تجزیه عامل و تنوع در بازار اوراق قرضه شرکت های بزرگ اروپا
|کد مقاله||سال انتشار||تعداد صفحات مقاله انگلیسی||ترجمه فارسی|
|15112||2013||20 صفحه PDF||سفارش دهید|
نسخه انگلیسی مقاله همین الان قابل دانلود است.
هزینه ترجمه مقاله بر اساس تعداد کلمات مقاله انگلیسی محاسبه می شود.
این مقاله تقریباً شامل 11773 کلمه می باشد.
هزینه ترجمه مقاله توسط مترجمان با تجربه، طبق جدول زیر محاسبه می شود:
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of International Money and Finance, Volume 32, February 2013, Pages 194–213
In this paper we present an analysis of diversification strategies on portfolios of European corporate bonds. From the perspective of a US-based investor we study whether mean–variance diversification strategies change as a result of the introduction of the European Economic and Monetary Union (EMU). Using a comprehensive and unique data set of European corporate bonds we show that country factors are more important than industry factors to describe the cross-section of European corporate bonds. In particular we find that in the Post-EMU period country factors remain important.
The central question of this study is whether a country or an industry diversification strategy performs better in European corporate bond markets. The answer to this question is of more than an academic interest, with potentially strong implications for the practice of bond fund management in Europe. A related issue that we address is whether diversification strategies change around the start of the European Economic and Monetary Union (EMU) and the introduction of the Euro in 1999. Galati and Tsatsaronis (2001) document a surge of corporate bond issuance in Europe in the immediate aftermath of EMU. If indeed, an industry sector allocation strategy in the European corporate bond markets is to be preferred, then this must be based on changing diversification benefits in portfolios of various allocations Pre- and Post-EMU, as can be discerned ex-post from realized returns. There is anecdotal evidence that fund managers of European bonds were intent on a shift from a ‘country’ to a ‘sector’ allocation just prior to EMU. Pieterse-Bloem and Lamedica (1998) report from a survey that 85% of European funds who anticipate changes in their EMU non-sovereign holdings, 70% expect an increase of the share of European corporate and bank bonds by an average of 21%. To our knowledge ours is the first study that analyzes diversification strategies for European corporate bonds over a long period. Unlike for equities, where the study of return volatility, correlations and comovements, their time-varying properties, factor determination and benefits for international diversification enjoys a long-standing history in financial economics, similar studies for bonds are missing. This is probably in some part due to the more esoteric nature of bond markets compared to equity markets and the fact that data for corporate bonds are not as readily available to academic researchers. Eurobonds are mostly traded over-the-counter as opposed to stocks, which are exchange-traded and for which prices are officially recorded, time-stamped and publicly available. We overcome these impediments and hand-collected a database of 4587 of European corporate Eurobond returns over the period from January 1991 to March 2008, covering nearly a decade before and a decade after the inception of EMU in 1999. Our contribution to the research field is not to provide a new methodology to study the presumed allocation shift for bonds, but rather to draw on earlier research conducted for equity markets.1 The benchmark study in this field is Heston and Rouwenhorst (1994), who propose a method for decomposing a set of equity returns into country and industry factors. These authors analyze whether the benefits of European equity portfolio diversification stems predominantly from a pure country or industry selections of stocks. The maintained hypothesis is that the relative impact of country factors should decline in favour of industry factors after the introduction of EMU. Heston and Rouwenhorst (1994) and many other studies typically conclude that country effects generally explain a larger portion of the cross-sectional variation in stock returns.2 Several studies do detect a rise in the relative importance of industry factors during the late 1990s, e.g. Galati and Tsatsaronis (2001), Flavin (2004), Ferreira and Ferreira (2006) and Phylaktis and Xia (2006). The rise in importance of industry effects versus country effects has to date not been confirmed as a long-term trend though. From a methodological perspective it is important to note that many of the studies in the earlier literature impose the assumption that country and industry factor loadings are fixed and constant over time. More recent studies overcome these limitations by relaxing either or both assumptions of unit and constant factor loadings. For example, Brooks and Del Negro (2004) use a latent factor model that varies exposures of stocks to country, industry and global shocks by virtue of which they demonstrate that the nested model of Heston and Rouwenhorst (1994) is less well supported. The assumption that firm returns have very little to no dynamic loadings to country and industry factors contrasts with evidence from Bekaert et al. (2009) who propose a dynamic arbitrage pricing theory model. Also Baele and Inghelbrecht, 2009 and Baele and Inghelbrecht, 2010 relax the assumption of constant factor loadings by proposing a generalized autoregressive conditional heteroskedasticity (GARCH) framework that allows both factor exposures and asset-specific volatilities to vary over time. While it is entirely plausible that models that overcome the restrictive assumptions of unit and constant factor loadings provide a better fit, they continue to rely on linear factor specifications. Consequently, results from these studies have by and large left the overall conclusion of the dominance of country factors over industry factors in stock return variation intact. Baele et al. (2010) focus on the time-varying features of correlation and covariance structures in country factors. Other studies from this branch, e.g. Hardouvelis et al. (2006), show that distance, information, common institutions and macroeconomic factors play a role in the explanation of equity return comovements between countries. Cross-fertilization from the existing equity-based literature on return decompositions and time-varying return distributions to the sphere of bonds has to date been rare. Baele et al. (2004) borrow from the literature of equity return decomposition models to devise a measure for the integration of corporate bond markets. Following Heston and Rouwenhorst (1994), they focus on the identification of country components in individual corporate bond yield spreads over government bonds. However, rather than decomposing the returns into country factors directly, the country component is defined as the residual from a second step after differences in specific risk characteristics of corporate bonds (such as maturity and rating) are filtered out. Baele et al. (2004) find