Testing for Government Bond Market Integration Case (Research Paper Sample)

Testing for government bond market integration. The case of the European Union
Andreea Avadanei
“Alexandru Ioan Cuza” University of Iasi, Romania
Faculty of Economics and Business Administration
Doctoral School of Economics
andreea_avadanei@yahoo.com
Abstract
The aim of this paper is to study the connections between euro and non-Euro government bond markets in the pre- and post- common currency launch period. We structured our research on chapters obeying the following order. Section 2 offers a brief review of the empirical literature on bond market integration within the Euro area. Section 3 presents the data and the methodological approach of the article. Section 4 illustrates the preliminary and main empirical results, while the fifth one analyses the findings in term of policy implications for both investors and policy makers. Section 6 concludes with some general remarks. The data used in this study are monthly interest rates on 10-year government bonds from ten Eurozone (EZ) member states (Austria, Belgium, Finland, France, Germany, Ireland, Italy, Netherlands, Portugal and Spain) and four non-Euro area countries (Denmark, Sweden, United Kingdom and United States) for the February 1990-January 2011 period. The sample period will be split into two subintervals, from February 1990 to December 2000 and from January 2001 to January 2011, in order to determine the different characteristics of the bond returns ex ante and ex post new common currency introduction in January 2001. The methodological approach will include the following steps. The first one entitles checking if each series is integrated of the same order, accomplished by the Augmented Dickey-Fuller test and Phillips-Peron tests. The second step is to test for cointegration to find out if bond market share a common lung-run path based on Johansen approach. Under the existence of a lung-run linkage, we should also analyze the possibility that short-term relationships among the bond markets might exist. This will be achieved via the Granger causality test. Our concluding remarks outline that EMU effects and the various changes and market actions led to a significant convergence in government bond yields since 2001.
Key words: economic and monetary union, financial integration, government bond market, cointegration analysis, Granger causality.
1. Introduction
The worldwide wave of financial market liberalization, facilitated by political, technological and financial advances, has led to increased interdependence among global equity and bond markets.
Financial integration indicators point out that the degree of integration in the euro area government bond markets has been very high since 1999. With the introduction of the common currency, government bond yields converged swiftly in all countries and yields became increasingly driven by common news. However, they also indicate that yields of sovereign bonds with similar, or in many cases identical, credit risk and maturity have not entirely converged. Differences in liquidity as well as in the availability of developed derivatives markets tied to the various individual bond markets may partly account for these spreads. Overall, the evidence shows that Eurozone government bond markets now exhibit a high degree of integration, albeit not as high as in the euro-area money market.
The reminder of the paper obeys the following order. Section 2 offers a brief review of the empirical literature on bond market integration within the Euro area. Section 3 presents the data and the methodological approach of the article. Section 4 illustrates the preliminary and main empirical results, while the fifth one analyses the findings in terms of policy implications for both investors and policy makers. Section 6 concludes with some general observations.
2. Literature review
In order to support empirical research on the connections between international bond markets, the academic literature offers a wide range of instruments and methodologies. One of the early studies investigating this issue is Ilmanen’s (1995), who used a linear regression model to assess the excess returns of long-term international bonds.
Its findings indicate a high degree of correlation, implying, in turn, international bond market integration. Similarly, Clare et al. (1995) highlighted the significance of global linkages in terms of portfolio diversification opportunities. Using variations of the GARCH modeling methodology, Capiello (2003) and Christiansen (2007) examined volatility spillovers across bond markets, noting increased effects between US and European segments, as well as a strong impact of euro introduction, which lead to almost perfect correlations among bond returns within the EMU member states.
Along the same line of research, Kim et al. (2006) studied the dynamics of bond markets in Japan, U.S. and four EU countries based on a bivariate EGARCH model involving time-varying conditional correlations. They results showed that economic integration has facilitated inter-bank integration, but the use of the euro has created investor uncertainty regarding the future evolution of macro fundamentals in the EZ countries, generating a “flight to government bonds”.
Other empirical research on government bond interest rate differentials focused on liquidity and default risk premiums. Based on panel data techniques to test indicators measuring the changes of the EU capital market integration, Adam et al. (2002) argued that the convergence in the Eurozone government bond markets has occurred before the introduction of the euro in January 1999. Baele et al. (2004) proved that bond yields were determined by common rather than local factors, which means a higher degree of market integration in the field after the launch of EMU third stage.
3. Data and methodology
The data used in this study are monthly returns on ten-years government bonds from ten Euro area member states (Austria, Belgium, Finland, France, Germany, Ireland, Italy, Netherlands, Portugal and Spain) and four non EZ countries (Denmark, Sweden, United Kingdom, United States), covering February 1990-January 2011 period. The time sample will be split into two subintervals, from February 1990 to December 2000 and from January 2001 to January 2011, in order to determine the different characteristics of the bond returns ex ante and ex post new common currency introduction.
The methodological design of our research will be comprised of the following steps. The first one entitles checking if each series is integrated of the same order, accomplished by the Augmented Dickey-Fuller test and Phillips-Peron tests. The second step is to test for cointegration to find out if bond market share a common lung-run path based on Johansen approach. Under the existence of a lung-run linkage, we should also analyze the possibility that short-term relationships among the bond markets might exist. This will be achieved via the Granger causality test.
The Augmented Dickey-Fuller test and Phillips-Peron test postulate that a series yt is assumed to be non-stationary under the null hypothesis. The Johansen test is a procedure for testing the cointegration of several time series. This test permits more than one cointegration relationships, so is more generally applicable than Engle-Granger test which is base don the Dickey Fuller (or the augmented) test for unit roots in the residual from a single (estimated) cointegration relationship. Johansen test employs two statistics, the trace and the maximum- eigenvalue. The trace test evaluates the null hypothesis that the number of distinct cointegration vectors is less than or equal to r (where 0<=r<- N and N is the number of variables) against a general alternative. The maximum eigenvalue test examines the number of cointegration vectors versus that number plus one. The critical values for these tests are given by Osterwald-Lenum (1992).
In the case of nonzero means, stochastic or/and deterministic trends, the cointegration equations may have intercepts and deterministic trends, so that it may not be the situation of a common χ2 distribution. Therefore, we need to formulate several assumptions regarding the trends underlying the series, as follows:
The level data yt have deterministic trends and the cointegration equations do not have intercepts;
The level data yt have no deterministic trend in data and cointegration equations have intercepts;
The level data yt have linear trends and the cointegration equations have only intercepts;
The level data yt and the cointegration equations have linear trends;
The level data yt have quadratic trends and the cointegration equations have linear trends.
Although for a comprehensive multivariate cointegration test we included all possible hypotheses, we select assumption number 3 as the most plausible one for the ensuing analysis.
Short-run relationships and/or causality are two issues not adequately captured by cointegration tests, so that we use a Granger causality test with an error correction term (is cointegration is found) or without an error-correction term (is cointegration is absent) to investigate these aspects.
4. Main empirical results
4.1. Descriptive statistics
Descriptive statistics for monthly returns of the 14 countries are presented in table no. 1. Ex ante EMU, EZ average returns vary from 1.8440% for Germany to 2.1839% for Portugal, and for non-EZ countries range from 1.8730% for U.S. to 2.0707% for Sweden. Analyzing post euro period we note much smaller values, Germany displaying the lowest value, of only 1.3426%.
Table 1: Descriptive statistics on monthly bond returns, pre- and post- euro period
(Feb. 1990-Jan. 2011)
Bond returnMeanStandard deviationSkewnessKurtosisPre-euro periodAT1.87150.2214-0.32922.2628BE1.92130.2538-0.20742.0918DE ...