Confidence and Decisions in Experimental Asset Markets [Website] [published version] (with Rasmus Pank Roulund), Journal of Economic Behavior and Organization.
This paper examines how traders' certainty and market certainty affect outcomes in an experimental asset market with known fundamental values. In this type of market, prices usually present large deviations from the fundamental value; in other words, bubbles are known to occur. We measure beliefs by asking participants to forecast the one-period-ahead price as a discrete probability mass distribution. We define certainty as the inverse of the dispersion of beliefs for each trader, and also create a market-wide measure of this to measure agreement across traders. We find that certainty affects price-formation and is also important in explaining the dynamics and size of the bubble. Moreover, as traders are successful they become increasingly more certain in their beliefs, even if these are on non-fundamental values, thus increasing the likelihood of price bubbles.
Behavioral Biases, the Current Account and Default, (Revise and Resubmit, MacroDynamics)
We include behavioral biases as in Barberis et al (1998) in a real business cycle model with default and trend shocks. Agents infer the trend process using misspecified mental models. The misspecification responds to two known heuristics: representativeness and conservatism, that leads to behavioral responses from the agents in the form of waves of optimism, pessimism, overreaction and underreaction. We present suggestive evidence from IMF forecasts that these biases are indeed present. The model can explain stylized facts found in developing economies, namely, high consumption to output volatility, counter-cyclical trade balance and frequent defaults. The model is observationally equivalent to other proposed explanations for these effects, but with potentially different policy implications.
After bubbles collapse, banks have often rolled-over debt at subsidized rates to insolvent borrowers or ``zombie firms.'' This paper explores the incentives to restructure debt in a game with risk shifting under debt overhang. We provide conditions under which it is privately optimal to zombie-lend even when it is socially inefficient. When a firm becomes insolvent, the firm loses access to competitive funding and its bank can exert monopoly power. The bank prefers to zombie-lend given that flowing funds for investment is not profitable due to risk shifting and liquidation entails costs. The model explains the inefficiency of traditional policies in the presence of zombies such as bank recapitalization and monetary policy and highlights the necessity of debt haircuts.
Asymmetric Effects of Trade and FDI: The Role of Country Size and Bridge Multinational Production. (with Lian Allub), submitted
Total gains from openness come through three channels: (i) trade, (ii) domestic multinational production (DMP), and (iii) bridge multinational production (BMP). We develop a quantitative theory to measure the effects of trade barriers and country size on the gains from openness through each of these channels. We show that these are sensitive to country size and larger in Europe than in South America. Country size is a crucial determinant of the contribution of each of these channels to the total gains from openness. DMP is more important in large countries, whereas the BMP channel is more important in small countries.
We study the effects of international flow volatility on the productivity of an economy and the desirability of a capital control. Productivity is endogenous, and depends on the allocations of resources. Entrepreneurs can decide either to do a small project (akin to services) or a big project (akin to manufacturing). Due to financial frictions they require some funding that they obtain via the international investors. International investors are of two types: short-term and long-term. Long-term investors are subject to a shock to its opportunity cost -- a sudden stop. Thus, there is expected volatility in the interest rate. This makes entrepreneurs decide on smaller or less productive projects, as to avoid the risk. Thus, an aggregate shock to the interest rate interacts with the financial friction to generate misallocation. A capital control has two effects. On the one one hand, it decreases the amount of liquidity of the economy. On the other hand, reduces the volatility of the interest rate. This increases productivity and thus increases the supply of funds from long term investors, thus partially offsetting the liquidity effect. The model is consistent with several facts regarding sudden stops. We explore the quantitative effects in a model of occupational choice with two sectors, and financial frictions as in Buera, Kaboski and Shin (2011), where we include international investors and aggregate uncertainty.
This paper studies the desirability of private debt haircuts in a simple model of heterogeneous firms and households. Households finance firm's working capital, and firms are credit constrained and differ in their debt levels. When there is an aggregate shock, less productive firms go bankrupt. This directly decreases the demand for labor and the wage receipts for households and indirectly decreases income from the defaulted loans to firms. The main result of the paper is that there is an optimal haircut for deposits such that both firms and families are better off. Moreover, there is a tension between maximizing welfare and maximizing output. This provides a rationale for the Cyprus, Greek and Argentinean experience. The quantitative exercise uses a model of heterogeneous firms and and matches the Argentinean devaluation episode of 2001 to an aggregate shock.
Work In Progress
Zombie firms and interbank linkeages: Evidence from Spain (draft coming soon)
A-Tree: A new o-Tree platform to run asset market experiments. (with Rasmus Pank Roulund)
Forecast.js: A Module for Measuring Expectations in Experiments. [Website] (with Rasmus Pank Roulund), submitted.
This paper presents a elicitation tool for economic experiments based on Harrison et al, 2017. The tool is user-friendly and enables subjects to forecast the movements of a continuous or discrete variable while addressing the main problems in eliciting beliefs. The module is easy to integrate with HTML-based experimental software kits, such as oTree (Chen et al., 2016)