MODELLING DAILY OIL PRICE DATA USING AUTO-REGRESSIVE JUMP-INTENSITY GARCH MODELS

Marc Gronwald, ifo Institute for Economic Research, Phone +49 89 9224 1400, e-mail:

Overview

The paper's contribution to the literature is twofold. First, itapplies Chan and Maheu's (2002) auto-regressive jump-intensity(ARJI)-GARCH model to daily oil price data from March 1983 toNovember 2008 in order to get a better understanding of the oilprice's behavior. Jump models, in general, have been proven to be auseful tool for capturing unexpected news.More specifically,ARJI-models allow the jumps to occur at differing size and

frequency. This model class has been successfully applied to varioustypes of financial market data, including exchange rates, interestrates andstock market returns. Second, it relates its empirical findings to a number of important theoreticalconsiderations. Firstly, Hotelling's (1931) seminal paper showsthat, as oil is an exhaustible resource, its price grows, in optimum, at the rate of interest.Secondly, oil isalso one of the main sources of carbon emissions. Sinn (2008)extends Hotelling's (1931) work by considering the issue of global warming andshows that ignoring this issue leads to a current overextraction of oil.Holland (2008), finally, shows that the oil price is the betterscarcity indicator than oil production is.

The paper is organized as follows: after motivating the paper, a descriptive analysis of the data is undertaken and the empirical method is outlined. Subsequently,the empirical results are presented and discussed.

Methods

Auto-regressive Jump Intensity GARCH models proposed by Chan and Maheu (2002).

Results

Strong evidence of conditional jump-intensity in daily oil pricechanges is found, indicated by highly significant jump coefficients.

Conclusions

The empirical findings imply that the oil price exhibits a behaviour that is frequently found in various types of financial market data. Conditional heteroscedasticity is present and the empirical distribution of oil price changes has heavy tails. Moreover, the oil price is very sensitive to news and the presence of jumps indicates that the oil price does not settle around a trend. Thus, these results are at odds with the notion of deterministic trends in natural resource prices [Slade, 1982; Lee et al., 2006]. Information extracted from the price of oil, however, is a crucial part of the theories by Hotelling (1931), Sinn (2008), and Holland (2008) sketched above. The empirical findings suggest that the traditional Hotelling-type models need to be extended by assuming the resource price to follow a jump process. This is likely to change the model outcomes.Most certainly, finding both optimal extraction paths and the optimaldecision regarding the transition to alternative technologies is hampered. What is more, the blatant non-existence of a long-run trend is likely to cause a current overextraction of oil, which has severe consequencesfor the global climate and, thus, for one of the greatest challengesof mankind.

References

Chan, W.H. and J.M.Maheu (2002), “Conditional Jump Dynamics in Stock Market Returns”, Journal of Business & Economic Statistics 20: 377-389

Hotelling, H. (1931), “The Economics of Exhaustible Resources”, The Journal of Political Economy 39, 137-175

Lee, J., J.A.List and M.C.Strazicich (2006), “Non-renewable Resource Prices: Deterministic or Stochastic Trends?”, Journal of Environmental Economics and Management, 51: 354-370

Sinn, H.W. (2008), “Public Policies against Global Warming: a Supply Side Approach”, International Tax and Public Finance 15, 360-394

Holland, S.P. (2008), “Modelling Peak Oil”, The Energy Journal 29: 61-79

Slade, M. (1982), “Trends in Natural-Resource Commodity Prices: An Analysis of the Time-Domain”, Journal of Environmental Economics and Management, 9, 122-137