Research by MIT Sloan Professor Reveals Merit-Based Reward Systems Can Increase Gender and Race Bias in Workplace

Keys to Reducing Inequity: Formalization, Accountability, and Transparency

CAMBRIDGE, Mass., August 22, 2008 — Merit-based rewards and other common performance management practices used by the majority of U.S. companies today can actually increase bias and reduce equity in the workplace, according to recent research by Professor Emilio J. Castilla of MIT's Sloan School of Management. Such practices and policies can result in women and minorities receiving less compensation than white men despite equal scores on their performance evaluations, he found, noting that such “performance-reward bias” can be overcome by increasing accountability and transparency in the organizational processes and routines that connect performance evaluations and wage increase decisions.

“Already in 2005, it was estimated that close to 70 percent of organizations offered variable bonuses based on employee performance. Today even more companies rely on performance-related rewards,” he said. “Perhaps these merit-based practices are intended to increase workers' job satisfaction and motivation to work hard, but to what extent are these practices working to solve workplace inequality? I found evidence of what I call ‘performance reward bias’ where even assuming that employers successfully find fair and unbiased ways of measuring employee merit during the performance evaluation stage, they can still potentially introduce bias and discretion in the way performance evaluations are used to determine employee compensation and promotions during the performance-reward stage.”

Castilla's article, “Gender, Race, and Meritocracy in Organizational Careers,” published in the most recent issue of the American Journal of Sociology, examines the relationship between performance evaluations and wage growth using personnel data from a large service organization in the U.S., which used a two-step process that separates performance evaluations from pay decisions. While the organization adopted a merit-based practice to ensure that rewards are allocated meritocratically, Castilla found evidence that this was not necessarily the case.

“The key finding was that women and minorities in the same job and work unit, with the same supervisor, and who had the same human capital received lower pay increases than white males, even when they were given the same performance evaluation scores,” he said, which is evidence of performance reward bias. While women and minorities did not initially receive lower starting salaries or performance ratings than white men, he found bias in the translation of performance ratings into amounts of salary over time.

Castilla established that this bias can be introduced at two points in the performance appraisal process. First, it can occur when a head of a unit (or head of supervisors) recommends to HR a specific salary increase amount for an employee using the performance evaluations received from the evaluating managers. The head of a unit may request a lower salary raise for an equally performing minority employee than for a nonminority employee, resulting in a lower average for minorities in reward recommendations going to HR. Second, bias also can occur when HR makes the decision to approve or reject a given salary increase recommendation made by the head of the unit, as HR may reject more minority rewards than nonminority rewards.

As large organizations may be unaware of the unintended consequences of such merit-reward practices at the workplace, increasing accountability and transparency in the performance-reward system will likely reduce bias, said Castilla. “We know from the social-psychological literature that accountability motivates decision makers to make fair decisions, which can help reduce judgmental biases,” he wrote. Also, “the timing of accountability is crucial, because accountability appears to be much more effective in preventing rather than in reversing biases.” In the organization he studied, unit heads were not accountable for their decisions regarding the amounts of salary increases.

He also found that while greater transparency reduces the incidence of bias, there are several reasons why lower levels of transparency continue to exist. For example, year-to-year salary increases for individual employees are generally not observable to the rest of employees and administrators, eliminating any salary comparisons among employees and potentially masking unfairness. Also, he wrote that “because most yearly salary increases are quite low, salary disparities among employees are so small that they are not noticeable overall.” In addition, because employees only tended to stay at the organization for 2.65 years, the shorter tenures minimized the long-term impact of the small differences in salary increases.

Castilla, author of Dynamic Analysis in the Social Sciences, stated that his findings demonstrate a critical challenge faced by employers who adopt merit-based practices to fairly reward and motivate their employees. “Ironically, although these merit-reward policies create the appearance of meritocracy, this study shows that the less formalized, less transparent, and less accountable stages of the performance appraisal process can actually create a greater opportunity for … bias to emerge, negatively affecting the fair distribution of rewards among employees in a way that is more or less invisible to everyone in the organizational setting,” he wrote.

Full Paper:

Related Links

For Media Inquiries

Paul Denning
Director of Media Relations
Tel: 617-253-0576
Fax: 617-253-5875

Patricia Favreau
Associate Director of Media Relations
Tel: 617-253-3492
Fax: 617-253-5875