Do monetary incentives, feedback and recognition matter for performance? Evidence from a field experiment in a retail services company

I use a field experiment in a retail services company to analyze the performance effects of providing monetary incentives, performance feedback, and recognition. I find that performance increases by 8 percentage points for sales reps who received only monetary incentives and by 12 percentage points for sales reps who received only recognition. These two effects are not statistically distinguishable from one another. The introduction of performance feedback does not have a statistically significant effect on performance. I further find that the percentage increase for the group that received both monetary incentives and recognition was 11 percentage points below the sum of the performance increases for the groups that only received one of the treatments. Hence, I find evidence that, because the two effects do not add up, monetary incentives and recognition are substitutes in this setting.;These treatment effects are heterogeneous in terms of the deciles of the performance distribution and demographic characteristics of the sales reps. I find that the treatments are most effective in the bottom deciles of the performance distribution, and that demographic variables, gender, age, and tenure, influence the effects of the different treatment conditions. I find that monetary incentives are effective for women but not for men, and vice versa for recognition. Older (more experienced) reps react more to monetary incentives and performance feedback and less to recognition, and vice versa for younger (less experienced) reps. Monetary incentives elicit a positive reaction from high tenure but not from low tenure reps, whereas recognition has a positive effect on the performance of low tenure but not high tenure reps.;Furthermore, I find that the performance increase due to the experimental intervention is reduced as the level of intrinsic motivation of the sales reps, measured in a pre- experimental questionnaire, increases. This negative interaction between the treatments and the intrinsic motivation variable is due mainly to the feedback condition. By comparing the pre and post scores of motivation, I find an increase in motivation for reps with a low initial level of motivation and who received good news throughout the experiment.