Abstract
To explain the interactive and causal relationships between gold and silver price movements, time series procedures are used to identify reduced-form parameters corresponding to a general arbitrage model. Enveloped in a framework of expectations about capital gains, current price changes are conjectured to be functions of lagged and present values of the substitute precious metal prices and the interest rate, plus lagged values of its own price. To determine the appropriate number of lags, to identify the appropriate interaction, and to examine the exogeneity between the gold and silver prices and the rate of interest, we use Schwarz's Bayesian information criterion and Akaike's final prediction error criterion. The resultant estimated model is simultaneous and provides information on participants' average lead times in forming these expectations.
KEY WORDS: