The effects of QE and monetary policy on implied stock market volatility: a VAR approach

Abstract

This thesis aims to investigate the relationships between monetary policy and implied stock market volatility in the euro area. Both types of monetary policies (conventional and unconventional monetary policies) are considered though we put a particular highlight on QE when it comes to explore unconventional effects. Our work aims to contribute to the growing literature which identifies structural monetary policy shocks from (intra-daily) financial market reactions to central bank announcements. We postulate, as do Jarociński and Karadi (2020), that monetary policy announcements simultaneously disseminate information on monetary policy and the central bank's assessment of the economic outlook. Based on this assumption, we capture and distinguish the respective effects of monetary policy (conventional and unconventional) separately from this information shock by imposing sign restrictions on these reactions of financial variables to the ECB's monetary policy decisions. In addition to this, we refine and distinguish conventional and unconventional monetary policy by imposing sign restrictions on the slope as Goodhead (2019) does. The responses to each shock on implied stock market volatility are then analysed through a VAR model which contains different interest rates measures (short and long), macroeconomic indicators and related stock market variables (Vstoxx and Euro Stoxx 50). Although our results depend on the specification of the VAR and on the selection of variables, we can suggest from them that implied stock market volatility is particularly affected by the information that investors might receive from a central bank when it motivates its monetary policy decisions. In addition, our results tend to show some degree of pass-through where implied stock market volatility can notably be affected by stock prices and the level of interest rates. These results, although promising, should be taken with caution notably due to the low number of observations we have, their sensitivity to model specification and a possible misidentification of the structural shocks previously mentioned.