Market Volatility Transmission and Central Banking: What Happened during the Subprime Crisis?

ABSTRACT

We examine market volatility spillover during calm and crisis periods. First, we define endogenous and exogenous market volatility: endogenous volatility refers to the early part of uncertainty in the market, while, exogenous volatility is not fully anticipated and occurs as a result of decisions taken by actors and institutions. Endogenous volatility is captured by the mean of the GARCH-type process. We compare market reaction to central banking for two states: outside the subprime crisis and during the subprime crisis. We evaluate the effectiveness of central banking during the crisis. We used a Multivariate GARCH model with structural breaks in variance. Our main findings confirm the American market's impact on European markets, and changes in cross-market spillover during the crisis. The results show the effect of communications, meeting days and policy decisions of the Fed on world markets.

Keywords:

• MGARCH
• structural breaks
• central banks communication
• subprime crisis
• volatility

JEL Classifications:

• C5
• E5
• F3
• G1

Notes

¹Studies on the relationship between monetary policy and financial market also consider the impact of monetary surprise on financial markets. Monetary surprise represents a non-anticipated part of monetary action and can be considered as a non-anticipated part of monetary news. Our study does not concern monetary surprise but the information contained in monetary policy actions and central bankers communications. Readers can find more details about monetary surprises in, for example, Kuttner (2001), Bernanke and Kuttner (2005) concerning US monetary policy surprise and bond markets, Haldane and Read (2000) concerning the British case, Craine and Martin (2008) about the FED and the Australian Central Bank and Bohl et al (2008) for a study on the impact of the monetary surprise of BCE on European stocks markets.

²European Central Bank (ECB), Federal Reserve (FED) and Bank of England (BoE).

³Connolly and Kohler (2004) used an exponential GARCH process.

⁴Edwards (1998) used a GARCH model to investigate the national and regional impact of interest rate spread during the Tequila crisis. Park and Song (2000) applied a similar GARCH model to test volatility spillover across Asian markets during the Asian flu epidemic.

⁵Φ(L)=I_m−Φ¹L···−Φⁿ¹Lⁿ¹, where n₁ is the VAR process order defined by sequential LR test (see details in Appendix A).

⁶ The DCC models of Engle and Sheppard (2001) and Tse and Tsui (2002) have the advantage of a two-step estimation. However, these templates provide a linear structure to the correlation dynamics and impose a similar dynamic conditional correlation. In addition, these models do not permit variance spillover.

⁷The Federal Reserve, the European Central Bank and the Bank of England.

⁸We did not find any significant ‘day-of-the-week effects’ in mean equations. Estimates are available upon request.

⁹Shock persistence is the measurement of cumulative effects of shocks on volatility. The persistence in each market (x) is equal to ϕ_x=(α_x)²+½(γ_x)²+(β_x)². The ‘half-life’ is the time period in which volatility is expected to be halved: hl_x=ln(0.5)/ln(ϕ_x).

¹⁰All estimations are available upon request.

¹¹Unconditional variance measurements confirm the presence of high and low variance regimes (Table 4). The measurement of unconditional variance is . Estimates are available upon request.

¹²Estimates are available upon request.

¹³Table 9 presents the coefficient estimates of the effects of central bank communication, of executive committee meeting days, and of monetary policy decisions, and then the effects of interventions by the Fed, the ECB and the BoE during the subprime crisis.

¹⁴We take account of key dates, such as the adoption of the Term Auction Facility by the FED in December 2007.