Do variations in the Baltic Dry Index (BDI), a weighted average of the Baltic Exchange shipping cost indexes for the four largest dry-vessel classes, serve as an early measure of global demand for raw materials and thereby predict asset class returns? In the January 2011 version of their paper entitled “The Baltic Dry Index as a Predictor of Global Stock Returns, Commodity Returns, and Global Economic Activity”, Gurdip Bakshi, George Panayotov and Georgios Skoulakis investigate the ability of BDI to predict stock market and commodity market returns. They focus on three-month changes in BDI as a predictor to smooth the high volatility of the monthly series. Using monthly BDI levels and returns for four MSCI regional stock indexes, 19 developed country stock indexes, 12 emerging country stock indexes, three spot commodity indexes and industrial production data for 20 countries mostly over the period May 1985 through September 2010 (305 months), they find that:
- The BDI three-month growth rate has a statistically and economically significant positive relationship with future monthly and quarterly returns for most global stock indexes, both in-sample and out-of-sample. However, predictive power essentially disappears at a six-month investment horizon.
- The predictive power of BDI persists after controlling for other widely used predictors such as past returns, the U.S. term premium and the U.S. default spread.
- Portfolio strategies designed to exploit the predictive power of BDI three-month growth rate generate higher certainty equivalent returns and Sharpe ratios than a passive benchmark, indicating economic value for asset allocation.
- The BDI three-month growth rate has a statistically significant positive relationship with future monthly changes in spot commodity indexes, with predictability derived essentially from raw industrial materials prices.
- The BDI three-month growth rate has a statistically significant positive relationship with future industrial production in most countries.
In summary, evidence from an array of tests indicates that investors may be able to exploit information in the Baltic Dry Index three-month growth rate about future returns for stock indexes and spot commodities.
Note that the statistical significance testing in this study generally assumes that variable distributions are well-behaved (normal). To the extent that BDI change and asset return distributions are wild, such testing loses meaning.
Reasons to be cautious about the findings of economic significance include:
- The use of indexes to measure strategy profitability ignores the frictions of forming and trading realistic portfolios of index components (stocks or commodities). Frictions include transaction costs and bid-ask spreads involved in buying and selling these components. These frictions may vary considerably over time (see “Trading Frictions Over the Long Run”), making realistic backtests very difficult. Alternatively, they involve tracking errors, management fees and switching frictions for funds employed as index proxies.
- The spot commodity indexes used in the study, which involve prices for immediate settlement and delivery of commodities (no roll return), may have monthly returns materially different from those for associated futures contracts traded by investors.
- There may be data snooping bias in selection of the three-month lagged growth rate as the BDI predictor. A sensitivity test using other past growth rate intervals would mitigate this concern.
- Results may be sensitive to the speed with which an investor implements BDI signals.
