The economics of investing in Strategic Petroleum Reserves
The optimal size of the SPR follows from the general model of investment in the servicing of a fluctuating or uncertain demand. The marginal (or average) costs per barrel per year are the annual capital costs per barrel of buying and storing the oil stock. The marginal benefits per barrel per year are the benefits of mitigating the oil price rise by releasing the last barrel, times the probability per year that the SPR is exhausted. This mitigation depends on the welfare loss of an oil price rise (for the oil consuming countries) and the price elasticity of the demand for oil. Also, being a small country which is part of a larger SPR system decreases the ratio between marginal costs and marginal benefits.
Using observed or assumed values of these quantities, one can compute whether a particular SPR is too large or too small. This is done in a spreadsheet-like table with a few input data and a few simple formulas, applied to the EU and the USA. In this way one can easily play with the essential assumptions behind decisions about SPR systems, and see what happens to the outcome. We find that the SPR seem too large. Of course, this result is subject to many limitations, such as the uncertainties surrounding the assumptions.
Published as section 3 of M. Mulder, A. ten Cate and G. Zwart, The economics of promoting security of energy supply, EIB Papers, volume 12, nr 2 (2007)
Comparison with other work at CPB
In Energy Policies and Risks on Energy Markets; A cost-benefit analysis, the break-even point (with zero net benefits) is computed, based on one fixed oil crisis, while here the net benefits are maximised, based on the distribution of oil crises. Also, "Energy Policies ..." takes into account the fact that the oil stocks can be released only in limited quantities, which is ignored here.
In Optimal safety standards for dike-ring areas, the residual damage plus the costs are minimised, while here the benefits (the avoided damage) minus the costs are maximised. (The latter formulation fits better with the term "cost-benefit analysis" and with the usual equality of marginal costs and marginal benefits in an optimum.) Of course the result is the same.