PERSONALITIES AND PUBLIC SECTOR PERFORMANCE: EXPERIMENTAL EVIDENCE FROM PAKISTAN
Arman Rezaee summarizes his working paper Personalities and Public Sector Performance: Experimental Evidence from Pakistan , co-authored with Michael Callen, Saad Gulzar, Ali Hasanain and Yasir Khan.
In most developing countries, small public health clinics constitute an important source of primary health care. Doctor absenteeism is often a problem at these clinics. This absenteeism may be driven by extrinsic factors, such as poor financial incentives or political patronage systems, and/or intrinsic factors, such as a lack of motivation. And these extrinsic and intrinsic factors may relate—for example, a health worker’s intrinsic incentive to perform may substitute for weak or deficient extrinsic incentives.
Information bottlenecks are also a problem and can make it difficult to address absenteeism. If data about absenteeism does not make it up the chain of command, managers may lack the information needed to address the problem. But even if they have the data, managers themselves may be driven by similar extrinsic and intrinsic factors to not use this data.
Understanding the relationship between intrinsic and extrinsic motivation can help developing countries better write and target their health care policies. For example, a policy meant to extrinsically motivate by monitoring absenteeism may not be necessary when intrinsic motivation is high but may help a great deal when intrinsic motivation is low. Similarly, providing absence information to managers may only be successful when intrinsic motivation is high. Learning how intrinsic motivation interacts with extrinsic motivation will help us invest in policies that improve outcomes and save our resources when policies are not needed.
Field Experiment Designed to Learn More
We tested two possible health care policies using a field experiment in Punjab, Pakistan, where small health clinics are the primary source of healthcare for rural populations. We implemented both policies within the Health Department across Punjab’s 36 districts.
The first policy targeted health inspector absenteeism by increasing the monitoring of health inspectors. We switched a random half of Punjab’s health inspectors from using a paper form that had to be hand delivered to headquarters (where they collected dust) to using a smartphone app form that is transmitted in real-time to headquarters. Inspectors knew that this electronic form was time- and location-stamped, and required a picture as proof of all health workers reported as present. This increased monitoring served as a source of extrinsic motivation for inspectors to carry out their assigned inspections.
The second policy changed how district headquarters received inspection information from these health inspectors. All health inspection data collected via our smartphone technology were aggregated for each clinic and presented to district-level health officials on an online dashboard. To manipulate the salience of the information provided to these officials, clinics with three or more absentee staff were highlighted in red on the dashboard. Because this cutoff (i.e. three or more absences and two or fewer absences) was arbitrary and never communicated to officials, it can be used to understand the impact of information salience, an extrinsic factor, on decision maker performance in terms of doctor absenteeism in clinics under that officials’ supervision.
For both policies, we were interested in learning if and how the success of the policy targeting absenteeism depended on workers’ degree of intrinsic motivation. In the case of the time- and location-stamped real-time inspection form, we wanted to know whether the policy increased inspection rates equally for all inspectors or differentially based on inspectors’ levels of intrinsic motivation. In the case of the highlighted information on the online dashboard, we wanted to know whether district health officials’ use of this salient information to decrease doctor absenteeism depended on their levels of intrinsic motivation.
To measure intrinsic motivation, we administered well-established personality tests (the Big 5 Index and Perry’s Public Sector Motivation index), validated for the Pakistani context, on the universe of health inspectors and district health officials in Punjab and a large representative sample of clinic doctors. To measure absenteeism, we conducted unannounced visits to health facilities. We also measured collusion between inspectors and clinic doctors, in the form of health inspectors informing doctors ahead of inspections to guarantee attendance when electronic reports were filed. If a doctor was absent during an independent assessment conducted by the experiment team yet reported as present during every official inspector checkup, we considered collusion likely. Finally, we measured the responsiveness of health officials to information on under-performing clinics by assessing clinic doctor attendance in a follow-up 15-45 days after a high-absentee clinic was highlighted on the online dashboard.
