Global financial crisis, liquidity pressure in stock markets and efficiency of central bank interventions

Abstract

In this article, we investigate the hypothesis of efficiency of central bank intervention policies within the current global financial crisis. We firstly discuss the major existing interventions of central banks around the world to improve liquidity, restore investor confidence and avoid a global credit crunch. We then evaluate the short-term efficiency of these policies in the context of the UK, the US and the French financial markets using different modelling techniques. On the one hand, the impulse response functions in a Structural Vector Autoregressive (SVAR) model are used to apprehend stock market reactions to central bank policies. On the other hand, since these reactions are likely to be of an asymmetric and nonlinear nature, a two-regime Smooth Transition Regression-Generalized Autoregressive Conditional Heteroscedasticity (STR-GARCH) model is estimated to explore the complexity and nonlinear responses of stock markets to exogenous shifts in monetary policy shocks. As expected, our findings show strong repercussions from interest rate changes on stock markets, indicating that investors keep a close eye on central bank intervention policies to make their trading decisions. The stock markets lead monetary markets, however, when central banks are slow to adjust their benchmark interest rates.

Notes

¹ Insofar as interest rate changes reflect shifts in monetary policy, we can note that between 1 January 2008 and 17 December 2008, the Fed lowered its main interest rates seven consecutive times from 4.25% to 0.25%. The Bank of England rate was reduced five times from 5.5% to 2%. The ECB announced two increases and three decreases in its interest rates with an interest rate on 10 December 2008 being equal to 2%.

² Results of unit root tests are not presented but may be obtained upon request.

³ For more details regarding the econometric analysis of VAR models, see Hamilton (1994).

⁴ The p order of a VAR model is determined using information criteria and autocorrelation functions.

⁵ See Hamilton (1994) for more details regarding the construction of impulse response functions.

⁶ See Van Dijk et al. (2002) for more details about the linearity tests.

⁷ The p and Q orders of GARCH models are also specified through information criteria and a general-to-specific estimation procedure of GARCH models.