In-kind incentives and health worker performance: Experimental evidence from El Salvador

Abstract

We experimentally evaluated the effects of in-kind team incentives on health worker performance in El Salvador, with 38 out of 75 community health teams randomly assigned to performance incentives over a 12-month period. All teams received monitoring, performance feedback and recognition for their achievements allowing us to isolate the effect of the incentive. While both treatment and control groups exhibit improvements in performance measures over time, the in-kind incentives generated significant improvements in community outreach, quality of care, timeliness of care, and utilization of maternal and child health services after 12 months. Gains were largest for teams at the bottom and top of the baseline performance distribution. We find no evidence of results being driven by changes in reporting or by shifting away effort from non-contracted outcomes. These results suggest that in-kind team incentives may be a viable alternative to monetary or individual incentives in certain contexts.

Keywords: El Salvador; Health services; In-kind; Pay for performance; Performance incentives; Team incentives.

Fig. 1

Team performance reports & certificates by treatment status. Notes: The top part of the figure presents a sample of the cover page of the performance report for each team. Highlighted in orange are the differences between the two. The bottom part of the figure presents a sample of the certificates of performance that were provided to teams that obtained 60% of more of the total points possible. The differences between the two are the amount obtained in treatment teams and the title of the certificate (a voucher in the case of treatment and just certificate for control). The identity of the teams was masked for confidentiality purposes.

Fig. 2

Distribution of team performance scores by treatment status and time period. Notes: The graphs present the kernel density of the distribution of the performance score of all teams by treatment assignment. The score is calculated as explained in Section 3 and was the one informed to teams during the period. The exact p-value of the Kolmogorov-Smirnov test of equality of the treatment and control distributions in each period is presented on the second column. During the 6^th and 12^th month follow-ups the experimental only treatment teams received incentives. At the 18^th month follow-up both treatment and control teams were eligible for incentives.

Fig. 3

Average standardized treatment effect of all contracted outcomes by category of baseline performance score. Notes: Categories of baseline performance score are constructed based on the sliding scale for incentives. The first category includes teams with a baseline performance below the threshold for incentives, i.e., 60. The second category includes teams just at or above the incentive threshold, i.e. 60–69 inclusive. The third category includes those teams with a baseline performance score of 70 or higher. The number of teams in each category is displayed in the figure label. Average standardized treatment effect of all eleven contracted outcomes, obtained as described in Section 5, using the model that pools both the 6^th and 12^th month follow-ups and with block effects and the baseline value of the outcome of interest in each regression. Confidence interval at 95% constructed from standard errors clustered at the team level, displayed on the red lines around each estimate.