In a New York Times article, Daniel Gross sympathetically discusses a paper by MIT economists Peter Temin and Frank Levy on the role of institutional behavior and social attitudes in income inequality. As a preface to his comments about their work, he sets forth some of the conventional arguments for the inevitability of inequality, namely, that it results from differences in skills and productivity:
Many economists, especially those who find themselves in the Bush administration, argue that the winner-take-all trend is fueled by other, unstoppable trends. After all, globalization, information technology and free trade place a premium on skills and education. “The good news is that most of the inequality reflects an increase in returns to ‘investing in skills’ — workers completing more school, getting more training and acquiring new capabilities,” as Edward P. Lazear, the chairman of the Council of Economic Advisers, put it last year.
It takes an optimist to find good news in the fact that the top 1 percent have steadily increased their haul while the other 99 percent haven’t; after all, many more than one in every 100 Americans are investing in skills and education.
But the orthodoxy surrounding income inequality is being undermined by research that looks at institutional issues: changes in the way the corporate world measures the performance of workers, the decline of unions, and government wage and tax policy. In this view, skills, education and trade aren’t the whole story. They’re simply “factors operating within a broader institutional story,” as Frank Levy, the Rose professor of urban economics at the Massachusetts Institute of Technology, describes it.
One big change in recent decades has been a rise in performance-based pay. Through the 1970s, thanks in part to unions that negotiated wages collectively, “people with different abilities and capabilities were frequently paid the same amount for doing similar jobs,” said W. Bentley MacLeod, an economics professor at Columbia.
But as companies and compensation consultants began using information technology to determine more accurately the contributions of individual employees, employers began to discriminate among employees based on performance. In a working paper, Professor MacLeod, along with Thomas Lemieux of the University of British Columbia and Daniel Parent of McGill University, mined census data and found that the proportion of jobs with a performance-pay component rose to 40 percent in the 1990s from 30 percent in the late 1970s.
“Since companies are better able to measure precisely what an employee contributes, we’ve seen a greater range of incomes among people doing roughly the same jobs,” Professor MacLeod said.
The fact that more Americans are paid less on the basis of a job title and more on their individual output inexorably leads to greater inequality. The authors’ conclusion is that the rise of performance-based pay has accounted for 25 percent of the growth in wage inequality among male workers from 1976 to 1993.
Arguments like this are prima facie evidence for why you can’t leave something as important as economics in the hands of economists.
“…companies are better able to measure precisely what an employee contributes” is a fiction because individual contribution is a fiction. Yes, employees can be more or less productive within certain confines. But their productivity is determined much more by institutional factors and environmental conditions than by their own efforts. Quality guru (and father of modern statistics) Edward Deming was vehemently opposed to the management by objectives construct created by Peter Drucker which was one of the early efforts to create individual accountability in the corporate context. Deming the statistician was adept at illustrating how differences in performance were often random or otherwise due to factors outside the employee’s control, and efforts to base rewards on apparent productivity could make matters worse.
Now Deming was concerned mainly with manufacturing, where individual output was much more measurable than in other environments, where the circumstances facing each employee are subject to greater variability than a worker on an assembly line.
Consider sales. Selling is one of the few areas in an organization where individual output can be measured, right? Yes, but assuming that results simply reflect talent and doggedness is a mistake. Anyone who has ever been in sales knows that who has the best results reflects in large measure who has the best territory. And in established businesses, territories are assigned, not developed.
Now one can argue that big companies have an incentive to put their best salesmen on their best territories, so talent will win out anyhow. But how true is that at the margin? With large accounts, they may have reasons for preferring your company that extend beyond the salesman’s initiative, such as unique product characteristics, warranty, or political controversy and risk involved in switching accounts. Thus, whole salesmen on the top accounts must be competent, they may not necessarily be the best.
Furthermore, organizations have well-documented tendencies to pay attractive and tall people better. One would expect these prejudices to operate in sales, with prettier and taller people being preferred as salespeople and given better assignments. The drug industry and certain Wall Street firms are know for hiring attractive, athletic types for client facing positions. Thus, what is being rewarded is not talent or effort, but advantages that have nothing to do with skills. So much for the argument that pay for performance reflects merit.
In fact, there is evidence that bonuses and supposed incentive based pay do not reflect productivity. Consider the most closely studied bonus-receiving population, CEOs. Most studies find no correlation between pay and performance and some have even found a negative correlation (the few that have were conducted by search firms, meaning the very group that profits from developing these pay pacakges and had to go through some gymnastics to produce the desired findings).
Moreover, performance appraisal systems, which are the foundation for bonuses and other merit based pay, are hopelessly and intrinsically flawed. Carnegie Mellon professors Patrick D. Larkey and Jonathan P. Caulkins’ 1992 paper “All Above Average and Other Unintended Consequences of Performance Evaluation Systems,” discuss how, despite 100 years of effort, performance appraisal systems fail to achieve their intended results due to romanticized notions about how organizations work and difficulties in making comparative rankings of workers engaged in different tasks. For example, the article discussed the many ways a boss’s motivations and quirks could lead to misleading ratings.
Caulkins and Larkey’s analysis showed that the idea that organizations are or can be meritocracies is a myth. Yet people have a powerful need to believe that society and the institutions they belong to are fair. These factors explain not only why increasing income inequality rankles, but why so many strive to attribute it to market forces or skill differences.
Yet even if you ignored the organizational factors discussed above, the NBER study attributed only about 25% of the increase in inequality to performance-related pay. That begs the question of what caused the other 75%.