advantage of the correspondence study is that there is random assignment of job applicants to otherwise similar resumes, and it requires no assumptions about reporting behaviour during a recession.

Interpreting our results, the firing margin is more likely to reflect taste-based discrimination compared to the hiring margin, as firms will have already had a chance to learn about a worker’s productivity (Altonji and Pierret 2001). We are quick to point out, however, that employers may be statistically discriminating based on expected future productivity for older versus younger workers. (We note it is not illegal to discriminate based on current productivity, even if it is related to age.)1 For the hiring margin, both taste-based and statistical discrimination are likely to play important roles. While we cannot distinguish between the two types of discrimination, we note that both are illegal under the ADEA, and from the worker’s perspective, equally harmful.

Our work suggests that whatever power disparities exist between workers and employers grow during recessions, so that employers can engage in higher levels of age discrimination. One possible policy response is to provide increased resources to the EEOC, and other defenders of older workers’ rights, during economic downturns. 

Authors’ note: This column is based on prior research by the authors (Dahl and Knepper 2020), which benefitted from helpful guidance from Ron Edwards at the EEOC and from Henry Farber for generously sharing the correspondence study data.

References

Altonji, J and C Pierret, (2011), “Employer Learning and Statistical Discrimination,” Quarterly Journal of Economics 116(1): 313-350.

Becker, G (1957), The Economics of Discrimination, University of Chicago Press.

Biddle, J and D Hamermesh (2013), “Wage Discrimination over the Business Cycle," IZA Journal of Labor Policy 2(1): 1-20.

Bosworth, B (2012), “Economic Consequences of the Great Recession: Evidence from the Panel Study of Income Dynamics,” Center for Retirement Research at Boston College Working Paper No. 2012-4.

Dahl, G B and M M Knepper (2020), “Age Discrimination across the Business Cycle,” NBER Working Paper No. 27581.

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

Farber, H, C Herbst, D Silverman, and T von Wachter (2019), “Whom Do Employers Want? The Role of Recent Employment and Unemployment Status and Age,” Journal of Labor Economics 37(2): 323-349.

Johnson, R and B Butrica (2012), “Age Disparities in Unemployment and Reemployment during The Great Recession and Recovery,” Unemployment and Recovery Project Brief.

Manning, A (2003), “The Real Thin Theory: Monopsony in Modern Labour Markets,” Labour Economics 10(2): 105-131.

Neumark, D and P Button (2014), “Did Age Discrimination Protections Help Older Workers Weather the Great Recession?,” Journal of Policy Analysis and Management 33(3): 566-601.

Toossi, M and E Torpey (2017), "Older workers: Labor force trends and career options," Career Outlook, U.S. Bureau of Labor Statistics, May.

Endnotes

[1] See Smith v. City of Jackson, Miss. (544 U.S. 228 (2005)).

1,260 Reads