Working Paper: NBER ID: w29923
Authors: James Dzansi; Anders Jensen; David Lagakos; Henry Telli
Abstract: This paper studies the role of technology to improve tax capacity in the developing world. We focus on local property taxation in Ghana, a potentially significant but under-collected source of revenue. We randomize the use of a new technology designed to help revenue collectors locate property owners to deliver tax bills, which is a major challenge in many developing countries with incomplete property addressing. We find that the technology increases bill deliveries by 27 percent and, surprisingly, increases tax collections by 103 percent. To reconcile these experimental findings, we build and estimate a dynamic time-use model in which revenue collectors respond to the new technology by shifting their allocation of time toward learning about households’ propensity to pay, which is initially hard to observe, and subsequently collecting from those with highest payment propensity. The model’s predictions are consistent with additional experimental results on time allocations, knowledge and collector strategies. Our explanation highlights how a technology that is designed to relax one constraint ultimately relaxes multiple additional constraints once behavioral changes by users of the technology are taken into account.
Keywords: technology; tax capacity; local governments; Ghana
JEL Codes: H2; H71; O12; O33
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| increase in number of bills delivered (L87) | increase in tax revenues (H29) |
| new revenue collection technology (H26) | improved targeting of households (R20) |
| improved targeting of households (R20) | increase in tax revenues (H29) |
| new revenue collection technology (H26) | reduction in navigational challenges (R41) |
| reduction in navigational challenges (R41) | increase in time for information gathering (D83) |
| increase in time for information gathering (D83) | improved targeting of households (R20) |
| new revenue collection technology (H26) | increase in number of bills delivered (L87) |
| new revenue collection technology (H26) | increase in tax revenues (H29) |