I'm reading up on discrimination in the labor market. In this study by Bertrand & Mullainathan (2004) they note the following:

Perhaps the skills of African-Americans are discounted because affirmative action makes it easier for African-Americans to get these skills. While this is plausible for creden- tials such as an employee-of-the-month honor, it is unclear why this would apply to more verifiable and harder skills.

Why would discounting not occur for harder and more verifiable skills?


My understanding is that this assumption is related to the models relying on bias in the observable signal (rather than differential variance or noise of these signals by race).

Discounting is less likely to apply to more verifiable and harder skills, because they are more verifiable in nature. Therefore an employer may more easily verify the quality of the signal. For example, hard skills such as writing, reading, math or ability to use computer programs are verifiable to some extent.

  • $\begingroup$ I don't know the underlying models. I simply know that affirmative action policies enable African Americans to get to college at a much higher likelihood than asians, whites or hispanics. (given other factors are equal) Thus it makes it easier for them to get these skills. Thus a college degree obtained by an African American is not the same as a college degree obtained by an Asian American, simply because the Asian American needed nearly perfect SAT scores to even get into college...Thus some discrimination should be expected even when controlled for educational status... $\endgroup$
    – Rubus
    Sep 12 '21 at 18:32
  • $\begingroup$ Given the study explores differences not based on actual skill but on characteristics of the applicants this result does not necessarily provide evidence for systemic racism - contrary to the conclusion of the authors... $\endgroup$
    – Rubus
    Sep 12 '21 at 18:35
  • $\begingroup$ @Rubus if two people behave identically in college, why would it matter what they did before college? Today X=5 and Y=5. Last decade X was 2 and Y was 4. Which is better? well, they are the same. Actually you may even extrapolate the growth rate and estimate that X is better, not Y. $\endgroup$
    – user253751
    Sep 15 '21 at 8:30
  • $\begingroup$ @user253751 Why would someone care about growth rate? Your reasoning assumes that education is completely repetitive, which is not true. Education is best modeled as all the knowledge acquired, so past performance does matter.. x1 = 3, x2 = 5, y1= 10, y2 = 5, x_tot = 8, y_tot = 15 $\endgroup$
    – Rubus
    Sep 15 '21 at 18:14
  • $\begingroup$ @Rubus "I don't know the underlying models." I recommend reading Aigner and Cain (1977) It is very informative and instructive. Otherwise, have I answered your question? $\endgroup$
    – emeryville
    Sep 15 '21 at 21:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.