for a sample of Euro zone corporate bond yields for 1998–2003 that the country-specific spread is statistically significant but relatively small in economic terms. Varotto (2003), whose study of corporate bond returns stays much closer to the standard Heston and Rouwenhorst (1994) decomposition model, finds that country factors dominate in corporate bond spread returns followed by maturity, rating and, lastly, industry factors. While Varotto’s (2003) result of dominating country factors in the cross-section of corporate Eurobond returns is substantial, it only provides part of the answer to our central research question for two reasons. First, results are based on a sample from 1993 to 1998, which ends before the introduction of the Euro. Second, even if country factors are identified as the dominating source of risk variation in corporate bond returns prior to EMU, care needs to be taken with the conclusion that a country-based allocation strategy would have yielded better results. The Heston and Rouwenhorst (1994) decomposition methodology determines in essence the extent to which separate factors explain return variation, but little more in terms of portfolio construction can be deferred from it. We therefore adopt a different methodological framework as our starting point and focus on explicit statistical tests that allow us to analyze which diversification strategy, based on either country or industry factors, yields better results in a mean–variance framework for European corporate bonds. To this end we apply spanning and efficiency tests in our empirical analysis. Again spanning and efficiency studies so far have been performed predominantly on equity returns. These studies have produced mixed results when applied to diversification strategies in Europe. While Moerman (2008) finds that a stock investor in Europe is better off diversifying over industries rather than over countries in the years between 1995 and 2004, Eiling et al. (2006) find that country- and industry-based portfolios cannot be distinguished in terms of mean–variance performance and Sharpe ratios for roughly the same period. In this paper, we follow similar lines on our data set of corporate Eurobonds. The outline of this paper is as follows. In Section 2 we present our database. Section 3 contains a description of the empirical methodology. In Section 4 our empirical results are presented and concluding remarks are in Section 5.
نتیجه گیری انگلیسی
This paper studies the factor decomposition and benefits of country versus industry diversification in European corporate bond markets. In our empirical analysis of corporate bond returns between January 1991 and March 2008 we borrow from the literature on international diversification of equity portfolios in assessing mean–variance performance. We use spanning tests to determine whether the mean–variance performance of country or industry portfolio indexes can be improved by adding assets from the opposite set, and efficiency tests to determine whether the maximum Sharpe ratios of country of industry indexes are statistically different from each other. We base our analysis on two types on index construction methods. The first set of portfolios is built from the direct country and industry classifications of each individual corporate Eurobond in our sample. We perform the spanning and efficiency tests on the inclusive set of country and industry indexes and on a set that excludes the overlapping components as these could potentially induce inherent covariance among the country and industry indexes. From the spanning tests we find that for value-weighted portfolios the mean–variance performance of both country and industry portfolios can be improved by adding industry and country indexes to the portfolio, respectively. This result applies to the full sample period and to the period before EMU inception. Both the decision to incorporate bonds in a value-weighted rather than an equal-weighted basis and the decision to remove overlapping components leads to a loss of diversification opportunities. In the Post-EMU era, both country and industry portfolios are mean–variance efficient, implying that investors could choose either one to construct a well-diversified portfolio. The second index construction method we consider in this paper is based on the decomposition methodology of Heston and Rouwenhorst (1994). We cannot reject the hypotheses that the correlation structures of the decomposed country and industry index returns differ significantly from correlation structures of the direct country and industry indexes. We find, in accordance with previous studies in equity markets, that after the decomposition of our European corporate bond returns, country effects dominate industry effects. The dominance of country factors rises significantly from the Pre-EMU to the Post-EMU era. While country effects dominate industry factors Pre-EMU by a factor of 1.6, they do so by a factor of 6.4 after EMU. A closer study of the variation of country effects among the countries included in the data set reveals convergence has taken place within the core countries of EMU but not with the UK and Sweden. The fading of industry effects leads us to conclude from our numerical results that industry dispersion is what likely seems to have taken place under EMU in Europe, rather than industry specialization. Using the decomposition-based country and industry portfolios, we find that spanning is rejected less often than in the case of portfolios directly constructed from Eurobond returns. Whenever spanning is rejected, it is only for the full sample period and for the value-weighted industry portfolio indexes. The latter is in line with the results from the decomposition analysis where we find that country effects contribute more to the variation in Eurobond returns than industry effects. The results from the efficiency tests show that country and industry portfolios cannot be statistically distinguished from each other in terms of their maximum Sharpe ratios. This is true for both types of country and industry portfolios, and for all periods. We do observe that country portfolios always have a higher maximum Sharpe ratio than industry portfolios albeit not significantly different. Our results for European corporate bonds are noteworthy; particularly if one considers that mean–variance tests performed on European stock returns have yielded mixed results at best. We are able to draw a general conclusion that allocations based on a country or an industry strategy in the Post-EMU era are both efficient in a mean–variance setting. In the Pre-EMU period a direct country- or industry-based allocation of European corporate bond portfolios or an industry-based portfolio using decomposed returns would have yielded inefficient portfolios. It does seem true, however, that over the long-term country-based corporate bond portfolios are rewarded with better risk-adjusted returns than industry-based portfolios. For Eurobonds country effects in portfolios matter, and contrary to common perception under fund managers, even more under EMU than before.