The Price of Gold – A simple Model (1989)
This 1989 paper builds an econometric model of the factors that the authors believe drive gold prices using annual data over 16 years. Their model has an R2 of 92%, explaining 92% of the variation in the annual gold price. The model can be criticised on a number of points.
Firstly the price of gold is a non-stationary time-series (See Diba and Grossman, 1984). This means that the assumptions of an OLS regression are violated and the results are questionable, to be interpreted with caution. Also the inclusion of a trade weighted dollar index has been criticised when used to explain movements in the gold price (See O’Connor and Lucey, 2012).
The authors aim to build a predictive model but as thedependant variable is not lagged it is not forward looking. Inflation data for example will only be available after the gold price is realised.
Method: Log-Log OLS Regressions
Data: Annual 1974 – 1988. Price of gold and gold supply from the Consolidated Gold Fields Annual Report. Dollar exchange rate index from the Federal Reserve. GNP Deflator from the US department of commerce.
Full Citation: Kaufmann, T. and R. Winters (1989). “The Price of Gold: A Simple Model.” Resources Policy 19: 309-318.
Abstract: Using regression techniques and seeking a simple predictive model the authors derived a formula for the annual price of gold based on changes in the rate of inflation in the USA, an index of the US dollar exchange rate and the annual world production of gold. Statistically the model shows a high correlation between the formula price and the market price over the past 16 years although many variables often considered important to the price of gold are ignored.