Abstract
My dissertation is grounded in policy issues that affect rural economic growth in emerging markets. In a global economic framework, where growth and redistribution are par for the course, policies promoting rural development deserve particular attention. In the first two chapters, I explore how property rights establishment affects trade, deforestation and agricultural productivity in the Brazilian Amazon. My third chapter presents substantial evidence that crises propagate differently across rural versus urban areas in Latin America.
In my first chapter, "Do Property Rights Solve the Tragedy of Commons under Free Trade? Evidence from Brazil," I find that the titling scheme, Terra Legal, increases wood exports from the Amazon, contradictory to policy expectations of reducing deforestation in the short run. Following property rights establishment, wood exporting intensifies initially and peaks at the third quarter post programme rollout. Evidence further suggests that medium-sized farms, which obtain the rights cheaply, clear forest land to make way for livestock production. A 10% increase in the area registered by such farms in a municipality increases average export value of wood-based products (by 2.5%), as well as animal-based products (by 4.3%), per quarter, within a municipality. To estimate the effect precisely, I use confidential, municipal-level export data and control for municipal-product unobserved heterogeneity and quarter-year trend, relying on the scheme's phased rollout for identification. I also present a dynamic model of land allocation, which shows that while deforestation could increase in the transition to steady state following titling, the agent reduces deforestation in the long run and allocates more land to sustainable forestry in the new steady state. To the best of my knowledge, this paper is one of the first to explore the effect of property rights establishment on exports, both empirically and theoretically.
The results indicate that the effect from removing the fear of appropriation, which induces agricultural expansion at the expense of forests, dominates the effect from correcting the inefficiency of open access. In my second chapter, "Property Rights, Agricultural Productivity and Deforestation in the Brazilian Amazon," co-authored with Molly Lipscomb, we explore the agricultural channel more thoroughly. Property rights establishment may increase a farmer's investment in his land due to lower risk of expropriation, easier access to credit, and the ability to sell the land. Empirically, we show that the titling scheme does indeed increase credit access---both at the intensive and extensive margins; the smallest farms appear to be the main beneficiaries. A 1 percentage point (p.p.) increase in the area registered by small farms leads to a 30% in credit financing and a 5% increase in the number of credit contracts for investment purposes. However, we find no evidence indicating that the small farms engage in deforestation; they decrease their temporary crop cultivation, but do not increase their permanent crop cultivation. Increasing the area registered by medium-sized farms that obtain their land cheaply leads to an increase in the area deforested, as well as an increase in the rate of deforestation, by 0.3% and 4.4%, respectively, in response to a 1 p.p increase in their area registered. Finally, removing the threat of expropriation, along with increased access to credit, results in a reduction of total area cultivated due to a decreased cultivation of temporary crop cultivation. However, the largest properties increase their overall cultivation, particularly of cash crops and cocoa, though we cannot attribute this to the credit channel.
My final chapter, "Asymmetric Effects of Crises in Urban vs. Rural Areas in Latin America: A Study Using Nightlights," more broadly assesses the impact of shocks in rural Latin America and the Caribbean. Using nightlights observed from space as a proxy for real GDP and population density maps, I construct a very precise measure of economic activity at five-square-kilometer grid level, a level of disaggregation never used before. To estimate the effect, I run separate regressions based on population density and cluster the errors at the country-level, using the pairs cluster bootstrap-t procedure to deal with the problem of too few clusters in my sample. Rural/semi-rural areas see their real income growth fall following systemic banking crises and currency crises, with the effect ranging between 0.5% to 1% based on the subsample and type of crisis. Notably, I fail to reject the null for their urban counterparts. These findings indicate that capital markets are not integrated by population at this disaggregate level and that the assumption of perfect risk-sharing is strongly violated.
