Alternation Bias and the Parameterization of Cumulative Prospect Theory

Abstract

Two recently published studies argue that conventional parameterizations of cumulative prospect theory (CPT) fail to resolve the St. Petersburg Paradox. Yet as a descriptive theory CPT is not intended to account for the local representativeness effect, which is known to induce 'alternation bias' on binary iid sequences such as those generated by coin tossing in St. Petersburg gambles. Once alternation bias is controlled for, conventional parameterizations of CPT yield finite certainty equivalents for the St. Petersburg gamble, negating the suggested need for reparameterization. Moreover, the associated willingness to pay estimates fall within the generally accepted empirical range.