[In his Jackson Hole speech, Monetary Policy since the Onset of the Crisis, Chairman Bernanke cited a Macro Advisers commentary from earlier this year written by Larry Meyer and Antulio Bomfim: “Not Your Father’s Yield Curve: Modeling the Impact of QE on Treasury Yields.” That paper is excerpted below.]
Traditional yield curve models are not well equipped to capture the effect of unconventional monetary policy on the term structure of interest rates. Who would have thought five years ago that the Fed’s balance sheet policies would become an important yield curve factor? Unconventional policies call for novel approaches to modeling the curve. Our proprietary yield curve model explicitly addresses the potential effects of Fed asset purchases.
Main Assumptions and Estimation Approach
The foundation of our model is what financial economists call the expectations hypothesis of the term structure of interest rates. That hypothesis says that the yield on a Treasury note that matures in N years can be decomposed into two main components: the market’s expected weighted average value of the overnight interest rate during the next N years and a term premium.
Of course, a challenge with our modeling approach is that neither the term premium nor the market’s expectation of the average value of the funds rate is directly observable. Both must be estimated or inferred from observed data. We estimate the rate expectations component by combining quotes from money market futures rates with forecasts derived from a small-scale macro model. In addition, we assume that the term premium component is driven primarily by four factors: Fed credibility, the relative stance of monetary policy, bouts of acute financial market stress, and expectations of the Fed’s balance sheet policies.
Using the ten-year Treasury yield to illustrate our estimation approach, we start by regressing the difference between the ten-year yield and its rate expectations component on our proxies for Fed credibility, the relative policy stance, market stress, and Fed balance sheet policy.
The model fits the data well, both across time and maturities. Figure 6 shows our estimates of the 2012Q1 values of actual and equilibrium term premiums for different maturities. Most of the gap between the actual and equilibrium premiums is attributable to the Fed’s large-scale asset purchase programs. The effect of these programs was felt most strongly around the five-to-seven-year maturity range, consistent with the overall allocation of net Fed purchases under QE1, QE2, and Operation Twist.
The Bottom Line
Unlike yield curves from before the 2008/2009 crisis, today’s curve is importantly affected by market participants’ views of the Fed’s future balance sheet policy. As a result, conventional yield curve models that worked well in the past may not work as well today. Our proprietary model includes expected Fed balance sheet policy as a term premium factor. In addition to fitting the data well, our model suggests that term premiums have been substantially depressed by Fed asset purchases.