Ebook Decomposing European Bond and Equity Volatility
This paper contains an analysis of factors affecting the variances of national European bond returns as well as the variances of national European stock returns. We apply the so-called volatility-spillover framework. The variance of the unexpected return of e.g. the German bond market is divided into a part caused by idiosyncratic US (global) bond effects, US stock effects, European (regional) bond effects, European stock effects, and pure German (local) bond effects. Equivalently, the variance of the unexpected German stock return is divided into the same five effects in addition to pure German stock effects. Bond and stock markets are investigated simultaneously, which - we believe - is new to the volatility-spillover literature. Moreover, to the best of our knowledge, we contribute methodologically to the literature by generalizing the volatility-spillover model.
The empirical analysis brings some light on the integration of the European financial markets. Local effects should be weaker, the more integrated the European financial markets are. Financial integration appears to be a major concern of the policy makers in the European Union (EU) as the EU has launched several policy initiatives to obtain financial integration, cf. Hartmann, Maddaloni and Manganelli (2003). Presumably, the introduction of the euro has worked in favor of financial integration. The observed home bias in Europe has decreased in the previous years, cf. e.g. Baele, Ferrando, Hördahl, Krylova and Monnet (2004). This might indicate that the European financial markets have become more integrated. When the importance of country specific effects are low, the potential benefits of diversification are also small. It is believed that stock and bond volatilities are linked via information spillover, cf Fleming, Kirby and Ostdiek (1998) for a model and analysis of US stock, bond, and money markets. Fleischer (2004) extends this model to also include the equivalent Australian markets. Here we investigate the European bond and stock market volatility linkages. Finally, we examine how important global and regional effects are for the European bond and equity volatility.
We provide a new volatility-spillover model that covers both bond and stock markets simultaneously. Our model is derived from Bekaert and Harvey (1997), Ng (2000), Bekaert, Harvey and Ng (forthcoming), and Baele (forth coming). Ng (2000) and Bekaert et al. (forthcoming) divide the conditional variance of the unexpected stock return for country i into three effects; global effects, regional effects, and own market effects. The literature applies multistep estimation procedures, e.g. Ng (2000) applies two (and a half) steps: The first step specifies an ordinary bivariate GARCH model for the US and Japanese stock returns. As an intermediate step, the residuals from the first step are orthogonalized. In the last step, the US and Japanese orthogonalized residuals are applied as additional explanatory variables in univariate models for the national stock returns. The orthogonalized residuals provide the volatility spillover in that they make the variance of the unexpected return of the individual stock market a linear function of contemporaneous US idiosyncratic variance, Japanese idiosyncratic variance, and own market idiosyncratic variance.
We also apply a multiple-step estimation procedure: In the first step, a multivariate (dynamic conditional correlation) DCC-GARCH model for the US bond return, the US stock return, the European bond return, and the European stock return is estimated. In the second step, the residuals are orthogonalized. In the third step, the orthogonalized residuals are applied as additional explanatory variables in a univariate model for country i’s bond return, hereby providing volatility-spillover from the US bond, US stock, European bond, and European stock markets into the individual bond markets. In the fourth step, the orthogonalized residuals from the second step as well as the own bond residual from the third step are applied as explanatory variables in univariate models for the return on country i’s stock market. Thus, there is also volatility spillover from own bonds. The volatility-spillover effects are allowed to change (independently of each other) after the launch of the euro in the beginning of 1999. The model allows us to divide the conditional variance of the unexpected return of bonds (stocks) into separate proportions caused by the five (six) different effects mentioned above. To the best of our knowledge, we add to the literature model-wise; our model allows volatility spillover from four global/regional markets instead of just two as in previous models and we also allow the own bond market to influence own stock market.
We investigate nine European Union member countries’ bond and stock markets. We apply weekly data that cover the period from 1988 to 2003. We find that the conditional bond-stock correlations have gone from positive to negative in the last part of the sample. Before the euro, there is significant volatility spillover to the individual bond markets from US bonds, US stocks, and European bonds and after the euro only from the US bonds and European bonds. Before the euro, the main part of the conditional variances of the unexpected return on the bond markets are caused by aggregate European bond effects and own bond market effects. US bond market effects and US stock market effects are also fairly large. After the euro, the European bond market effects are strongest followed by US bond market effects. Own bond market effects have decreased dramatically. Before the euro, there is significant volatility spillover to the stock markets from US bonds, US stocks, European bonds, and own bonds and after the euro only from US stocks and European stocks. Before the euro, the own stock market effects and US stock market effects are the most important for the conditional variance of the unexpected return of the stock market.
The European stock effects and the US bond effects are small. After the euro, own stock market effects, US stock market effects, and European stock market effects are all strong. The results are to some extend influenced by the ordering of the variables in the orthogonalization. Our results indicate that bond (stock) market volatility is mainly influenced by bond (stock) market effects. This might suggest to analyze bond (stock) market variability separately from stock (bond) market variability. After the introduction of the euro, the local bond market effects have become smaller, whereas the local stock market effects are still sizeable. There appears to be room for further integration in the European financial markets, especially for the equity markets. Both global and regional effects are important for European bond and stock variances.
Download
PDF Ebook Decomposing European Bond and Equity Volatility
Posted in :
