Merchant Commodity Storage Practice Revisited

Abstract

Commodity merchants use various heuristics to value leasing contracts on storage facilities as real options and make inventory trading decisions. Two prominent heuristics sequentially reoptimize simple models, leading to the so-called rolling intrinsic (RI) policy and rolling basket of spread options (RSO) policy. The extant literature numerically demonstrates that these two policies are nearly optimal in many realistic settings and can be used with Monte Carlo simulation to obtain fairly accurate estimates of the value of storage contracts. This paper provides a theoretical basis for the observed benefit of reoptimization with these heuristics and additional numerical evidence for the near optimal performance of the RI and RSO policies in several practical cases, but shows that the RI policy significantly outperforms the RSO policy in some of these cases. This research also proves that the RSO policy has a double basestock target structure, a known property of an optimal policy that is trivially true for the RI policy. Moreover, this work develops efficient and effective dual bounds to assess the performance of merchant commodity storage heuristics. In particular, these bounds are immediately relevant to the developers and users of the two considered heuristics.