A long-commodity-cycle model of the world economy over a century and a half: Making bricks with little straw

In this paper we consider the world business cycle and specifically relate it to two main shocks: real commodity supply and prices, and money supply policy. Our model suggests that because there is a long gestation of commodity supply capacity, commodity prices are capable of very large swings; these swings in turn cause swings in costs for non-commodity industries which act as a source of real business cycle shocks to output. Investment in the non-commodity sector is subject to quite short lags and so these shocks cause investment booms and busts in this sector which in turn generate booms and busts in general output. Our model therefore refers back to the ideas of Kondratieff and Schumpeter concerning the importance of '’long waves' ’and commodities in the business cycle. We estimated a DSGE model with a long time to build in commodities to try and explain the long wave commodity cycle in the world economy. Using data from 1860-2008 we found a model that easily passes the indirect inference Wald test. Inflation and interest rate data were obtained from measuringworth.com. Other data sources include Historical Statistics of the World Economy: 1-2008 AD (GDP data), Monetary Statistics of the United States, Abstract of British Historical Statistics, and the World Bank. The model exhibits long commodity cycles, with a half-cycle being approximately 30 years. This seems to fit with the data, for example, the commodity peak of 1980 then three decades of falling/very low/sharply rising prices which peaked at 2007. The three main shocks to the model are found to be money supply (affecting output), materials productivity (affecting commodity production) and a Phillips Curve shock (affecting inflation), which echoes the findings of Friedman and Schwarz (1963). As a robustness check we calculated the welfare loss for various different monetary rules including a Taylor Rule, a nominal GDP targeting rule and a price level targeting rule. We found that neither could improve on the original money supply growth rule. We found that Friedman's k-percent rule is a slight improvement.