Evaluating the impact of targeted decentralization on household consumption: Evidence from marginalized and privileged regions in Kenya
利用肯尼亚2010年宪法改革后的分权政策，研究发现定向财政转移支付显著增加了边缘化地区的家庭消费，且贫困家庭受益更大，但富裕家庭在教育医疗支出上增长更多。
This paper evaluates the impact of Kenya’s 2010 constitutional reform, which established 47 semi-autonomous counties and implemented a nationwide decentralization framework with redistributive fiscal transfers favoring 14 historically marginalized regions through a revenue-sharing formula and Equalization Fund. While decentralization is widely promoted as a tool for improving citizen welfare, evidence on its distributional consequences—particularly when exposure varies across regions—remains limited. Using household-level data from the 2009–2018 Kenya FinAccess Survey, this study exploits differential exposure to targeted fiscal transfers following the 2013 decentralization rollout to examine effects on six categories of household consumption: mobile airtime, education, household bills, medical expenses, rent, and family transfers. A difference-in-differences framework reveals significant post-reform increases in spending across several categories in more heavily exposed (marginalized) regions relative to less exposed counties. Quantile regressions show that these gains were unevenly distributed: poorer households experienced larger relative increases in total consumption, while richer households saw stronger growth in education and medical expenditures. Causal mediation analysis indicates that higher county-level budget allocations and household income partially explain these effects. A comprehensive set of robustness checks—including parallel trends verification, entropy balancing, falsification tests, alternative treatment definitions, and sensitivity to inference choices—supports the internal validity of the findings. Overall, the results suggest that targeted fiscal decentralization, when embedded within a nationwide reform, can reduce regional disparities while generating heterogeneous welfare gains across income groups.