An enigma inside the conundrum

It is sometimes argued that foreign exchange intervention in support of the US dollar by dollar-pegging Asian central banks, which was especially heavy from 2002 onwards, contributed to the ongoing financial crisis which began in the US last summer. The idea is that the investment of the proceeds of this intervention in dollar bonds represented a "saving glut" that depressed bond yields (by up to 60bps in the period up to May 2005, according to Warnock and Warnock's Federal Reserve International Finance Discussion Paper 840). The fact that this fall in yields occurred despite tightening US monetary policy from 2004 was described at the time by Federal Reserve Chairman Greenspan as a "conundrum". In an attempt to maintain their returns in the face of lower yields, the argument goes, investors became more willing to buy bonds which offered a relatively high yield to compensate for higher risk, including risks of default, illiquidity and operational problems such as legal ambiguity. This increased demand for "spread product" (so-called because the relatively high yield of these bonds is characterised as a yield spread over benchmark government bonds) "crowded in" additional supply, including bonds involving previously almost unsaleable risk such as marginally creditworthy borrowers and complex contracts. It was the eventual realisation of these risks, notably in collateralised debt obligations (CDOs) constructed from mortgages owed by sub-prime borrowers, that precipitated the crisis.

The problem that RebelEconomist has with this explanation is that, if the growth in spread product had been driven by central banks, which tend to buy relatively safe government or government-sponsored agency bonds, the yields on this safe debt should have been depressed the most. It would be expected that some relative decrease in yield would be required to induce other buyers of safe bonds, such as pension funds and insurance companies, to switch out of safe debt into riskier and less liquid substitutes. In other words, spread product spreads ought to have widened. On the contrary, until the crisis hit, spreads narrowed, as exemplified by the five year (a maturity representative of central bank foreign exchange reserves portfolios) swap spread (ie representing commercial bank debt) shown in Figure 1.




















Of course, it may well be the case that, rather than simply representing the expected value of credit, liquidity and other losses suffered by investors in spread product, yield spreads depend on investor preferences for risk, which they trade off against expected return, as in the classic mean-variance model of portfolio selection. If so, however, it seems reasonable to expect the spread to shrink in line with yield levels. But this was not the case either; even expressed as a proportion of the yield on the corresponding treasury benchmark, spreads narrowed, also as shown in Figure 1. Moreover, if buying of riskier bonds had been motivated by inadequate yields on safe bonds, investors might have expressed concern about having to take these risks, whereas in fact surveys of investor sentiment, such as the Gallup/UBS investor optimism index, indicated a gradual recovery of confidence following the dotcom bust which culminated in the bankruptcy of Worldcom in 2002.

My conclusion is that the idea that central bank buying of the safest debt displaced existing investors reluctantly into spread product is implausible. Perhaps there was some less direct link between central bank dollar purchases and the growth of spread product, such as a surge in US economic optimism that encouraged Americans to both consume more, including imports from dollar-pegging countries, and regard investment risks like default as less likely to be realised. Alternatively, a growing perception that a "Greenspan Put" existed (ie that the Fed would ease in response to financial crisis) may have led investors to regard spread product as less risky than previously and no longer requiring such a wide yield spread. Whatever, it seems that the explanation for the narrowing of spreads associated with the conundrum is itself an enigma worthy of more detailed investigation.