Tactical Asset Allocation with Commodity Futures
Buch in deiner Nähe kaufen
Beschreibung
The thesis is based on three separate studies, each analyzing different properties of commodity derivatives and using different methodologies.
Chapter 2 provides an introduction to the specific features of commodity derivatives as an asset class and empirically tests the theory of storage for a set of 19 futures and forward contracts. There is a negative inventory - convenience yield relationship for most commodities in the sample and marginal convenience yields are falling as discretionary inventory increases. Scarcity, defined as times when the stock of discretionary inventory is reduced or negative, respectively, is a significant risk factor for futures returns. Hence, a lagged proxy for scarcity serves as predictive variable for returns of industrial metal and energy contracts. Simple trading strategies going long futures contracts in states of scarcity and short otherwise yield statistically and economically significant excess returns for a number of commodity contracts. These abnormal returns cannot be explained by neither a commodity risk factor, an equity risk factor or net hedging pressure. Implementing the scarcity proxy as conditioning information in a stochastic discount factor setting, conditional bounds are significantly sharper than classical volatility bounds and mean-variance frontiers are wider than fixed weight frontiers for commodity futures with industrial use.
In chapter 3, I study conditional volatilities and correlations over a broad sample of 25 commodity futures and the S&P500 equity index. Fitting various GARCH models to the data, volatilities are significantly asymmetric for about half of the contracts in the sample. Seven futures show higher volatility when past returns were positive while six commodities and the equity index are more volatile in down states. Conditional correlations within the commodity complexes and between commodities and the S&P500 equity index are evaluated by an asymmetric, multivariate DCC-GARCH model. Correlation parameters are significantly asymmetric for a number of futures contracts and the equity index. In general, correlations tend to rise in joint bad states where return innovations of both assets have been negative in the past period. For Crude Oil and equity returns, unconditional correlations are close to zero but correlations increase to 0.6 for joint bad states. Furthermore, correlations between commodity futures and equity returns tend to increase in bad states of the economy; hence, reducing the benefit of commodities for portfolio diversification. However, correlation dynamics between precious metals, notably Gold futures, and equity returns are fairly symmetric and correlations remain low in joint bad states as well as over the business cycle.
Chapter 4 builds on two major insights of the previous studies. The predictive power of storage for futures returns and the need for active asset allocation within portfolios of commodity futures. I construct long/short portfolios of futures contracts based on two distinct trading signals, price momentum and changes in discretionary inventory. Both, the momentum and the storage signal allow for an outperformance to the passive GSCI benchmark index but the double-sort strategies, jointly evaluating both signals, achieve the highest returns. The return differentials between buy portfolios, holding futures contracts with previously high returns and relatively low inventory, and the sell portfolios (low momentum, large inventories) are significantly positive and economically relevant. Average returns for the best performing active strategy are over four times as large as for the benchmark index (Annualized geometric total return of 28.9% vs. 6.1% over the sample period of 28 years.). Active strategies are riskier than the passive index but the volatility and the selection risks involved in the long/short trading methodology can be reduced considerably if the buy and sell portfolios are diversified over a larger number of individual contracts.
Autor*in
Stefan Dätwyler