Are asset prices predictable? : evidence from the UK futures market
The purpose of this thesis is to provide evidence on the predictability of asset prices: does it exist and what are its characteristics? Therefore processes underlying asset price behaviour are the central concern of this thesis. Utilising UK futures market data, initial time-series tests consider if, ex post, prices depart from random behaviour and if linkages exist between prices set in different markets. Results show that departures from randomness exist, namely, persistence and mean reversion, and suggest that prices in some markets may be used to forecast prices set in other markets. The thesis then investigates the mean reversion characteristics of the data by conducting 'model-based' tests on the existence and sources of mean reversion. Subject to the maintained hypothesis of the cost-of-carry model, tests reveal expected transitory components in commodity and metals prices across maturities. For financial assets, expected mean reversion is found only at the near to maturity horizons. Implied cash flow yields and interest rate movements have a role in driving the transitory process but this varies across assets and maturities. Using multivariate estimation procedures and focussing on asset bases, the thesis then explores whether predictability may be due to a common component driven by an unobservable source of risk. Movement in the price of systematic risk is proxied by ex ante variables that have been shown to have predictive power for returns from bond and stock markets. For financial assets, the evidence cannot reject common movement at short horizons but rejects common movement at longer horizons. Further tests reveal that the observed short run commonalties are dominated by systematic components associated with expected future spot rates. In contrast, tests reveal that agents' in metal and energy markets appear to price the systematic risk associated with their futures position as well that associated with spot price forecasts.