مقاله انگلیسی رایگان در مورد بحران مالی و بایاس برآورد در بازارهای اوراق قرضه ( الزویر )

مقاله سال 2016

The past couple decades saw an admittedly large number of crises including the Russian Default Crisis of autumn 1998, Y2K in 2000, the Dot Com Bubble spanning the period 1997-2000, the recent Global Financial Crisis of 2007-2010 and the Eurozone sovereign debt crisis that began in October of 2009 (Bussiere and Fratzscher 2006; Juneja and Pukthuanthong 2015). It is not surprising that much recent work documenting their consequences for financial markets has emerged in the extant literature (e.g., Bowe and Doumta 2001; Forbes and Rigobon 2002; Corsetti, Pericoli, and Sbracia 2005; Inyeob Ji and In 2010; Ivashina and Scharfstein 2010; Kenourigos, Samitas, and Paltalidis 2011; Wang 2014). Recently, authors have also begun to explore consequences for international bond markets (e.g., Dungey, Fry, Gonzܽ́lez-Hermosillo, and Martin 2006; Beetsma, Guiliodori, de Jong, and Widijanto 2013; Philippas and Siriopoulos 2013). Concurrent with this literature, scholars have examined the impact of estimation bias on the empirical performance of models for the dynamics of bond markets (e.g., Dempster and Tang 2011; Bauer, Rudebusch, and Wu 2012; Bauer, Rudebusch, and Wu 2014; Wright 2014; Juneja 2016). Based upon these studies, there is reason to believe that the impact of estimation bias on the dynamics of international bond markets is exacerbated during financial crises. Such bias likely creates patterns of elevated volatility in measures of empirical performance associated with international bond markets and those crisis periods.

The primary focus of this research is to study the impact of estimation bias on the dynamics of international bond markets during recent financial crises. We focus on the US and its main trading partners, Canada, China, the eurozone, Japan, and Mexico; due to the fact that countries that are intimately involved with each other from the perspective of international trade would presumably be characterized by similarities in bond market dynamics.1 To provide motivation for this selection, we run principal components analysis on zero coupon yields corresponding to each country in our data sample over the period under investigation in the current study; March 5, 2004 through December 12, 2014. Carrying out principal components analysis enables us to extract the main factors driving the variation in interest rates for each country. For each country, only one factor explains the majority of variation in interest rates. 97.73% of the variation in US interest rates can be explained by the first factor. 97.42% of the variation in Mexican interest rates can be explained by the first factor. 97.95% of the variation in interest rates in the eurozone can be explained by the first factor. 94.68% of the variation in Japanese interest rates can be explained by the first factor. 93.03% of the variation in Chinese interest rates can be explained by the first factor, and finally 93.06% of the variation in Canadian interest rates can be explained by the first factor. In Figure 1, we present a time-series plot of the first factor for the US and its main trading partners. Patterns in co-movements across the first principal component of the US and its main trading partners appear to be quite strongly related as the first factor generally has the same shape. In the case of Mexico, its first principal component moves inversely with that of the US. In fact, the correlation coefficient between the first principal component of the US and Mexico is -0.861, while the correlation coefficients between the first principal component and the remaining countries are 0.855 for the eurozone, 0.841 for Japan, -0.064 for China, and 0.897 for Canada. With the exception of China, the time series dynamics of the main factor driving variation in interest rates between the US and its main trading partners are very strongly interrelated. Indeed, the main factor influencing China’s interest rates also exhibits similar patterns in co-movements. Byrne, Fazio, and Fiess (2012) study the co-movement in long-term interest rates for eight industrialized countries, including the US, Canada, and Japan, over the period January 1988 through July 2006. They find that yields on government debt at the 10-year maturity for the eight nations included in the sample display a remarkable degree of co-movement that increases toward the end of their sample period, which coincides with the beginning of our sample period. Taken together, Figure 1 and this prior finding suggest that co-movements across factors were especially pronounced during the Global Financial Crisis of 2007-2010 and the ongoing eurozone sovereign debt crisis that began in October 2009. Additionally, these remarkable patterns in co-movements support the notion that the US and its main trading partners are quite similar from the standpoint of bond market dynamics and this provides motivation for the inclusion of these countries in the study.

We carry out our assessment of model accuracy by constructing estimates of three measures; model forecast error, long maturity term premia, and long maturity risk premia and follow the implementation of Juneja (2016) in our empirical design. Therefore, our analysis relies on the data which we believe represents an advantage relative to prior approaches (e.g., Yang and Wang, 2010, Juneja, 2014, Juneja, 2015) which rely on observation or theory (e.g., Dempster and Tang 2011; Bauer, Rudebusch, and Wu 2012). Additionally, we focus exclusively on the class of affine term structure models because it received a large amount of attention in the literature (e.g., Dai and Singleton, 2000, Dejong, 2000, Collin-Dufresne, Goldstein, and Jones, 2008, Christensen, Diebold, and Rudebusch, 2011, Duffee and Stanton, 2012, Hamilton and Wu, 2014).2