Bahram Pesaran

Modelling Volatilities and Conditional Correlations in Futures Markets with a Multivariate T Distribution

This paper considers a multivariate t version of the Gaussian dynamic conditional correlation (DCC) model proposed by Engle (2002), and suggests the use of devolatized returns computed as returns standardized by realized volatilities rather than by GARCH type volatility estimates. The t-DCC estimation procedure is applied to a portfolio of daily returns on currency futures, government bonds and equity index futures. The results strongly reject the normal-DCC model in favour of a t-DCC specification. The t-DCC model also passes a number of VaR diagnostic tests over an evaluation sample. The estimation results suggest a general trend towards a lower level of return volatility, accompanied by a rising trend in conditional cross correlations in most markets; possibly reflecting the advent of euro in 1999 and increased interdependence of financial markets.

An Assessment of the Relative Importance of Real Interest Rates, Inflation, and Term Premiums in Determining the Prices of Real and Nominal U.K. Bonds

We use a vector autoregression (VAR) to decompose unanticipated bond returns into news about fundamentals (expected real interest and inflation rates) and expected risk premiums. This decomposition is applied to U.K. short- and long-maturity nominal bonds, and to U.K. index-linked bonds. We also examine the sources of relative conventional and real bond returns. The results suggest that for both bond types, real-rate news plays an insignificant role, and that even for real bonds inflation news is important. Both bonds are strongly influenced by news about future risk premiums, but these appear to reflect a common factor that has little influence on their relative returns. News about inflation dominates unanticipated relative returns, which appear to provide a reliable source of information about inflation expectations.

A Non-Nested Test of Level-Differenced versus Log-Differenced Stationary Models

This paper considersthe applicationof the simulated Cox test procedure developed in Pesaran and Pesaran (1993) to test linear versus log-linear models. The test procedure can also be applied to other generalized linear regression models such as level-difference stationary models versus the log-difference stationary models. In order to compare the small sample performanceof the proposed test with other tests extant in the literature, the paper also reports the resultsof a numberof Monte Carlo experiments using the experimental framework of Godfrey et al. (1988). The Monte Carlo results provide strong support for a simplified version of the simulatedCox test over the PE and the BM tests, but suggest that there is little to choose between the simulated Cox test and the DL test.

The Use of Spreads in Forecasting Medium Term U.K Interest Rates

This paper aims to extend recent work on the term structure of interest rates by establishing, in the context of the medium term UK interbank market, forecasting models which make use of market spreads as error correction terms. These models are then used within a trading scenario to test the short run efficiency of the market. The results indicate that this market is inefficient in the short run. Furthermore, the performance of the multi-step-ahead forecasts from the models suggest that this may be a fruitful avenue for further research into longer maturity rates.