Journal articles

Refereed book chapters

Working paper and research in progress

Credit Lines in Microfinance - Evidence from a Field Experiment in India (with F. Aragon and K. Krishnaswamy), revise-and-resubmit, Journal of Development Economics

This paper studies the effect of flexible microfinance loans on small businesses’ performance. We use data from a field experiment that offered a credit line to female vendors in India. This new product retains several features of standard microfinance loans, such as joint liability and weekly meetings, but allows flexible withdrawals and repayments like a credit card. We find a positive incremental effect on vendors’ gross profits: on average, a credit line increases profits by 7 percent more than a standard microfinance loan. The effect increases with time passed since loan disbursal, rising to about 15% after 18 weeks. The observed increase in profits is driven by the flexibility of the credit line which also allows vendors to invest in riskier, but more profitable, goods. Our findings highlight the role of loan flexibility as a viable strategy for raising the impact of microfinance.

We build and structurally estimate a model of occupational choice between entrepreneurship and wage work. We explicitly distinguish involuntary entrepreneurship (running a business out of necessity) from running a business by choice. Involuntary entrepreneurs are agents who would earn higher income in wage work but cannot obtain a job due to a labor market friction. We estimate the model via the simulated method of moments using Thai urban data. Our results imply a 19% fraction of involuntary entrepreneurs among all businesses. Involuntary entrepreneurs earn much lower income (80% less) than the voluntary and are more likely among low-wealth and low-schooling households. We also quantify and distinguish the misallocations in occupational choice and investment resulting from labor and credit market frictions. Our results imply 17% excess (involuntary) entrepreneurs relative to the first best because of the labor market friction and 1% less entrepreneurs because of the credit friction. Evaluating counterfactuals using the estimated model shows that reducing the credit constraint has only minor impact on occupational misallocations but policies that relax either the labor or credit constraints (e.g., access to microcredit) yield sizable income gains for the poor.

Bogus Joint Liability Groups in Microfinance (with Y. Xue and X. Xing), revise-and-resubmit, European Economic Review

In a random sample of clients of CFPAM, the largest microlender in China, 73% of all joint-liability groups practice Lei Da Hu. That is, one person uses all group members' loans in a single project. We call such borrower groups `bogus groups'. The Lei Da Hu practice violates the key premise of group lending, that each borrower must use their loan in a separate project (what we call `standard group'). We extend the theory of group lending by analyzing the endogenous formation and coexistence of standard and bogus groups and characterize the efficient lending terms. The chosen group form depends on the borrower productivities and probability of success. Bogus groups are formed by heterogeneous borrowers, when the gains from larger expected output exceed the foregone default risk diversification. Accounting for bogus groups in their lending strategy can help MFIs raise productive efficiency and borrower welfare.

Distinguishing Across Models of International Capital Flows (with M. Wright), new version coming soon

We formulate and solve a range of dynamic models of international capital flows and risk sharing with imperfect capital markets. We feature both models of exogenously incomplete markets (debt with tax on borrowing or on capital outflows, non-defaultable debt) and models with endogenously incomplete markets (defaultable debt, limited commitment), as well as the complete markets benchmark. All models share common preferences and technology. We use computational methods based on mechanism design, linear programming, and maximum likelihood to estimate and statistically test across the alternative models of international capital markets. Our methods work with cross-sectional or panel data and allow for measurement error and unobserved heterogeneity. We study which models fit best and also what type of data (income, investment, capital, consumption, or all together) can be used to distinguish across the alternative models. Empirically, we use panel data on GDP, government expenditure, consumption, capital stock and investment per capita for 175 countries in 1993-2002. We find that, overall, the defaultable debt and autarky models fit the data best. The complete markets and limited commitment models are rejected in all estimation runs.

Economics of Crime Networks (with R. Dastranj and S. Easton), in progress

We study a network-based model of criminal activity. Agents' payoffs depend on the number and structure of links among them and are determined in a Nash equilibrium of a crime effort supply game. Unlike much of the existing literature that takes network structure as given, we analyze optimal network structures, defined as maximizing aggregate payoff. Using potential functions, we give necessary and sufficient conditions that guarantee the existence and uniqueness of equilibria with non-negativity constraints on effort. These results can be used to identify optimal networks for given cost and benefit parameter configurations drawing on graph theory and using a computational algorithm that searches over all possible non-isomorphic networks of a given size. Our results can be also used to study, via numerical simulations, the effects of alternative crime reducing policies on the network structure and crime level - removing agents, removing links or varying the probability of apprehension.

Blockchains, Collateral and Financial Contracts, in progress

Social Insurance and Status (with B. Xia), 2012

Development Dynamics with Credit Rationing and Occupational Choice, 2005

The paper presents a stylized general equilibrium model of a developing economy in which the wealth distribution, the interest rate, and the wage rate are endogenous and interact dynamically. A credit market imperfection stemming from limited commitment results in allocative inefficiency due to credit rationing and occupational choice constraints. Credit rationing is shown to persist as the economy develops. The proposed model is shown to match both general empirical regularities pertaining to developing economies and macroeconomic data from Thailand. Furthermore, wealth inequality in this setting may be detrimental for economic development, providing a rationale for redistribution policies.

This paper provides a step-by-step hands-on introduction to the techniques used in setting up and solving moral hazard programs with lotteries using Matlab. It uses a linear programming approach due to its relative simplicity and the high reliability of the available optimization algorithms.

Other publications