Rational speculative bubbles and commodities markets: application of duration dependence test

Abstract

The presence of rational speculative bubbles in 28 commodities is investigated using the duration dependence test on the stochastic interest-adjusted basis. Eleven of 28 commodities experienced some episodes of rational speculative bubble. These commodities are West Texas Intermediate (WTI) crude oil, coffee, corn, soya bean No. 2, soya bean meal and oil, wheat No. 2 soft red and hard winter wheat, lean hogs, gold and platinum. Additionally, natural gas, propane, live cattle, and pork bellies exhibit mean-reversion in the interest-adjusted basis.

Keywords:

• commodities markets
• commodities futures
• duration dependence
• nonparametric methods
• convenience yield
• speculative bubble

JEL Classification::

• C14
• C22
• C32
• F31
• G12
• G15

Notes

¹ For details, see Milonas (1986), Ng and Pirrong (1994), and Heaney (2002).

² Both Gibson and Schwartz (1990) and Schwartz (1997) developed stochastic convenience yield models. The practical benefit of these stochastic convenience yield models lies in developing term structures of commodities future prices. They do not, however, describe the endogenous factors determining the convenience yield.

³ Normally, convenience yield models incorporate the effects of reduced marketability due to the daily mark-to-market and the possibly incorrect pricing of the contract in relation to the underlying due to maturity, storage and convenience effects (Brenner and Kroner, 1995; Milonas and Henker, 2001; Ludkovski and Carmona, 2004; Casassus and Collin-Dufresne, 2005).

⁴ For full derivation of this equation, see McQueen and Thorley (1994).

⁵ Since each series is analysed individually, using all the data available is better than cutting the most data to match the shortest series in the group.

⁶ The IAB basis estimates are comparable to those derived by Heaney (2002).

⁷ To conserve some space, the autocorrelation coefficients and Ljung–Box Q-statistics for each commodity are not reported here but are available from authors upon request.

⁸ For Brent crude oil, there are exactly four runs that last for 4 months and 13 runs that last at least for 4 months. The hazard rate states that if a positive run persists for four consecutive months, there is a 30.80% probability that the bubble will burst in the next month. For the WTI, there is a 25.00% probability that positive run lasting for 4 months will revert to a negative return in the next period. The hazard rate of other commodities can be interpreted in the same way.

⁹ Some authors even dispute 1929 crash as bubble. See McGrattan and Prescott (2001) for example.