choosing the explanatory variables for the present study largely follows these papers.
Chen (1991) argues that dividend yield and the default spread are associated with
business cycle conditions and hence, we include annualized dividend yields on both S&P
500 and the World excluding the US stock indexes in our analysis. Also, we use the
annualized yield spread between corporate bonds, given as the yield spread between
Moody’s rated BAA- and AAA-bonds, as our default spread. Additionally, we include
change in industrial production, calculated as change in year-over-year figures, to capture
the link between the business cycle and annual production growth.
With regard to monetary environment variables, Gorton and Rouwenhortst (2006) argue
commodities are inflation hedges and hence, we include the rate of inflation (calculated
as year-over-year) in the data set. Furthermore, we include the monetary aggregate M2,
which is likely to be important to describe changes in monetary conditions in the
economy and for financial market sentiment variables, we rely on the stock market
returns, using both the total return on the S&P 500 and World ex US indexes. In addition,
we include the consumer confidence index for the US, since the US is the most important
market for the consumption of commodities, and the trade weighted US dollar index,
similarly since precious metals are denominated in US dollars. All data are obtained from
Datastream.
(Insert Table 1 about here)
4. Empirical Findings
(a) Preliminary Analysis