Erb and Harvey (2006), however, focus on the composition of the main commodity
market indexes used in practice, including the GSCI, and investigate whether these
indexes provide a representation of the aggregate commodity market. They question
whether commodity markets can actually be considered as a single asset class since
differences in the behavior of prices between individual commodities seem significant. In
their empirical work, they demonstrate that historically commodity futures returns have
largely been uncorrelated with one other and they caution against extrapolating historical
returns on an index like the GSCI into the future. In fact, they argue that it should be
questioned whether the commodity markets can be represented by a single index, which
appears contrary to the contentions raised by Gorton and Rouwenhorst (2006).
3. Method and Data
(a) Method
Following the results of prior work, (e.g. Strongin and Petsch, 1995; Vrugt et al.,2004;
Rouwenhurst and Gorton, 2006; Fleming et al. 2006) we posit that precious metals prices
are related to a key set of macroeconomic variables that represent the broader monetary
environment (such as inflation and monetary aggregates), the business cycle (such as
industrial production) and financial markets conditions (such as the US dollar exchange
rate, stock index returns and consumer confidence indexes). These are discussed in
greater detail in the next section. Hence, the expected returns can be expressed as:
Et(rtM|It-1)= f(Et|Ct(Xt)) (1)
where r is the return on a precious metal (M = gold, palladium, platinum or silver) at time
= t, conditional on the information (I) available at the previous time interval time (t-1).