Abstract: Developing countries face massive urbanization under weak property rights. Slum upgrading is a popular policy to improve shelter for many, but preserving slums at the expense of formal developments may entail future opportunity costs. We investigate these dynamic inefficiency concerns by estimating the long-term impacts of the 1969-1984 KIP program, which provided basic upgrades to 5 million residents in Jakarta, Indonesia. We assemble high-resolution data on program boundaries and current outcomes, including novel photographs-based slum indexes. Among historical slums, KIP areas today have on average 15% lower land values, 50% fewer high-rises, and are more informal, consistent with delayed formalization. A boundary discontinuity design yields similar results. Surplus calculations show heterogeneous opportunity costs, with 90% of the losses concentrated in half of the program areas, where land values are high. Elsewhere, KIP delivers sizable surplus. Our exercise informs the debate on whether to upgrade or formalize slums as cities expand.

Abstract: We develop a generalized neighborhood choice model allowing for heterogeneity in knowledge about amenities. When people sort into neighborhoods under imperfect information, all amenities are potentially endogenous. We construct a latent quality index to address this bias, using panel data from a neighborhood choice program that provided information about rents and same-school network to graduating students. Individuals switch into neighborhoods with larger networks and lower rents, and the effects persisted after graduation, influencing actual residential choices. Marginal willingness-to-pay estimates are $123 per month to live with a larger network, and not accounting for endogeneity will bias it up by 70%.

Abstract: We investigate how neighborhood preferences and choices changed one year after the beginning of the COVID pandemic. We study a Neighborhood Choice Program that helped graduating students choose where to live by providing new information about rents and amenities. Using panel data on neighborhood rankings before and after information, we find that changes in rankings favor neighborhoods where social and professional network shares are higher by 2.2 percentage points, rents are lower by $432, and are 2.4 kilometers farther from the city center. Interestingly, we did not detect this movement away from downtowns when the program was offered prior to the pandemic. We then estimate a neighborhood choice model to recover MWTP for amenities both before and after the pandemic. Our estimates reveal that MWTP for network shares post COVID is markedly lower than prior to COVID. Finally, we perform counterfactuals to quantitatively assess how changes in preferences affect where people live, and find that weaker network preferences are most impactful, while heterogeneity by commute and work-from-home are less relevant.