سهم تقاضای اوراق قرضه آمریکا به مسئله پیچیده بازدهی اوراق قرضه آمریکا 2004-2007: تحقیقات تجربی

Although the federal funds rate started rising from mid-2004 US long term rates continued to fall. A likely contributory factor to this ‘conundrum’ was the contemporaneous increase in US bond demand. Using ARDL based models, which accommodate structural breaks, this paper estimates the impact of foreign and domestic demand on AAA rated US bond yields in the ‘conundrum’ period. This impact is shown to have been everywhere significantly negative. The fact that our model fully explains the ‘bond yield conundrum’ gives support to the hypothesis that the US CDO market was rapidly expanded before 2007 chiefly to absorb the overspill of global demand for safe assets. Moreover, our models demonstrate that there are strong linkages between the 10-year Treasury yield and the long term yields of AAA rated non-Treasury bonds.

From 2002 to mid-2007 when the US subprime crisis broke out US bond yields were at unusually low levels. Before mid-2004 these levels could be explained by the greater stability of ‘fundamentals’ and low short term interest rates (the ‘great moderation’), but the persistence of these low yields after that point in time was puzzling. Financial markets expected long term rates to rise in tandem with the rise in the federal funds rate as was the case in previous periods of monetary tightening. This did not happen. On the contrary, not only did long term rates not rise they actually continued to fall3 (see Fig. 1). As Alan Greenspan, the then Chairman of the Federal Reserve, stated before Congress in June 2005: “Among the biggest surprises of the past year has been the pronounced decline in long term interest rates on U.S. Treasury securities despite a 2-percentage-point increase in the federal funds rate. This is clearly without recent precedent. … Moreover, even after the recent backup in credit risk spreads, yields for … corporate bonds have declined even more than Treasuries over the same period.” (Greenspan, 2005, p.1).What caused this ‘bond yield conundrum’? Considering that its appearance coincided with a marked upswing in investor demand for US bonds (see Fig. 2) it is possible that a considerable part of the downward pressure on US bond yields stemmed from that demand (Bernanke et al., 2011). To verify this possibility, a number of empirical studies have focused specifically on the impact of foreign government demand for US Treasuries on long term Treasury yields. Foreign official investor demand began to increase after February 1994 when China devalued its currency, but the rate of increase in that demand accelerated even more sharply after 2003 as many emerging market economy governments sought to preserve part of their increasing commodity revenues and export surpluses in safe stores of value. While some studies found no evidence of a long term demand impact on Treasury yields (e.g. ECB, 2006 and Rudebusch et al., 2006), the majority of recent studies have found evidence of a negative impact, albeit that the estimated size of the impact varied from study to study (e.g. Idier et al., 2007, Bandholz et al., 2009, Craine and Martin, 2009, Warnock and Cacdac Warnock, 2009 and Bertaut et al., 2011).In this paper we assess the impact of investor demand on long term Treasury yields using an autoregressive distributed lag (ARDL) based econometric model. Since it has become well established that the increase in demand for US bonds stemmed not only from foreign official investors but also from private foreign (mainly European) and domestic investors (Bernanke et al., 2011) we consider the impact of all of these sources of demands on yields. Further, given that the ‘bond yield conundrum’ applied as much to the other major US bond markets as to the Treasury market, one would have expected an analysis of the impact of demand on long term yields in these other markets. As there has been no such analysis,4 this paper seeks to fill this lacuna by modeling the impact of foreign and domestic investor demand on US agency, and AAA rated corporate and municipal bond yields using ARDL-based models. As an additional novelty we also consider whether there are linkages between the 10-year Treasury bond yield and the long term yields of AAA rated non-Treasury bonds. In our view, an important by-product of this econometric assessment of the contribution of demand to the US ‘bond yield conundrum’ is that it may help to resolve the question as to why the US collateralized debt obligation (CDO) market was allowed to grow in a very short time to a size that was able to trigger widespread financial panic when this market suddenly collapsed in August 2007.5 The conventional answer to this question is one that places the major burden of responsibility on the US financial system itself. However, there is a minority view that, while the US banks and their associates cannot be absolved from blame in accelerating the rate of production of CDOs right up to mid-2007, the major driving force behind that acceleration was the pressure of demand for US safe assets spilling over from other major US debt security markets (e.g. Caballero and Krishnamurthy, 2009, Gros, 2009, Lysandrou, 2009 and Caballero, 2010). Clearly, this alternative view, and its ensuing policy implications, would command far more attention were it to be convincingly demonstrated that the rise in foreign and domestic demand for US bonds in the period leading up to the outbreak of the subprime crisis did indeed have a substantial significant negative impact on all highly rated traditional fixed income products. The layout of this paper is as follows. Section two gives details of the model specification, the data used and the chosen sample period. Section three presents and discusses the estimation results while section four briefly comments on their policy implications. Section five concludes.

Our models fully explain the US ‘bond yield conundrum’ of 2004–2007 as found not only in relation to US Treasuries but also in relation to all of the other traditional AAA rated US debt securities, something that has not been achieved in the previous literature. We attribute this result to the incorporation of a broader set of variables than is usual in our models, this being made possible by the adoption of the ARDL approach, and to the allowance for evident structural change at the end of the 1990s (the latter confirming findings of previous authors). We find that long term non-Treasury yields are significantly influenced by the 10-year Treasury yield and that the latter is important to explain the low pre-crisis long term yields of the former (the 10-year Treasury yield lowered non-Treasury yields by up to 64 bp in the ‘conundrum’ period). This is especially true for 10-year agency bonds that investors seemingly regarded as the closest substitute for 10-year Treasury bonds. We furthermore find that, if one accounts for the negative indirect effect of foreign official investor demand on non-Treasury yields via their depressing effects on the Treasury yield, investor demand seemingly is the most prominent factor in explaining the unusually low long term US bond yields in the ‘conundrum’ period. The direct negative impact of foreign official investor demand was up to 60 bp in the case of the 10-year Treasury yield and up to 43 bp in the case of the 10-year agency bond yield, whereas foreign and domestic private demand reduced the 10-year agency bond yield by up to 29 bp, and the AAA rated long term corporate and municipal bond yields by up to 19 bp. This finding has substantial policy implications in that it provides support for the hitherto underexplored hypothesis that next to regulatory and market failures an excess safe asset demand on the part of investors was a chief force that drove the expansion of the US CDO market well beyond what was prudent.