James Bessen, Erich Denk, and Chen Meng
Why do wage gaps persist between men and women and between white and black workers? For a long time there were substantial differences between these groups in education and work experience. But those differences have now diminished or even reversed. Yet economists still find significant gaps remain even after accounting for a large variety of observed characteristics.
Is this evidence of discrimination? Some economists question this explanation, arguing that unobserved differences between groups might affect their productivity. Some suggest that free labor markets solve problems of discrimination–a worker paid less than her marginal productivity because her employer discriminates can just switch jobs to a non-discriminatory employer.
TPRI’s new research paper, based on a natural economic experiment, firmly rejects that view as too simplistic. We find that the story is different if employers know the salary history of job applicants. If a worker is underpaid because her employer discriminates, then prospective employers bargaining over wages will make offers based on that discriminatory pay, rather than on her productivity. This mechanism appears to perpetuate economic inequality.
We know this because about a dozen states in the US have recently banned employer access to salary history information. Using a rich difference-in-differences analysis, we find that salary history bans (SHBs) are followed by:
- A sharp increase in the rate at which online job advertisements post salary information. A large number of employers apparently prefer to post wages once they can no longer take advantage of salary history information in bargaining.
- A significant increase in the pay of workers who change jobs. After controlling for a large number of characteristics, the pay of job changers rises 5-6% more on average in states with an SHB than in states without.
- An even larger increase in the average pay of women and African-Americans who change jobs. The pay of women rises 8-9% more under an SHB; the pay of African-Americans rises 13-16% more. Moreover, these increases account for most of the residual wage gap.
- No change in the rates of worker turnover (possibly arising from bad matches) or the rates at which highly paid workers change jobs (adverse selection).
The SHBs reveal that women and African-Americans have indeed been paid less for reasons unrelated to their productivity. Employers using salary information may perpetuate past discrimination and other inequities. Note that these employers are not necessarily biased; they are just looking for the best deal. Economists tend to focus on individual rationales for discrimination such as a “taste for discrimination” or statistical discrimination. But the use of salary history in bargaining is an institutional mechanism that can perpetuate discrimination.
Fortunately, at a time when systemic discrimination against women and minorities is being confronted, and remedies are being sought to level the playing field for all, salary history bans appear to be an existing and effective means for doing so without apparent detriment to employers or employees.
Also: Wall Street Journal coverage.