Closer Look at Flight-to-Safety: A Mixed-Frequency VAR Approach (as of March 2024)

The paper investigates flight-to-safety from the US stock market to the US treasury bond market following a stock market uncertainty increase under a vector autoregression (VAR) framework. It proposes a relatively new VAR approach, mixed-frequency VAR (MF-VAR), to incorporate variables with different observation frequencies. This enables the inclusion of both macroeconomic and financial variables in the same VAR equation, which typically have different available data frequencies and different data generating process characteristics. With this new tool, the paper uncovers the fact that flight-to-safety from the stock to the bond market is extremely short-lived in the US. Additionally, it proposes a potential solution for similar future research related to macroeconomic causal inference with variables having different characteristics and data frequencies.

Monetary Policy and (Dis-)Inflationary Risk: A Quantile Local Projection Approach (with Ji Hyung Lee) (as of May 2024)

In this paper, we employ the local projection methodology at the quantile level to explore the predictive impact of moentary policy on the United States inflationary and disinflationary risks, as measured by Inflation-at-Risk (IaR) - the tail values of the future inflation rate distribution. Our findings reveal an increased level of uncertainty in future inflation rate following contractionary monetary policy surprises. The phenomenon is mainly driven by the lower tail risk (disinflationary or deflationary risk) than by the upper tail risk (inflationary risk), and the divergence persists for approximately 2-4 months after the initial shock. Our results from the empirical analysis suggests the importance of a deliberate control variable setting stage for similar research investigating quantile impulse responses through the local projection framework.

Are Regional Housing Markets at Risk after Tornadoes? A Panel Quantile Local Projection Approach (as of June 2024)

This paper introduces a novel integration of panel quantile regression with fixed effects and local projection methodologies to explore the dynamic impacts of tornadoes on regional housing market risks in the United States. This approach extends the application to include two-way fixed effects for both individuals and time, aiming to capture more comprehensive and accurate risk dynamics compared to previous studies. The empirical analysis reveals significant variability in the impact of tornadoes based on their strength and destructiveness. Weaker tornadoes are more associated with decreased housing demand, and stronger tornadoes are more associated with decreased housing supply. By applying two different groups of tornadoes as different shock measures in panel quantile local projection, I show that the weaker, demand effect dominant tornadoes lead to downward locational shift in regional house price growth rate distribution, whereas the stronger, supply effect dominant tornadoes lead to a dispersion in the distribution. These findings highlight the critical role of natural disasters, particularly the destruction of houses caused by them, as a potential key driver of increased risk level in the regional housing market. The methodological application and results of the paper offer valuable insights for policymakers and real estate investors, emphasizing the importance of considering natural disaster induced risks in housing markets.

Forecasting the Distribution of Inflation Expectations: Machine Learning Applications (with Humberto Martinez-Garcia)

Systemic and Individual Risks: How Do They Respond to Monetary Policy?