Monetary policy and systemic risk

Abstract

This thesis investigates how and through which channels conventional U.S. monetary policy affects systemic risk from February 1990 to February 2020. Systemic risk is measured with a market capitalization-weighted ΔCoVaR framework applied to U.S. Global Systemically Important Banks, directly linked to two emphasized channels: the credit cost channel and the risk-taking channel. The study employs structural vector autoregressions with external instrument identification, using high-frequency monetary policy surprises around FOMC announcements to isolate exogenous policy shocks. This strategy is benchmarked against traditional Cholesky identification across specifications, including industrial production, consumer prices, the federal funds rate, and financial indicators such as the excess bond premium, National Financial Conditions Index Risk, and the CBOE Volatility Index. External instruments perform better than Cholesky, eliminating price puzzles and generating stronger, theoretically consistent responses. A contractionary monetary policy shock reduces systemic risk, with ΔCoVaR falling by 0.12 percentage points on impact and reaching −0.60 percentage points within two months. These effects persist for nearly three quarters. Variance decomposition shows that policy shocks account for up to 39% of systemic risk variation under external instruments, compared to less than 10% under Cholesky identification. Transmission occurs primarily through credit costs and risk-taking. The excess bond premium rises immediately after tightening, reflecting higher funding costs, while NFCIRISK and the VIX increase gradually, capturing broader market risk perceptions. Overall, contractionary policy shifts risk away from systemic exposures toward costlier credit and market risks, lowering interconnectedness while raising borrowing costs.