A gold pricing model (1983)
Sherman (1980) provides the first model of the gold price in the literature using a number of variables but with little theoretical backup for his choice of variables from any previous studies as well as measuring the elasticity’s of commercial demand for gold. These estimates involve OLS regressions without accounting for unit roots in the data. In the elasticity’s equation he regresses logs of the real gold price and world GDP on the log of the quantity of commercial demand using 11 years of observations. Demand for gold for commercial purposes is found to be above unit elasticity with respect to the real price of gold and world GDP as a proxy for world income. While both findings seem sensible the lack of data and the questionable estimation process means that further work on these questions is required, as noted by the author.
The model of the gold price uses variables at a monthly frequency from 1972 which have become common in other more recent studies and it seems this work has had an impact on later researchers thinking. Again an OLS regression is carried out in levels without addressing unit root issues. The variables listed are not always attached to sources.
He uses a tension index to reflect precautionary demand for gold, or its safe haven status as it is now described. As in 1983 gold did not have a yield through the London leasing market the real Eurodollar rate is used to show the opportunity cost of holding gold. The US dollar trade weighted index serves to reflect the fact that old is measured in dollars and to capture a number of unspecified macroeconomic variables not included elsewhere in the model. Excess liquidity serves to reflect a proxy for anticipated future inflation and is measured as an author constructed measure of global M1 indexed to 100 at the beginning of 1972 divided by a GNP/GDP index for the big seven industrial countries. Finally a variable to reflect unanticipated inflation is given as e difference between inflation at t and average inflation over the previous 18 months.
The authors note issues of multicolinearity between a number of these variables. The Eurodollar interest rate, US dollar trade weighted exchange rate, world GDP, excess liquidity and unanticipated inflation are all found to be se significant explanatory variable for the gold price in the expected directions. However the lack of theoretical clarity in the choice of variables included and their construction coupled with the issue of unit roots means that this attempt at modelling the gold price left significant room for improvement in later studies.
Reference: Sherman, Eugene J. “A gold pricing model.” The Journal of Portfolio Management 9.3 (1983): 68-70.