Police Pay and Performance

"Police performance … declines sharply when officers lose arbitration cases. The per capita number of crimes cleared (crimes resulting in arrests) is 12 percent higher in the months following arbitration rulings in favor of police officers."

"Crime doesn't pay" may be a debatable axiom, but new evidence strongly suggests that the more crime-fighters are paid, the better they will combat crime. In Pay, Reference Points, and Police Performance (NBER Working Paper No. 12202), Alexandre Mas maintains that when police officers are awarded salaries below their desires and expectations, both arrest rates and average sentence length will decline, but when police receive their salary demands, arrest rates will rise.

Mas analyzes data on final offer arbitration rulings in compensation disputes between police bargaining units and major New Jersey cities from 1978 to 1996. (In 1977, New Jersey law mandated that wage disputes of police and fire fighters be submitted to final offer arbitration; about 9 percent of such disputes during the years of the study went to arbitration.) His dataset encompasses some 383 arbitration cases from 225 cities, and the study reveals that the employers won only 34 percent of their cases. The apparent "success" of the police bargaining units, however, may be explained by the possibility that union negotiators are more risk-averse than city negotiators and therefore submit more conservative pay demands.

What does seem clear from the researcher's analysis is that workers are unsatisfied not just with low wages, but with wages below a reference point that they consider fair. Mas uncovers just such a phenomenon when he compares data on the New Jersey arbitration awards and subsequent police performance.

The main measure of police performance in Mas's study is the number of crimes cleared by arrest monthly per 100,000 residents in a municipality. The number of arrests may be affected by the amount of overtime put in by police, by police absenteeism, or simply through the portion of the workday devoted to actual policing. Arrests represent costly effort in crime solving, in the arrest procedure, and in the subsequent paperwork. Police departments often base their own internal evaluations on the number of arrests. Therefore, Mas compares the average number of clearances in the months prior to arbitration of police wage disputes to the average in the months after arbitration.

Police performance, Mas finds, declines sharply when officers lose arbitration cases. The per capita number of crimes cleared (crimes resulting in arrests) is 12 percent higher in the months following arbitration rulings in favor of police officers. Felony arrests in cities where police unions lost in arbitration are also associated with lower incarceration probabilities and shorter jail sentences, suggesting that police may reduce their efforts and cooperation with prosecutors following arbitration losses. That is, the police expend less energy in gathering evidence, or at least in presenting evidence to prosecutors. In addition, police bargaining unit losses are associated with a 5.5 percent increase in reported crime rates in the months following arbitration rulings, suggesting less active policing.

Mas finds that the change in performance of New Jersey police officers depends not only on the amount of the pay raise but also on the counter-offer that was proposed but rejected. Comparisons of pay raises to counterfactuals, Mas says, influence police effort when the police lose in arbitration, but such comparisons are not relevant when police win, suggesting that these workers are subject to a form of loss aversion. The degree to which performance declines after an arbitration loss also depends on whether the loss was anticipated, suggesting that whether an arbitration decision is considered a win or a loss depends on employee expectations prior to arbitration. "On the whole," he surmises, "these results highlight the importance of managing and, in particular, lowering employee expectations prior to manipulating wage policy in organizations."

The author adds that it is well known that final offer arbitration awards are viewed as low quality because they are not the result of negotiations. He says his study shows that final offer arbitration can have additional inefficiencies in terms of the resulting response of participants to unfavorable judgments. In this regard, models of final offer arbitration can be developed to take into account the effect of differential rulings on productivity, for example, by employers managing worker expectations to minimize the behavioral costs arising from not meeting those expectations.

Mas suggests that additional work is needed to determine whether productivity responses to arbitration are exacerbated by the fact that the arbitration rulings in his study represent group-level outcomes. That is, are the effects of failing to achieve a reference point increased when the resulting disappointment affects an entire group of workers?

Finally, Mas says that additional studies should consider whether or not the behavioral responses associated with the differential arbitration outcomes observed in this study represent a general phenomenon relating to allocative mechanisms that clearly distinguish winners from losers, such as negotiations that involve a single and discrete high-stakes issue.

--Matt Nesvisky

The Digest is not copyrighted and may be reproduced freely with appropriate attribution of source.
NBER Videos

National Bureau of Economic Research, 1050 Massachusetts Ave., Cambridge, MA 02138; 617-868-3900; email:

Contact Us