Based on what you have learned throughout this course; summarize and critically analyze the work of Sebastian Sotelo titled ‘Domestic Trade Frictions and Agriculture’ and prepare it for submission for the Hypothetical Journal of Political EconomyAll Domestic Trade Frictions and Agriculture Sebastian Sotelo University of Michigan I a for th mous Calie Helle Salga in sem neap Unive Unive Burea this p cussio per w Electro [ Journa © 2020 use su I develop a model of agriculture on heterogeneous land to study the relation between trade, productivity, and welfare in Peru, where far- mers face high internal and external trade costs. I quantify the model with new data on crop prices, yields, and land allocations. I then mea- sure the effects of changes to trade opportunities. A policy of paving roads raises aggregate productivity (4.9%) and the median farmer’s welfare (2.7%), but increased competition from remote suppliers harms 20% of farmers. An increase in international grain prices spreads unevenly across regions, benefiting farmers but hurting urban consumers close to ports. I. Introduction In developing countries, a large majority of the poor live in rural areas. Their livelihoods are often tied to subsistence agriculture, limited by ma- jor barriers to trade, such as weak infrastructure, adverse geography, and m indebted to Sam Kortum, Nancy Stokey, Thomas Chaney, and Andrei Levchenko eir advice and encouragement. I also thank the editor (Ali Hortaçsu), three anony- referees, Fernando Alvarez, Costas Arkolakis, David Atkin, Saki Bigio, Lorenzo ndo, Kerem Cosar, Javier Cravino, Alan Deardorff, Jonathan Eaton, Jeremy Fox, Sara r, Erik Hurst, Gita Khun Jush, Joaquin Lopez, Sara Moreira, Ralph Ossa, Edgar do, Danny Tannenbaum, Adriaan Ten Kate, and Mike Waugh, as well as participants inars at the University of Chicago, Yale University, the Federal Reserve Bank of Min- olis, the Federal Reserve Bank of St. Louis, the University of Michigan, Penn State rsity, the Federal Reserve Board, Drexel University, the University of Toronto, Brown rsity, DePaul University, the Federal Reserve Bank of Philadelphia, and the National u of Economic Research for helpful comments and suggestions at different stages of roject. Special thanks to Alan Sanchez, Ben Faber, and Thibault Fally for detailed dis- ns. Maria Alejandra Zegarra provided excellent research assistance. Part of this pa- as written while I was visiting the Economics Department at Yale University. Funding nically published June 3, 2020 l of Political Economy, 2020, vol. 128, no. 7] by The University of Chicago. All rights reserved. 0022-3808/2020/12807-0006$10.00 2690 This content downloaded from 099.230.201.231 on July 25, 2020 07:00:20 AM bject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). domestic trade frictions and agriculture 2691 the spatial dispersion characteristic of rural populations. Not surprisingly, researchers and policy makers perceive the costs of domestic and interna- tional trade as a drag on incomes and productivity.1 Assessing the welfare and productivity effects of improvements in trade opportunities, such as infrastructure, requires an understanding of how individual farmers and consumers react to these improvements and how their choices interact in the aggregate. On one hand, policies that reduce trade costs can increase allocative efficiency and welfare by unlocking the forces of comparative advantage and the use of modern inputs. On the other hand, such policies also affect the equilibriumprices of crops, espe- cially those that are traded only domestically, thereby potentially harming producers who face increased competition. Because these crops usually constitute an important part of the diet of food buyers, improved ex- change opportunities can also affect welfare through their effect on con- sumption prices. In this paper, I develop a framework to measure the consequences of trade-related shocks to the agricultural sector and quantify it using Peruvian data. I start by documenting four facts about agriculture in Peru that guide my modeling choices. First, there is substantial price disper- sion across regions, which is indicative of domestic trade costs. Consistent with the presence of domestic and international trade, prices are also higher in urban areas, and the prices of export crops rise with proximity to ports. Second, farmers within a region allocate land tomany crops, and third, average revenue per unit of land varies substantially across crops. These last two facts suggest that land quality varies within regions and that land intensity differs across crops. Fourth, the quality of roads varies through- out the country, which produces spatial variation in trade costs. Informed by these facts, I then develop a quantitative model of special- ization and trade. In the model, farmers can grow various crops on land plots of