Cost-Benefit Analysis as a Statistical Hypothesis Test: An Example from Urban Transportation

Abstract

A method to evaluate the estimated social benefits and costs of many urban transportation decisions is presented, based on the techniques of statistical hypothesis testing. After a slight critique of current practice in cost-benefit analyses, a general equilibrium model is formulated that includes many of the relevant variables and decisions which interact with an urban transportation system. The model is based on individual behavior assumptions (as opposed to macro-behavior) and uses the concept that commodities provide attributes. The significant omissions are mentioned and discussed. It is next shown that if the model is true, then hypotheses involving the Pareto-superiority (and Pareto-optimality) of various decisions with respect to transportation systems imply hypotheses restricting the values of parameters of a single linear equation. This allows the testing (i.e. falsification) of the original hypotheses by using the familiar T and F tests of regression analysis. These tests are based on a single-equation regression only using observations on the rates of use of, the attributes provided by, and the inputs to, the transportation system. The simplicity of the tests seems to indicate that revisions of the model towards reality would be very productive.