First, smartphone monitoring has the largest impact on health inspectors with high Big 5 characteristics—a one standard deviation increase in the Big 5 personality traits index is associated with a 27 percentage point differential increase in inspections in response to increased monitoring. In other words, inspectors with more intrinsic motivation are much more likely to respond to increased extrinsic monitoring by decreasing absence.
Second, district health officials with high Big 5 characteristics are most likely to respond to a report of an underperforming clinic as measured by reduced doctor absenteeism at the facility—one standard deviation higher official Big 5 index is associated with an additional 40 percentage point reduction in doctor absence following an underperforming facility highlight in treatment districts. This means that more intrinsically motivated health officials respond to salient information on absenteeism by nearly guaranteeing subsequent doctor attendance.
There are four central implications of these results. First, intrinsic measures such as the Big 5 index could potentially provide useful diagnostics in public sector hiring, training, and promotion decisions. If, for example, a district health official with average intrinsic motivation is replaced by an official that is one SD higher (or becomes one), we could expect large results when that official has the information needed to do his/her job.
Second, improvements in performance may be achievable even in a system where everyone has the same weak extrinsic incentives to work. In Punjab, patronage is common, government jobs are not particularly lucrative, and there is little chance of being fired for being absent. Yet, more intrinsically motivated health workers and health officials respond positively to policy changes.
Third, personalities are a major predictor of which officials will use actionable evidence in making policy decisions. This is extremely promising since replacing a paper based form with an app and a dashboard is cheap as smartphones become common, is easily scalable, and is applicable to many different settings.
Last, heterogeneity in intrinsic motivation matters for the effectiveness of policy interventions. Even if the distribution of public workers is fixed in the short term, understanding the workers’ intrinsic motivation can allow for drafting better policies, for more cost effective targeting of policies, and even for knowing when a policy may be unsuccessful and should be avoided.
 Citation: Callen, M., Gulzar, S., Hasanain, A., Khan, Y., Rezaee, A., (2013) Personalities and Public Sector Performance: Experimental Evidence from Pakistan. Working paper. Chaudhury, Nazmul, Jeffrey Hammer, Michael Kremer, Karthik Muralidharan, and F Halsey Rogers, “Missing in action: teacher and health worker absence in developing countries,” The Journal of Economic Perspectives, 2006, 20 (1), 91–116.
 For further details on the field experiment, please see Callen, Michael, SaadGulzar, Ali Hasanain, and Yasir Khan, “The Political Economy of Public Employee Absence: Experimental Evidence from Pakistan,” 2013. Unpublished manuscript.
 It is important to note that there were some health inspectors that were rarely absent before our intervention, and thus a decrease in absenteeism was not possible for them regardless of increased monitoring or levels of intrinsic motivation. The same is true on the other extreme, there are some health inspectors that will never show up to work in any policy scenario. This highlights the fact that the results presented are Localized Average Treatment Effects (LATEs).
 We cannot causally isolate the mechanism by which district health officials increase doctor attendance. Anecdotally, they do not fire or transfer absent doctors but rather rely on stern phone calls or meetings. They also report spending more time monitoring clinics themselves.
 There is some evidence that personality traits are malleable. For example, see Roberts, Brent W., Kate E. Walton, and Wolfgang Viechtbauer, “Patterns of Mean-Level Change in Personality Traits across the Life Course: A Meta-Analysis of Longitudinal Studies,” Psychological Bulletin, 2006, 132 (1), 1–25.
Arman is a PhD Candidate in economics at the University of California, San Diego. Before starting his PhD, he obtained a Master’s in Public Policy from the Harvard Kennedy School and a Bachelor’s from the University of California, Berkeley. His primary research aims to strengthen governance and government service provision in developing countries through information communication technology (ICT) interventions, in fields such as health, education, and livestock. He also conducts research on the historical causes and long-term impacts of institutions, and he is interested in a broad range of political economy issues in developing countries. Arman lives in Berkeley, CA with his wife and 12-month old daughter.
Disclaimer: The views expressed here are those of the authors and do not necessarily represent the views of the Institute of Development and Economic Alternatives