Academics habitually tax the public's patience with stupid questions. It's easy to see why practical folk ignore us. Every now and then, though, an academic asks a truly profound question that seems stupid on the surface. Case in point: David Romer's still unpublished 1992 paper "Why Do Firms Prefer More Able Workers?," newly available on my website with his kind permission.
What on earth could be puzzling about the fact that firms prefer more able workers? Well, if firms pay everyone what they're worth, then hiring the most-able is no more profitable than hiring the least-able. As Romer explains:
In most markets, prices rise with general quality so that buyers have optimal levels of quality. In the automobile market, for example, a typical buyer identifies some target level or range of general quality and then searches for cars within that general category... The cost of a car rises essentially one-for-one with its general quality, and thus there is no reason to expect the highest quality cars to generate the most surplus for a buyer; rather, large surplus is generated when a buyer is able to find a car that fits his or her particular needs and tastes well.
This account does not appear to describe what occurs in the labor market. Consider a firm that is attempting to fill a position. There will generally be a variety of applicants, and they will typically differ in terms of observable characteristics that are relevant to likely performance on the job - education, experience, punctuality, articulateness, and so on. In most circumstances, the firm does not have some target level of ability; rather, faced with a variety of applicants who are interested in the position, it offers the position to the most able. Equivalently, typically the workers that a firm would least like to lose are its most able. Personnel management texts and hiring and interviewing guides, for example, simply take it as given that the goal of the hiring process is to find the most talented possible worker.
After putting the puzzle on the table, Romer offers an elegant solution: Firms prefer more able workers because pay is more equal than productivity. Why? Because inequality hurts morale of lower-paid workers. After presenting a mathematical model to show that his story is internally consistent, Romer discusses empirical evidence in favor of his story. He starts by surveying "observers of the labor market":
Consider for example the evidence provided by the authors of personnel management textbooks. An important message of such textbooks is that it is the effort of low-ability workers than is most easily affected by pay policies. Fairness and rewarding ability and accomplishment are generally viewed as distinct functions of a pay system...
The authors of such books simply presume that compensation should (and does) rise less than one-for-one with workers' contributions, and that the only question is to what extent pay should be related to performance.
Then he covers experimental evidence showing that:
...when experimental subjects are asked to allocate rewards among participants in a project to promote harmony and minimize conflict, they reduce the dispersion in the rewards. Such reallocations make sense only if notions of fairness involve an element of absolute equality rather than simply rewards proportional to contributions.
Finally, Romer points out a bunch of odd facts that his model readily explains:
Unions simultaneously make compensation more open and more compressed.
Why some firms and industries pay higher wages to all their workers.
Empirical work on the last point is intriguing. Romer:
Dickens and Katz (1987a) find that one of the variables most strongly associated with inter-industry wage differences is the average education level of workers in the industry; that is, a given worker on average earns a higher wage in an industry where his or her fellow workers have greater education. This is precisely in accord with the prediction of the theory that a worker of any given ability level obtains a higher wage in a firm where average ability levels are higher. Dickens and Katz find that industry wage differences are moderately associated with other worker characteristics associated with higher wages, notably average experience and tenure and the fraction of workers who are male.
Also check out Romer's explanation for large firms' tendency to pay all their workers more.
Overall, a profound paper. I'm not quite ready to say "Read the whole thing"; pondering the mathematical sections equation-by-equation could easily take all day. But I will say "Read all the English." You won't just learn important lessons about labor economics. You'll also learn the meta-lesson that the distinguishing between banal and profound questions is harder than it looks.