Working Paper: NBER ID: w26388
Authors: Thomas Bossuroy; Clara Delavallade; Vincent Pons
Abstract: Developing countries increasingly use biometric identification technology in hopes of improving the reliability of administrative information and delivering social services more efficiently. This paper exploits the random placement of biometric tracking devices in tuberculosis treatment centers in urban slums across four Indian states to measure their effects both on disease control and on the quality of health records. The devices record health worker attendance and patient adherence to treatment, and they automatically generate prompts to follow up with patients who miss doses. Combining data from patient and health worker surveys, independent field visits, and government registers, we first find that patients enrolled at biometric-equipped centers are 25 percent less likely to interrupt treatment—an improvement driven by increased attendance and efforts by health workers and greater treatment adherence by patients. Second, biometric tracking decreases data forgery: it reduces overreporting of patient numbers in both NGO data and government registers and underreporting of treatment interruptions. Third, the impact of biometric tracking is sustained over time and it decreases neither health worker satisfaction nor patient satisfaction. Overall, our results suggest biometric tracking technology is both an effective and sustainable way to improve the state's capacity to deliver healthcare in challenging areas.
Keywords: biometric tracking; healthcare provision; data quality; tuberculosis; randomized controlled trial
JEL Codes: I12; I18; O15; O33
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| biometric tracking devices (Y90) | reduction in treatment interruptions (C22) |
| biometric tracking devices (Y90) | increase in health worker attendance (I15) |
| biometric tracking devices (Y90) | increase in follow-up home visits (I15) |
| biometric tracking devices (Y90) | decrease in data forgery (K24) |
| biometric tracking devices (Y90) | reduction in overreporting of patient numbers (C83) |
| biometric tracking devices (Y90) | reduction in underreporting of treatment interruptions (C22) |