Rowan Shi

I am an Assistant Professor of Economics at Ryerson University in Toronto, Canada.

research interests

International trade, urban/spatial economics, computational methods


GitHub: rowanxshi
CV: web or PDF

working papers

Combinatorial discrete choice, with Costas Arkolakis, Fabian Eckert

Abstract We study combinatorial discrete choice problems in which an agent chooses items from a set and the return to each item depends on which others are chosen. We develop an approach to solve such decision problems and to aggregate optimal decisions across heterogeneous agents. In order to solve a single agent’s decision problem, we impose a restriction on the complementarities between items. To aggregate across heterogeneous agents requires an additional restriction on the complementarities between items and agent type. We argue that both restrictions naturally arise in many economic models. We present a generalized treatment of plant location and multi-stage input sourcing problems, highlighting the role that our restrictions play to solve and aggregate individual agents’ problems.

Julia Package

Solving combinatorial discrete choice problems in heterogeneous agent models: theory and an application to corporate tax harmonization in the European Union, with Kathleen Hu

Abstract This paper develops a solution method for computing optimal decisions to combinatorial discrete choice problems (CDCPs) in heterogeneous agent settings. With an arbitrary type distribution over any number of differentiated characteristics, it quickly computes the policy function mapping the entire type space to corresponding optimal actions. The binary decisions can display either supermodular or submodular interactions. Problems of this structure arise naturally ineconomic settings, especially in international trade and industrial organization. The proposed algorithm is particularly well suited for estimating or computing general equilibrium models incorporating heterogeneous agents solving CDCPs, including choices on plant locations, input sourcing partners, or export market entry. As an illustration of the algorithm in practice, the paper then turns to evaluating the effects of a counterfactual policy equalizing corporate tax rates across the European Union using a quantitative general equilibrium model where heterogeneous firms optimally select a set of countries in which to operate affiliates.

works in progress

Non-traded gains from trade: Evidence from Brazil, with Rafael Parente

Abstract We investigate the impact of trade shocks on the labor allocation within industries at the local labor market level. Using the Brazilian import liberalization of the 1990s as the empirical setting, we uncover a novel margin for the gains from trade: industrial reorganization among non-traded producers. We begin by showing empirically that local labor markets more exposed to the policy experienced more job reallocation within industries, both traded and non-traded, compared to those less exposed. Moreover, small establishments were less likely to survive compared to large establishments; among survivors, they were less likely to grow. To explain these empirical regularities, we provide reduced-form evidence that non-traded producers select into importing: plants in high exposure regions were more likely to start importing, with new importers originating from the middle of the size distribution but growing the most over the liberalization period. Motivated by these findings, we develop a parsimonious model of heterogeneous producers incorporating this mechanism. The theory implies that reallocation among non-traded producers is welfare-enhancing. In contrast, in a special case where all non-traded producers engage in importing identically, there is no reallocation within non-traded industries.