The return behavior of commodities is different. If you don't believe this, read the recent paper, "Anomalies in Commodity Futures Markets: Risk and Mispricing". This exhaustive research paper looks at a battery of prominent risk factors and premia that have been tested in other financial markets against a large dataset of commodity prices. The authors attempt to catalogue where there are sizable risk premia priced in the market and measure factors which do not seem to be priced in commodities. There is a lot of information to be digested with a wide range of conclusions; however, it provides more information on the uniqueness of these markets relative to equity and fixed income sectors.
Of course, commodities are diverse and generally have lower correlations across markets than what can be found in equity and fixed income markets. Still, there is a likelihood that risk premia fund in traditional markets should be present in this asset class. The authors look at three major categories for commodity sorts based on risk or mispricing variables, trading frictions, and moments. The research is conducted on 26 commodity markets from data that spans from 1959 through 2015.
Of course, commodities are diverse and generally have lower correlations across markets than what can be found in equity and fixed income markets. Still, there is a likelihood that risk premia fund in traditional markets should be present in this asset class. The authors look at three major categories for commodity sorts based on risk or mispricing variables, trading frictions, and moments. The research is conducted on 26 commodity markets from data that spans from 1959 through 2015.
These sorts are categorized by returns as well as factor models which include: CAPM, 3, 4, and 5-factor alpha models, a commodity model and the commodity-focused FFFM alpha model. The details of the sort specifications can be found in the paper. These allow for comparisons against similar analysis done in equity and fixed income markets. We have provided a table of results between the high and low sorts and the level of significance.
The testing shows that there are significant premia pricing with respect to aggregate jump risk, momentum, historical skew and kurtosis. There is marginal significance with 3 and 5-year reversals and the volatility of volatility. a number of tests prove to be inconclusive. Some of these results have been found by other researchers, but it tells a good story about why certain strategies like trend-following with some form or risk management can be successful in these markets.
1 comment:
You'd benefit from Popper's book "The Poverty of Historicism."
History moves only forward. The world that existed when I began to write a few seconds ago is gone forever and cannot be recreated. Every event, every moment, is unique.
We have no idea how commodities will behave in the future; we certainly cannot discover how they will behave by studying the past.
Studies that use data mining are enticing, tempting, and fruitless.
Of course, the real question is why we fall for them?
Post a Comment