Pay-for-Performance Failure

Definition

Merit pay and incentive plans apply the same extrinsic-reward logic that Kohn shows fails elsewhere, and he argues the workplace evidence bears this out: pay tied to performance has little or no demonstrated relationship to better organizational results, and where positive effects appear they are almost always confined to short-term, quantity-based measures on simple tasks — never to quality.

In the Book

Kohn surveys the empirical record directly. G. Douglas Jenkins Jr.'s 1986 review of twenty-eight studies of financial incentives found positive effects in 57 percent of cases, but nearly all measured short-term quantity on clerical or physical tasks (assembly work, tree planting, card sorting); of the five studies that measured quality, none found a positive effect. A study of welders found production dropped when an incentive system was abruptly removed — but then rose back to or above its prior level within months. A large meta-analysis (Guzzo et al., roughly 330 comparisons across ninety-eight studies) found no statistically significant overall effect of financial incentives on productivity, while training and goal-setting programs outperformed anything involving payment. In the public sector, the Civil Service Reform Act of 1978's merit-pay program was judged a disaster by its own director, and University of California researchers tracking twenty Social Security Administration offices over four years found merit pay "had no significant effects on organizational performance," using the very metrics the pay was based on. Kohn also notes that Japan and Germany — economic competitors held up as successful — rarely used incentive-based pay at the time, undercutting the claim that reward-driven motivation is simply human nature.

Why It Matters

This is a direct empirical challenge to the default assumption behind bonus structures, commission plans, and merit-pay systems: that tying pay to measured output improves that output. The pattern the evidence shows — real gains only on simple, quantity-measured tasks, and none or negative effects on quality or on knowledge work — gives a concrete filter for evaluating any proposed incentive scheme: what exactly is being measured, and is the underlying work the kind where "more, faster" is actually the goal.