Estimating the Term Premium: Sample Periods Matter

Zehao Li

♦ Estimated risk-neutral rates from canonical affine term structure models are highly sensitive to sample periods. For example, the 5-5 forward risk-neutral rate for September 1981 can differ by 4.6 percentage points (98%) depending on whether the sample starts in 1961 or 1981. Additionally, the estimated response of this rate to high-frequency monetary policy shocks varies severalfold, even within the same sample for monetary policy transmission regressions. We propose that a shifting endpoint model can mitigate these issues and provide new estimates of the effects of monetary policy shocks on long-term risk-neutral rates.

Read it here.