Information in the Term Structure – The Indian Evidence (I): Modeling the Term Structure and Information at the Short End for Future Inflation

This study is part of an on-going work on assessing the information content of the term structure in India for future inflation, future short rates and real interest rates. In this part, first the Indian term structure is modeled using three alternative specifications and changes in slope of the term structure at the short-end assessed for forecastability of inflation. Performance of two atheoretical (Nelson and Siegel, 1987 and Svensson, 1994) models is compared against empirical implications of a general equilibrium (Cox, Ingersoll and Ross, 1985) model. While Svensson is seen to offer no improvement over Nelson-Siegel, Cox-Ingersoll-Ross comes out as marginally superior to both on the criteria of mean absolute pricing and yield errors (both in-sample and out-of-sample), behaviour of the short and the long rates, stability of the parameters and behaviour of forward rates for maturities 1-8 years. This is encouraging because models like Nelson-Siegel and Svensson are designed to fit the observed yield curves, while Cox-Ingersoll-Ross is a theoretical model derived from intertemporal description of a competitive economy. On the information content of the term structure, in the sample under study, change in the slope of the term structure seems to have no information for inflation changes over the horizon 1 month to 2 years. Results could be sample and/or sampling-frequency specific. Results for the long end of the term structure (from a bigger sample) follows.