Was Harold Zurcher myopic after all? : replicating Rust's engine replacement estimates

Rust (1987) concludes that the data "clearly reject" the hypothesis that Harold Zurcher made bus replacement decisions using a monthly (and myopic) discount rate of 0 in favor of the factor 0.9999. The alternative model requires the nested fixed point algorithm developed in the paper which became the basis of an ongoing empirical literature. The p-value of the likelihood ratio test was 0.053. Recoding the preferred model and re-processing the raw data reveals two types of errors in the original analysis and a revised p-value of 0.078. This remains below .10, which can be inferred as the significance level that clear rejection was based on. Thus the myopic hypothesis is again rejected although for lower conventional significance levels it would not be.