In this paper we propose a panel data approach to modeling the risk premium in the term structure of interest rates. Specifically, we develop a fixed maturity/random time effects model, which implies a time-invariant one-factor model. Our approach allows us to disentangle risk premia and unexpected excess returns, which is not possible in the standard time series approach. In addition, small sample bias is alleviated and statistical efficiency improved. Our results allow for interesting inferences about maturity-specific effects in the term structure. First, the expectations hypothesis is soundly rejected for our full data panel of us treasury securities. Second, a considerable degree of mean reversion is present in the risk premia. Third, our findings shed new light on the magnitude of the slope coefficient in regressions of the yield onto the forward curve.
|Number of pages||25|
|Journal||Journal of International Financial Markets, Institutions & Money|
|Publication status||Published - 1 Jan 2003|