- Applied Time Series (Macroeconometrics)
Click here to view my research summary.
"The Time-Varying Response of Hours Worked to a Productivity Shock" (Job Market Paper) Appendix
Macroeconomists appear to have not reached a consensus about the response of hours worked to a total factor productivity (TFP) shock. This paper updates evidence on the TFP-hours worked relationship by extending Galí (1999). A TFP shock is identified as the only source of variation in average labor productivity in the long-run. I estimate a structural vector autoregression (SVAR) that includes time-varying parameters (TVPs) and stochastic volatility (SV) on average labor productivity and hours worked. The estimation follows Canova and Perez-Forero [2015. Estimating Overidentified, Non-recursive, Time-Varying Coefficients Structural VARs. Quantitative Economics]. This algorithm produces structural intercepts and slope parameters that are consistent with the long-run restriction. I find evidence in support for RBC and New Keynesian theories. The impulse response functions of hours worked with respect to a TFP shock are negative on impact and at the business cycle horizons. Results from forecast error variance decomposition show TFP shocks dominate the fluctuations of ALP. Time variation in the slope parameters and SV of the ALP regression are responsible for the conflicting evidence in the literature. The reasons for the breaks are several episodes of structural change, such as the productivity slowdown and the Great Recession.
"The Time-Varying and Volatile Macroeconomic Effects of Immigration"
This paper studies the impact of immigration on the U.S. macroeconomy. I develop novel identification schemes to estimate structural VARs. The identification begins from Galí (1999). He identifies a productivity shock by assuming it is the only shock to drive average labor productivity in the long-run. I add immigration and consumption to the SVARs. Identification of immigration shock requires short- and long-run restrictions. I impose neutrality on immigration with respect to all other shocks in the short-run while allowing labor productivity and hours worked to respond to an immigration shock at impact. In the long-run, I consider two alternatives for whether labor productivity is neutral to an immigration shock. Consumption is added to the SVARs to identify a permanent income shock. The permanent income shock is included because the immigration decision also respond to prospects for lifetime earnings. The data favors an identification scheme with immigration neutrality on impact and average labor productivity responds only to productivity and immigration shocks in the long-run. My estimates show productivity and hours worked decline while consumption rises in response to the immigration shock at the business cycle horizons. Immigration's impact on the economy is state-dependent, while the response of immigration to macroeconomic shocks has been stable over time.
Work in Progress
"An Alternative Method for Estimating Stochastic Volatility in Structural VARs"
(with James M. Nason)
Stochastic volatility has become a standard element when estimating SVARs. The standard approach builds on Primiceri (2005). He uses an algorithm from Chib et al. (2002) which assumes SV can be well approximated by a mixture normal distribution. This introduces an additional hidden state variable into the Markov Chain Monte Carlo (MCMC) sampler. More recently, Monfort et al. (2015) propose to estimate SV models with a quadratic Kalman filter. They replace the hidden state with squared residuals of the SV model. The QKF approach reduces bias and improves efficiency, according to Monfort et al. (2015). In this paper, we show that the QKF can be incorporated to estimate SV in SVARs. We do this by revising a Metropolis-within-Gibbs sampler developed by Canova and Perez-Forero (2015). We aim to provide Monte Carlo evidence that QKF reduces bias and is more efficient statistically and computationally than the standard approach. The paper concludes with an empirical example illustrating the benefits of the QKF for estimating SVARs with SV.