Application of bootstrap techniques in econometrics: the example of cost estimation in the automotive industry

Bootstrap methods applied in regression models help to approximate the distributions of the coefficients and the prediction errors. In this paper, we apply bootstrap techniques to determine prediction intervals from econometric models when the regressors are known. Ve investigate problems associated with their application: determining the number of replications, choosing the method to calculate the least-squares estimator (pseudo-inverse or inverse) and sorting algorithm? of the statistic of interest. This investigation arises from? the need in the automotive industry to predict costs in the early phases of development of a new vehicle. Generally, the sample size is small and the model's error term of the model is not Gaussian. Consequently, bootstrap techniques strongly improve prediction intervals by reflecting the original distribution of the data. Two examples (engine and fuel tank) illustrate the technique.