A threshold VAR model of the propagation of U.S. financial stress to production and employment.

Author:Potts, Todd B.

    The performance of the U.S. economy over the last 4-5 years has served as a painful reminder of just how long-lasting and severe financial-crisis-induced economic downturns can be. Over the course of the Great Recession, the unemployment rate rose from under 5% to nearly 11%, and even though the recession officially ended in 2009, the rate remains over 8%, well above the long-run average for the U.S. economy. Financial crises are nothing new to the U.S. and other developed countries, but the fact that these increases in financial stress have led to such steep and persistent drops in real economic activity only in the current crisis and the Great Depression indicates that financial stress may propagate through to the real economy differently based on whether not a nation is in a "high-stress" vs. "low-stress" environment. Even before the recent financial collapse, researchers studied how adverse shocks to the financial system affect the business cycle. (Bernanke and Gertler, 1995) discuss how the credit channel is an important part of the monetary transmission mechanism and (Bernanke, Gertler, and Gilchrist, 1999) explain that the financial accelerator can exacerbate reductions in output because an adverse financial shock increases the potential riskiness of borrowers and reduces the flow of credit. Since the onset of the crisis, a proliferation of associated research has emerged. (IMF, 2008) points out that financial crises tend to generate longer-lasting and more severe economic downturns than those emanating from other adverse shocks. (Gilcrhist et al., 2009) examine how credit market shocks lead to economic fluctuations through changes in asset prices and (Gilcrhist and Zakrajsek, 2011) construct a credit-spread index to show how the excess bond premium can lead to substantial declines in real economic activity and equity prices. Other researchers have even developed indices that measure the degree of financial stress in an economy. Two of the most widely used are (Hakkio and Keaton, 2009), from the Federal Reserve Bank of Kansas City and (Cardarelli et al., 2011) from the International Monetary Fund. (Balakrishnan, 2009) produces a similar stress index for emerging economies. In addition to developing a financial stress index (FSI), (Balakrishnan et al., 2009) identifies "high-stress" periods as being when the FSI is over 1.5 standard deviations above its long-run average. Based on the downturns experienced in the Great Depression and the recent meltdown, it stands to reason that when the financial system is already highly stressed, further increases in financial stress may have more deleterious consequences than when we are in a relatively low-stress environment; but certainly, a more scientific approach in determining whether or not the financial system is highly stressed or not is preferred to relying on an a priori assumption. Fortunately, a number of methods have been developed for testing for the presence of such a structural break when the exact breakpoint is unknown. (Andrews, 1993) presents a technique applicable to single-equation estimation, and (Tsay, 1998) and (Hansen, 2000) outline multivariate threshold analyses.

    Recent papers have utilized such multivariate threshold analysis to study macroeconomic dynamics. (Galvao and Beatriz, 2011) use a version of Tsay's threshold VAR to examine how the monetary transmission mechanism differs depending on the current stance of monetary policy and (Afonso, et al., 2011) employ a threshold VAR incorporating...

To continue reading