What Diversity can do, Design can do better.

People sometime argue that diversity in an organization is good because it improves the quality of ideas generated. Different backgrounds bring different perspectives, which provide different insights, whereas having many people with the same background causes partial redundancy.

This argument seems to be mainly made as a rationalization rather than as a true reason. It’s typically employed to support hiring more blacks, or less frequently more women and hispanics. But rarely do advocates explain exactly what new perspectives these people are meant to bring. Does one’s race give one a unique insight into how to write good code? If not, this argument seems pretty poor as a justification for discriminating in favor of blacks for programming jobs. Do women have special, vagina-based insights into maths or physics? If not, it doesn’t seem to work as a justification for discriminating in favor of women for STEM positions.

Indeed, if you actually wanted a diversity of opinions, you would probably just seek to hire that directly. Maybe your investment team should have majored in Economics, Physics, History and Statistics rather than Economics, Economics, Economics and Economics. Perhaps you should hire some social conservatives to your sociology department rather than actively and openly discriminating against them. Sometimes this strategy is employed – Corporate Boards do try to have people from a wide variety of backgrounds, both inside the company, different companies and even different industries. But I’ve never seen the pro-diversity crowd realize this purported benefit of racial diversity could be much more directly achieved.

Indeed, suppose different races did have different insights into programming. Then you would probably benefit from seeing each race represented. But while you might want some people from each race, there’s no reason to think you’d want them in the same fractions as appear in the overall population. At the moment having a racial breakdown significantly different from the US is enough to have you branded as un-diverse, but is there any reason to think the overall US has the optimal racial make-up for your company? Probably not. Indeed, as the racial make-up of the US is changing over time, even if your organization’s optimal make-up was fashionably diverse at the moment, it won’t be in the future, as the hispanic share increases and the white share decreases.

And if you were actually looking to take advantage of different racial perspectives and advantages, you wouldn’t have a corporate-wide quota or such. Instead, individual job openings would come with desired races attached. We would see a return to “No Blacks or Irish” notices on job postings, brought back at the auspices of political correctness.

Economies of Scale in Individual Labor Supply

Here are two stylized facts about labor economics:

  1. Utility is roughly logarithmic in income
  2. People who earn more per hour also work longer hours.

Together they present a puzzle – those higher income people are higher income primarily because they earn more dollars/hour, not because they work more hours. Yet if utility is logarithmic, there are diminishing returns to income, so we should expect people with higher hourly rates to work fewer hours.

Essentially, the first stylized fact suggests the income effect dominates, while the second suggests the substitution effect dominates.

One solution to this conundrum would be if the hourly rate changed with the number of hours worked. Maybe there are some jobs that simply cannot be done unless you put a huge amount of effort into them: you can’t be a part-time investment banker or corporate lawyer. If so, your productivity would increase dramatically with hours worked, so the demand curve for your labor would be upwards sloping. It’s a bit like a Giffen Good, except the causation goes

  • Higher Quantity -> More Valuable -> Higher Demand -> Higher Price

rather than

  • Higher Price -> More Valuable -> Higher Demand -> Higher Quantity

At the same time, every extra dollar is worth less and less to you, and each hour of leisure lost hurts more than the previous one, so you demand a higher hourly wage the more hours you work. So your supply curve is upwards sloping, roughly exponentially (to offset the logarithmic dollars->utility conversion)

When both supply and demand curves and upwards sloping, it is not clear there is a unique equilibrium – there could be multiple equilibria.

This could explain why we see such a difference between the incomes of

  1. The increasing number of people who do not work at all
  2. People who do ordinary jobs for around 40 hours a week
  3. Extremely high earning extremely hard working people

each group occupies a different one of these equilibria.