Economic Letters, 2015, 126, 96-100. Libo Yin*, Liyan Han

The feature of co-movements in commodity prices prompts a new search for the fundamentals. From a theoretical point of view, Alquist and Coibion (2013) show a factor structure for commodity prices in which the common factor captures the combined contribution of all aggregate shocks that affect commodity markets.

However, research from an empirical standpoint generates mixed views. Byrne et al. (2013) document a significant degree of co-movements. West and Wong (2014) find that commodity prices consistently display a tendency to revert towards one factor. However, evidence for co-movements has recently been challenged by Daskalaki et al. (2014) who find that none of the factors could explain the co-movements of commodity returns.

This paper contributes to the empirical evidence on the significance and the structure of the common dynamic properties of commodity price fluctuations. By employing a Bayesian dynamic latent factor model, we relate international commodity returns to one global, six sectoral, and 24 commodity-specific factors. This decomposition measures the extent to which global, sectoral, and commodity-specific components explain the variation in commodity prices.

Perhaps the most related approach to our own are Gospodinov and Ng (2013) and Moench et al. (2013). The first one decomposes commodity convenience yields into factors and uses these estimated factors for forecasting inflation. However, it does not put any limitations on the structure of factors. The second one puts a block structure on the panel and estimates global and sectoral macroeconomic factors.

We find that there are two sources of co-movements present in the fluctuations in almost all of the commodities, which together explain almost half of commodity returns variations. A subsample analysis further reveals that the global factor typically increases significantly in importance for all commodities, likely as the results of increased index investment of commodities.