Research
Research
This work explores a novel identification method in the setting of monetary policy. This method, called the negative control approach, borrows ideas from the conditional exogeneity and the IV methods but dispenses with the exogeneity assumptions while shifting the focus to the confounding mechanism. An appealing conceptual feature of this model is the fact that endogeneity can be exploited as much as exogeneity. Pioneered in Miao et al. (2018), the approach relies on the use of additional variables and a set of orthogonality conditions to construct a proxy of the endogenous component and block the omitted variable bias. This work is the first application of the negative control approach in the macroeconomic setting. In identifying the causal effects of the market interest rates on the macroeconomic outcomes of interest, we employ macroeconomic news and measures of Fed policy in a novel fashion as the additional variables satisfying orthogonality restrictions derived from the analysis of the information flow. Two key orthogonality conditions are used: (1) macroeconomic news only affects the Fed policy and the market rates by influencing the expectations of the unobserved state of the economy and (2) Fed policy only affects the macroeconomic outcomes through the influence on the market interest rates. A simple state space model supporting the reasoning is presented and the empirical application is benchmarked against the IV approach used in Bauer and Swanson (2023). Resulting estimates support the idea that the improper use of additional observables might underestimate the impact of monetary policy. This work also provides ample ground for further research in expanding the set of auxiliary variables and exploring nonlinear extensions.
Research In Progress
The macroeconomic dynamics at higher frequencies are analyzed in answering whether the dynamic in times of Covid has been different with respect to the past crises. To tackle this problem, we fit a dynamic factor model using macroeconomic data in conjunction with epidemiological data from the times of the pandemic. We find that the pandemic time dynamics were significantly different in comparison to other crises. An additional dynamic factor is required to better fit the uncommon dynamics. The application of the dynamic model with a Covid-specific factor allows to strip the data of the erratic behavior of the pandemic times by orthogonalizing with respect to the Covid dynamic factor and allow the researchers to use the adjusted data for macroeconomic analysis.
This work tackles the question of identifying and estimating US regional multipliers of federal spending using a rich data set of the government contract awards. Using the data description, a distinction is made between the contract signing time and the timing of the contract payment to ensure correct spending attribution. Variation across states, times, firms, and industries is employed to obtain the instruments for the federal government spending on the state level and estimate the fiscal multipliers. Several versions of shift-share instruments are constructed, including classical Bartik, fixed share approach, and the instrument based on heterogeneous spending.