Long-Term Cycles and the Prediction of Petroleum Demand

Abstract

Summary A great amount of effort is expended in attempting to predict both short- and long-term petroleum demand. This effort has become more difficult because oil markets have become more unpredictable. This paper presents a new look at long-term petroleum demand trends in terms of the presents a new look at long-term petroleum demand trends in terms of the derivative of Hubbert's life-cycle curve. The data reveal a long-term trend with fluctuations that result from recurring business cycles. It seems reasonable to extrapolate this trend into the future. A straight-line equation was derived by the least-squares method and extrapolated to the year 2000. Results suggest that average year-to-year growth of demand will be slightly positive until 2000, but the peak oil consumption that was reached in 1979 will not be exceeded until well into the 1990's, if at all. Introduction Strategic planning in any business must operate with certain assumptions about overall product demand. Estimation of both short- and long-term market movement requires much effort, particularly as oil markets have become more unpredictable. It is useful to diverge from the focus on short-term uncertainties and determine whether recent events can fit into a coherent, longterm viewpoint. Fig. 1 gives total U.S. liquid-hydrocarbon demand through 1984. Before 1973, demand increased almost continuously along a smooth curve. Planning could be done with the ostensible assurance of an expanded future market to absorb new production capacity. Although forecasters were aware, on an abstract basis, that petroleum is a finite resource and that exponential growth, therefore, must cease at some distant point, few near-term (5- to 10-year) projections assumed anything other than continued exponential growth (Fig. 1). The previously unprecedented changes in price and demand after 1973 spawned a decade of heightened efforts to predict and to influence (by government action) the demand for liquid hydrocarbons. After 1975, a resumption of exponential demand growth led many forecasters to the simplistic assumption that the 1973–75 period was only an aberration. It was asserted that after the oil-price increases had been absorbed into the economy, the market would continue to expand at an exponential rate until some time in the future as determined by economics and the depletion of the resource. Others, however, seriously discussed whether the limits to growth were being approached and whether demand would soon level out. It was pointed out that continuing the historical demand growth rate of 6%/yr would result in a clearly impossible 18-fold increase in only 50 years. In fact, any rate of exponential growth (a constant percentage rate of increase per year) becomes impossible if extrapolated far enough into the future. A hybrid forecast from the 1975–79 period would have assumed continuing market growth, but perhaps at a constant rate rather than an exponentially increasing rate (Fig. 1). What, then, can be expected for the future? It is obvious from Fig. 1 that there is no longer any basis for the simple extrapolation of past trends to predict future market demand. We could be poised to resume demand growth, to enter a period of stable demand, or to encounter further reductions. None of these scenarios appears to be incompatible with history. There are, of course, extensive efforts under way to apply increasingly sophisticated computer models and detailed economic analyses to the problem of predicting future demand patterns. This paper will not delve into those efforts but will be paper will not delve into those efforts but will be confined to looking at broad historical trends. Petroleum Production Petroleum Production Hubbert first drew widespread public attention to the finite life cycle of petroleum resources. In a remarkably prescient work, he fitted a logistic curve to U.S. prescient work, he fitted a logistic curve to U.S. production data through 1961 and predicted that U.S. domestic production data through 1961 and predicted that U.S. domestic production would reach a maximum in 1970. As shown production would reach a maximum in 1970. As shown in Fig. 2, U.S. domestic production (Curve B) did peak in 1970, although at a higher value than calculated by the Hubbert curve (Curve A). The Hubbert curve can be strictly constrained by geological analysis. The logistic equation used is Np = npa/(1+ae(-bt),...........................(1) where a = constant, b = constant, Np = cumulative production, nps = ultimately discoverable reserves, and t = time. A value for npa was calculated from geological analysis, and then values for a and b were determined by curvefitting the actual production data. The value of npa fixes the total area under the curve of production rate, Q = dNp/dt, vs. time. JPT P. 669