varying quality. They can also trade their crops, at a cost, for in- termediate inputs, nonagricultural goods, and other crops in local, ur- ban, and internationalmarkets. Differences in land quality across regions and in land intensity across crops generate domestic and international 1 According to the World Bank’s (2007) World Development Report, as of 2002, 75% of the world’s poor were rural dwellers. The same report relates developing countries’ agricultural performance to within-country variation in access to markets and land quality (World Bank 2007, 54). Likewise, a recent Inter-American Development Bank report addresses how transport costs limit overall exporting activity: “High domestic transport costs can push exports to concentrate in just a few areas . . . while squeezing gains or simply locking out of trade large swaths of the country” (Mesquita Moreira et al. 2013, 3). from the Sjaastad Research Fellowship Fund is gratefully acknowledged. Finally, I thank Ministerio de Transportes y Comunicaciones de Peru and Ministerio de Agricultura y Riego de Peru for sharing their data with me. All opinions and any remaining errors are my own. This paper circulated previously as “Trade Frictions and Agricultural Productivity: Theory and Evidence from Peru.” Data are provided as supplementary material online. This content downloaded from 099.230.201.231 on July 25, 2020 07:00:20 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). 2692 journal of political economy All trade, while domestic trade costs discourage it. The resultingmodel is a hy- brid between a small open economy, which takes international prices as given, and a closed economy, in which prices are determined by regional trade within the country. The equilibrium features price dispersion and in- complete land specialization across regions, as in the data. Productivity and welfare, moreover, are determined by market access and comparative advantage. To quantify the theory, I construct a novel data set that combines sev- eral sources of data on Peruvian agriculture. I use government statistics on land allocation, production, and prices to estimate crop-specific land quality across and within regions. To estimate within-country trade fric- tions, I bring in a complete data set of the transportation system of Peru. Finally, I use disaggregated household survey data to estimate the elastic- ity of substitution across crops in consumption. I next conduct two sets of counterfactual experiments to understand and quantify how changes in trade opportunities affect welfare and pro- ductivity in the context of imperfect regional integration. First, informed by policy plans of the Ministry of Transportation of Peru, I simulate two infrastructure shocks that improve the transportation system and reduce domestic trade costs unevenly across regions. Second, I simulate a shock to international crop prices based on a World Bank scenario for the ef- fects of the Doha trade talks. There are two key points to understanding regional responses in these counterfactual scenarios. First, the initial production and consumption allocations govern the extent to which changes in prices translate into changes in productivity and welfare; these allocations, in turn, are driven by the interaction of comparative advantage and market access. Second, equilibrium prices respond endogenously and transmit shocks across re- gions. These price responses also depend on the substitutability of crops in production—which is linked to the degree of land heterogeneity—and in consumption. To understand, for example, how improved transportation directly im- pacts farmers’ productivity and welfare, note that when access to markets is costly, farmers pay high prices for their purchases and collect low prices for their sales. By reducing the cost of accessing domestic and foreignmar- kets, better roads allow farmers to specialize according to comparative ad- vantage and increase their use of imported intermediate inputs. The general equilibrium effects of this policy are more subtle. While infrastructure policy increases the price that a farmer gets for his prod- ucts by improving his own access to domestic and international markets, the policy also improves the access of farmers in other regions and in- creases the supply of crops to domestic markets, thus decreasing their price. Increased competition from remote suppliers therefore leads to re- duced prices for farmers who originally had better access to markets. This content downloaded from 099.230.201.231 on July 25, 2020 07:00:20 AM use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). domestic trade frictions and agriculture 2693 Using this quantitative model, I measure the aggregate and distribu- tional effects of two infrastructure policies. I find that a Ministry of Trans- portation plan to pave major roads increases productivity