Replication Data for: The Emergence of the Poverty-Mental Health Gradient and the Great Pakistan Earthquake of 2005

Replication package for the forthcoming Journal of Development Economics paper by Andrabi, Daniels, and Das. Abstract. The relationship between poverty and mental health is of fundamental interest, both because of the equity implications of public investments in mental health and because mental health may be an important channel through which poverty traps persist. However consensus on this link has been difficult to establish, with substantial heterogeneity in the poverty-mental health gradient observed across studies. Taking the heterogeneity in the poverty-mental health gradient as our object of empirical interest, we use data collected four years after a devastating earthquake struck the Northern reaches of Pakistan to show that, in regions unaffected by the earthquake, there is no correlation between poverty and mental health. By contrast, a strong gradient between consumption poverty and mental health emerged in regions exposed to the earthquake. To the extent that this gradient can be interpreted causally, and we present a number of tests within the confines of our cross-sectional data in support of such a causal interpretation, our results suggest that associations between poverty and mental health gradient arise in populations subjected to large adverse shocks. The link between poverty and mental health is therefore crucially mediated by the extent of overall distress in the local ecology. Data. Five de-identified Stata .dta files (2,456 survey households; 10,538 mental-health respondents; 156,775 census individuals across 150 villages in four earthquake-affected districts of northern Pakistan: Abbottabad, Bagh, Mansehra, Muzaffarabad). Only variables actually referenced in the analysis do-files are retained. Codebooks (.xlsx) and integrity signatures (-sig.txt) ship with each .dta. Code. All Stata do-files that produce the paper's exhibits end-to-end (figures F1-F5, tables T1-T3, appendix A-D) plus vendored third-party Stata packages. Run do runfile.do; approximately 2 minutes on Stata 18 MP. Ethics. IRB exemption granted by Georgetown-MedStar IRB, study ID STUDY00002237. License. Code: MIT. Data: Restricted-access, governed by this deposit's access conditions.