In a previous issue of EF we attempted to cobble together an index of precious metals equity from the Morningstar and VanEck data, and came up with an annualized return of 12.81% from 1/69 to 9/96.

While we thought this was a useful estimate, we were concerned that it seemed "too high," and were worried that it encompassed only 27.75 years.

After a great deal searching, it turns out that the obscure S&P Gold Mining Index actually goes back to 1942. Courtesy of Brian Taylor at Global Financial Data we now have this data. The price we pay for this much more reliable index is that it does not include dividends. For the 55 year 1942-96 period the principal only return was 6.45%. If one assumes an average dividend of 3%, then the long term return over the period was 9.45%. This compares with 16.53% for small stocks, 13.96% for the S&P 500, 5.20% for the 20 year treasury, and 4.47% for t-bills over the same period.

The correlations with the other assets in the Ibbotson data base are as follows: US Small Stocks 0.23, S&P 0.24, Long Corporates 0.02, 20 year treasuries 0.00, 5 year treasuries 0.07, and T-Bills 0.09. I've plotted the annual returns and wealth index for the period below:







As you can see, this asset is not for the faint of heart -- in 1969 the index lost 75% of its value. Nonetheless, optimizations of the 1942-96 historical data for the 6 Ibbotson assets and the S&P Gold Index (assuming the 9.45% dividend inclusive return) yields an optimal precious metals equity allocation of about 10% over the mid range of risk (10%-15% of SD).

In 1942 the price of gold buillion itself was fixed at $35, and at year end 1996 was $340, for a 55 year return of 4.22%. It should not surprise that the return of gold stocks was higher. Over the long haul you do much better investing in United Fruit than in a load of banannas.

copyright (c) 1997, William J. Bernstein