The Winner's Curse and Cost Estimation Bias in Pioneer Projects

Abstract

Cost overruns are almost as massive and almost as pervasive on private "first of a kind" projects as on defense contracts. This paper examines the cost overrun problem in terms of the methodology of cost estimation, abstracting from moral hazard problems. The main results of the paper are these: first, if cost estimators are an unbiased estimation methodology, then under certain monotonicity conditions, this will produce an observed cost underestimation bias, because cost estimates are used as a guide to project decision making; second, the more uncertainty there is with respect to the costs of a project, the larger will be the observed cost underestimation bias, assuming the estimator uses an unbiased estimation methodology; third, in bottoms up estimation, the most accurate of cost estimation methodologies, there is a built in cost underestimation bias because the value of information is not incorporated into the cost estimate; fourth, the size of the underestimation bias in bottoms up estimation definitely-increases with uncertainty only under rather stringent conditions on the construction production function.