Essays in Macroeconomics
Author(s)
de la Barrera i Bardalet, Marc
DownloadThesis PDF (2.364Mb)
Advisor
Werning, Iván
Schmidt, Lawrence D.W.
Terms of use
Metadata
Show full item recordAbstract
The thesis consists of three essays on macroeconomics. In the first essay, I study the price and wage-setting implications of monopsony models with nominal rigidities. I develop a New Keynesian model with wage posting and on-the-job search. I show how wage markdowns are related to the importance of hiring costs, and those are estimated to be an order of magnitude larger than previous calibrations. I show how at the individual level, both higher monopsony power and higher wage rigidity amplify the price response of idiosyncratic demand shocks. At the aggregate level, the main driver of inflation is not an increase in real wages but rather an increase in the cost of hiring workers. Given that firms have problems finding workers, they raise prices. In a calibrated model, I show how negative labor supply shocks reduce the real wage when the nominal wage increase is offset by the nominal price increase. In the second essay (joint with Marc de la Barrera and Masao Fukui), we study how the interaction between China’s productivity growth and currency peg to the US dollar affected the labor market and trade imbalance in the United States. Empirically, we document that in response to similar exposure to Chinese exports, countries pegging to the US dollar experienced larger unemployment and trade deficits compared to floating countries. Theoretically, we develop a dynamic model of trade featuring endogenous imbalances and nominal rigidity, and show that Foreign growth may hurt Home welfare and characterize optimal trade and monetary policy in this environment. Quantitatively, we find that China’s currency peg is responsible for 447 thousand manufacturing jobs lost in the US over 2000-2012, one third of the total US trade deficit over the same period, and reduced US lifetime welfare gains from Chinese growth by 32% compared to an economy where an otherwise identically growing China had its currency floating. A short-run safeguard tariff may have effectively accommodated China’s currency peg and ameliorated the labor market distortions. In the third essay (joint with Tim de Silva), we explore a novel field that uses machine learning techniques to solve dynamic stochastic optimization problems. While most traditional approaches require the knowledge of a law of motion for exogenous states like income, we show a methodology that allows us to remain agnostic about the data-generating process of the state. Instead of calibrating a model mimic the dynamics of the state, we need to ob2serve realizations of such state. Parametrizing the policy function with a neural network, we are able to solve the value function problem without ever knowing the law of motion of the state, which the neural network endogenously learns. We test our approach with the income fluctuations problem and show how our methodology is able to learn the income process when it is an AR(1), and is also able to solve the problem for an unspecified income process. We then compare the welfare loss of specifying a particular income process and evaluating the policy function without making any assumption on the income process, and we find that the miss optimization loss is negligible. A byproduct of this project is the publication of the python package nndp that is available for use and solve a wide array of finite horizon, dynamic stochastic optimization problems.
JEL Classification C63, E24, E32, F16, F31, G51, J63
Date issued
2024-05Department
Massachusetts Institute of Technology. Department of EconomicsPublisher
Massachusetts Institute of Technology