Job Market Paper
Abstract: I study how a credit crunch affects output price dynamics. I build a unique micro-level dataset that combines scanner-level prices and quantities with producer information, including the producer's banking relationships, inventory, and cash holdings. I exploit the Lehman Brothers' failure as a quasi-experiment and find that firms facing a negative credit supply shock decrease their output prices approximately 15% relative to their unaffected counterparts. I hypothesize that such firms reduce prices to liquidate inventory and to generate additional cash flow from the product market. I find strong empirical support for this hypothesis: (i) firms facing a negative bank shock temporarily decrease their prices and inventory and increase their market share and cash holdings relative to their counterparts, and (ii) this effect is stronger for firms and sectors with high initial inventory or small initial cash holdings. To discuss the aggregate implications of these findings, I integrate this micro-level study into a business cycle model by explicitly allowing for two identical groups of producers facing different degrees of credit supply shock. The model predicts that a negative credit supply shock leads to a large temporary drop in aggregate inflation - as a result of the aggressive liquidation of inventory - followed by an increase in inflation as producers eventually run out of inventory. This prediction for inflation and inventory dynamics is fully consistent with observations for the 2007-09 recession.

Working Papers
Abstract: Existing empirical evidence on price-cost markup cyclicality is mixed. This paper finds that markups are procyclical unconditionally, and procyclical conditional on demand shock using a flexible production function. The estimated production function features a larger input complementarity than in a tightly parameterized production function (Cobb-Douglas and CES), leading to greater efficiency and higher markups during an expansion. These results have two striking implications: (i) much of the cyclicality in markups arises from input complementarity, rather than nominal rigidity, and (ii) the U.S. economy behaves as if it has increasing returns to scale.

Abstract: We study the business cycle with a Translog production function. We empirically identify a complementarity between labor and energy that leads to procyclical returns to scale, which is not compatible with the tightly parameterized production functions commonly used in the literature (Cobb-Douglas and CES). We therefore propose a flexible Translog production function that not only features complementarity-induced procyclical returns to scale but is also consistent with a balanced growth path. A simple calibrated business cycle model with the proposed production function generates strikingly data-consistent dynamics following demand shock without relying on either nominal rigidities or countercyclical markups. Our model also produces a stronger amplification effect than a model without complementarity. We then incorporate our production function into a benchmark medium-scale New Keynesian model (Smets and Wouters 2007) and repeat the business cycle accounting exercise. We find that input complementarity leads to a more dramatic decrease in the role of ''suspicious shocks" than of ''structural shocks."

Work in Progress
''The Margins of Trade and Inequality," joint with Jonathan Vogel (UCLA)
''Quantifying the Effect of Financial Development on Productivity and Misallocation," joint with Mauricio Larrain (Columbia) and Sebastian Stumpner (University of Montreal)
''Propagation of the Housing Market Shocks through the Production Networks," joint with Jungsik Hyun (Columbia) 
''Rising Global Corporate Saving: Drivers and Macroeconomic Implications," joint with Hiroko Oura (IMF)