As the conviction grows that the Federal Reserve will adopt a second round of quantitative easing (dubbed by some as “QE2″), I thought it might be helpful to survey some of the different estimates of what effect this might have on long-term interest rates.
Although we are used to thinking of the Federal Reserve as playing a key role in determining interest rates, it is far from clear that the Fed matters that much for interest rates at the moment. The Fed’s traditional influence comes from changing the supply of reserves injected into the banking system, which in normal times would quickly change the interest rate at which banks lend those reserves to each other overnight. But with over a trillion dollars in excess reserves and the overnight rate practically at zero, the Fed’s primary policy tool is completely irrelevant at the moment.
There might still be some ability to affect longer-term interest rates from much more massive operations. The theory is that by taking some of the supply of longer-term bonds off the market, this might raise the price and thus lower the yield on those bonds. There are a number of recent studies that try to measure the potential size of this effect. I’ve described my own research with Cynthia Wu on this topic to Econbrowser readers, and today would like to take a look at how our estimates compare with those obtained by other researchers.
It’s not trivial to make these comparisons, because different researchers have posed the question as to what the Fed might do in different ways, and used different measures to summarize what the effects would be. One benefit of our approach is that it provides a complete description of what would happen to the entire yield curve, both in normal times and in the current environment in which the overnight rate is at the zero lower bound. That allows us to calculate predicted values for the various magnitudes that other researchers have studied.
My paper with Cynthia examined what would happen if the Fed were to have bought $400 billion in long-term Treasuries, using December 2006 as our base for the pre-crisis evaluation. In the pre-crisis environment, it is quite important in our framework to assume that the Fed sterilized this with $400 billion in offsetting sales of short-term Treasuries. By contrast, when at the zero lower bound as now, sterilization is irrelevant, and we have alternative estimates of the effects if the Fed simply purchased $400 billion in long-term securities outright with newly created reserves in the current zero-lower-bound (ZLB) environment.
One of the most influential studies of the effects of such operations is by Joseph Gagnon and colleagues, who report estimates from a variety of methods. The estimate we highlight in the table below comes from a regression of the 10-year Treasury yield on the supply of Treasury debt of more than one-year duration. Gagnon and colleagues measured the latter as a fraction of GDP. A number of other variables thought to influence yields were also included in the regression, which was estimated over 1986:M12 to 2008:M6. Their estimated coefficient on the relative debt supply is 0.069. In 2006:Q4, $400 billion would correspond to about 2.9% of GDP, for a predicted drop in the 10-year yield of (2.9)(0.069) = 20 basis points. By contrast, Cynthia and I would predict about a 14-basis-point decline from such an operation if it had been implemented in December 2006 and almost the same effect if implemented today.
The Gagnon study reported a range of estimates from other methods, some of which would correspond to a predicted effect as low as 12 basis points.
Greenwood and Vayanos arrived at estimates in a rather different way. They regressed the yield spread between assets of different maturities on the share of outstanding Treasury debt that was of more than 10 years in duration. They found that a one-percentage-point increase in the share resulted in a 4-basis-point increase in the 5-year over 1-year spread over the period 1952-2006. In the sample we studied (1990-2007), a maturity swap of the size considered above would have lowered the share of debt with maturity greater than 10 years by 9.8 percentage points, which gives an effect implied by the Greenwood-Vayanos estimates of (9.8)(4) = 39 basis points. By contrast, the estimate based on our empirical analysis of the size of the effect would be 17 basis points if implemented in 2006, but only 9 basis points if implemented in the current environment. The reason for the difference is that, in our framework, part of the drop in the spread if the policy had been implemented in 2006 would have come from an increase in short-term yields, something that would not happen if the same purchase were implemented today.
A third study has just been released by Fed researchers Stefania D’Amico and Thomas King. They look at the change in yields of different maturities during the Fed’s purchase of $300 billion in long-term securities between March and October of 2009. They conclude that these purchases lowered the yield on 10-year Treasuries by about 50 basis points, which would translate into an effect of (4/3)(50) = 67 basis points for a $400 B purchase analyzed in the table above. Our estimate of 13 basis points is quite a bit below theirs. However, the 10-year yield was where these purchases were concentrated and where D’Amico and King found the biggest effects, and large standard errors are associated with any of these estimates.
Also last week, Deutsche Bank in the September 29 issue of its Global Economic Perspectives reported “guesstimated effects” of $1 trillion in new long-term security asset purchases by the Fed. These were based on the Gagnon study mentioned above along with separate estimates put together by Macroeconomic Advisers and Deutsche Bank staff. Deutsche Bank’s expectation is that $1 trillion in long-term purchases in the current setting might produce a 50-basis-point decline in long-term yields, which we’ve translated as a 20-basis-point decline for the $400 billion purchase analyzed in the table above.
Given the great differences in the methods by which these estimates were arrived, there is a surprising degree of consensus, at least as far as the rough magnitudes are concerned. Our estimates are a bit below those of other researchers, but everyone has found that there will be some effect if the purchases are of a sufficiently large magnitude. As mentioned above, there is considerable uncertainty associated with any of these estimates.
These also appear to be the sort of numbers that the FOMC may be thinking in terms of as well. In a speech given Friday, Federal Reserve Bank of New York President William Dudley conveyed the following:
some simple calculations based on recent experience suggest that $500 billion of purchases would provide about as much stimulus as a reduction in the federal funds rate of between half a point and three quarters of a point. But this estimate is sensitive to how long market participants expected the Fed to hold on to these assets.
In recent years, a 100-basis-point move in the fed funds rate has translated into about a 40-basis-point move in the 10-year yield (e.g., Table 2 of my 2008 study). Hence if we use the lower end of Dudley’s range, we might come up with a number of (400/500)(1/2)(40) = 16 basis points as another ballpark estimate of the effect on the 10-year rate of another $400 B in long-term bond purchases.
But even if we agree that the Fed could depress long-term yields with these kinds of measures, it is a separate question as to whether it should. I discussed this issue a few weeks ago. I remain of the opinion that while the Fed is understandably reluctant to embrace QE2, it may have little other choice.