It's no secret that credit spreads, or the difference between U.S. Treasury and corporate debt yields, have been trending ever lower.
Investors around the world have been chasing the juicy returns on offer from buying corporate debt, driving yields on that debt to historic lows and diminishing the extra "spread" one might earn when compared to buying ultra-safe U.S. government debt. Years of low interest rates and easy money have kept the number of corporate defaults ultra-low, giving even the bonds issued by junk-rated companies a new "safe haven" status and making them an added attraction for yield-starved investors.
A new Federal Reserve working paper by David Lopez-Salido, Jeremy Stein, and Egon Zakrajsek attempts to quantify the effects of credit-market sentiment on business cycles, using data from 1929 to 2013. While the paper is heavy on statistical analysis and formulas, it does find a link between compressed credit spreads and a slowdown in the economy:
This paper emphasizes the role of credit-market sentiment as an important driver of the business cycle. ... More specifically, we establish two basic findings about the importance of time-variation in the expected returns to credit-market investors. First, using almost a century of U.S. data, we show that when our sentiment proxies indicate that credit risk is aggressively priced, this tends to be followed by a subsequent widening of credit spreads, and the timing of this widening is, in turn, closely tied to the onset of a contraction in economic activity.
In other words, animal spirits in the credit market are associated with an eventual correction and then a subsequent decline in economic activity.
The findings come at a potentially sensitive time for the Federal Reserve. The U.S. central bank is trying to figure out how to raise interest rates without roiling markets, but it's also been accused of prolonging its easy monetary policy and inflating a great bubble in bonds (and maybe other assets) in the process. While the paper's authors skirt making direct recommendations, they do suggest that their findings have implications for future monetary policy.
The question our research raises is whether there is a future price to be paid for today’s highly accommodative policy. Specifically, if a policy-induced compression of credit-risk premiums tends to reverse itself in the same way that unconditional movements in credit-risk premiums do, then our results might lead one to believe that easy policy today would be associated with an increase in the expected unemployment rate somewhere between two and four years down the road. If so, the central bank would face a nontrivial intertemporal tradeoff, even in the absence of any tension between its unemployment and inflation goals: an aggressively accommodative policy would move it closer to its unemployment target today, but might, at the same time, risk pushing unemployment rates in the future further away from the target value.
How should the central bank seek to handle this tradeoff? Clearly, it depends on how far unemployment is from target today. ... The marginal benefit of reducing unemployment at any point in time is linearly increasing in the distance from target. So if the unemployment rate today is very high, this marginal benefit is likely to loom large in relation to any future marginal costs, which are evaluated around a presumably lower expected level of unemployment. In contrast, if unemployment today is only slightly above target, the marginal benefit of accommodation could be less than the expected marginal cost of increased unemployment in the future. Thus, even without taking the threat of inflation into consideration, there may be a reason for the central bank to begin gradually removing accommodation as unemployment approaches its target level, especially if credit-market sentiment appears to be elevated.
To be clear, this discussion is intended to be speculative. And even if one agrees with the qualitative arguments, our results are not sufficient to allow for the monetary-policy tradeoff we have outlined to be quantified in such a way as to make it operational. To get anywhere close to this point will require a good deal of further work, both conceptual and empirical. Nevertheless, we do want to highlight what we see as a potentially useful direction for future research.