Robust optimization policy benchmarks and modeling errors in natural gas

Abstract

The problem of regulating natural gas procurement has become a huge burden to regulators, especially due to the plethora of complicated financial contracts that are now being used by local distribution companies (LDCs) for risk management purposes. Muthuraman, Aouam, and Rardin (2008) proposed a new benchmarking scheme, called policy benchmarks and showed that these benchmarks do not suffer from the usual criticisms that are made against existing regulatory methods. Such policy benchmarks based regulation has however faced hurdles in being adopted. One of the primary reasons has been concerns over its robustness.We demonstrate in this paper that when modeling errors are present, the policy benchmarks proposed earlier can backfire and are hence, as suspected, not well suited for regulation. We begin our analysis with a more general model than the one that has been used earlier by accommodating the LDC’s ability to reduce cost by exerting effort, as in classical economics. We derive solutions to the LDCÕs problem, find closed form solutions for the regulator’s optimal fee fraction along with risk sharing implications, and provide insights into the policy benchmark selection. We then construct a robust-optimization based policy benchmarking mechanism that inherits all the original benefits. We further demonstrate that these, unlike the earlier benchmarks, are robust against modeling errors.