in agriculture, measured as a multicrop index of productivity (3.8% for the average re- gion and 4.9% in aggregate). Themedian farmer, moreover, experiences a 2.7%welfare gain. There is also substantial heterogeneity: the farmer in the 25th percentile of the welfare change distribution gains 0.1%, and the farmer in the 75th percentile gains 13.2%. Rural dwellers employed in the nonagricultural sector mostly benefit as a result of the policy, but their gains are limited in remote areas because consumption prices in- crease and their ability to export nonagricultural goods is limited. Along the same lines, a different policy of building new roads in targeted re- mote areas allows the median farmer to gain 0.02% of welfare, while far- mers at the 25th percentile of the welfare gains distribution lose 0.1% and those at the 75th percentile experience 0.25% gains. In both simulated policies, an increased supply of crops not traded internationally drives down the welfare of farmers who were originally well connected to urban markets.2 In the second counterfactual exercise, I simulate a shock to interna- tional crop prices based on a World Bank analysis of the potential effects of the Doha trade talks. This exogenous shock mildly increases the inter- national price of cereals and cotton. Unlike in a small open economy, however, international shocks are unevenly spread in regional markets because many Home regions do not trade every crop with the rest of the world. Although this is a relatively small shock, its general equilibrium effects are quantitatively important and lead to the opposite welfare pre- diction of a simple small open economy model in about 23% of regions. Proximity to ports makes it more likely that a region will trade with the rest of the world and therefore strengthens the transmission of interna- tional price shocks to regionalmarkets—including to crops that are traded only domestically. Among nonagricultural workers, those in regions close toports tend to losemore, because their consumptionbasketsmore closely reflect the price increase of high-consumption cereals. In addition to producing substantive results for Peru, this paper also makes three methodological contributions. First, the theory connects tightly with data on land allocations and productivity; hence, the model can be estimated based solely on agricultural and aggregate trade statis- tics, which are collected by many countries. This approach is especially useful for studying economies in which trade also occurs within borders, because it sidesteps the need to use domestic trade data, which are typi- cally unavailable. 2 Sizable welfare reductions are not uncommon, even though both policies reduce trade costs for all regions. The farmer in the 5th percentile of the welfare gains distribution loses 2.4% in the first policy and 0.7% in the second. This content downloaded from 099.230.201.231 on July 25, 2020 07:00:20 AM All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c). 2694 journal of political economy All Second, themodel treats each crop as a homogeneous good, instead of using the standardArmingtonmodeling device in which the same good is regionally differentiated. This approach allows for a simple analysis of the dissemination of price shocks across regional markets—using only data on initial consumption and land shares—and the elasticities of supply and demand. Moreover, when studying trade within a country, this treat- ment provides an alternative to regional differentiation, which, besides being implausible at high spatial resolutions, forces all regions to trade bilaterally in all goods. In contrast, the model in this paper generates sparse trade patterns—a well-known feature of trade data—using finite trade costs. Third, I obtain a simple estimating equation for the elasticity of land allocation with respect to relative prices, which is a key quantity that gov- erns adjustments to shocks. The estimating equation captures a basic eco- nomic intuition inherent to models in which production factors are het- erogeneous—namely, that factors are optimally allocated to their best use. Hence, increasing a crop’s land allocation reduces yields because it re- quires incorporating land less suited to that use. The strength of this effect is directly related to the heterogeneity of land. This parameter also governs all cross-elasticies in production—a practical compromise when the num- ber of parameters that can be estimated is limited by the availability of a few cross sections of data. Peru is an ideal setting for this study, because its geography is diverse and its agricultural sector resembles that of both developed and develop- ing countries. It is a middle-income country in which a few large urban markets are often the destination for traded agricultural produce but some well-connected regions produce for export markets.