Measuring port efficiency using bootstrapped DEA: the case of Vietnamese ports

摘要

Standard data envelopment analysis (DEA) tends to be sensitive to the number of variables of a chosen sample, and it is unable to account for their random nature. Standard DEA can exhibit statistical inconsistency, biased results, and an arguable inference process. Thus, in this study, an efficiency evaluation method is used to overcome these limitations, especially since no studies of port efficiency have addressed this issue. This study applies bootstrapped DEA to a sample of the 43 largest Vietnamese ports and compares the results with those from stochastic frontier analysis (SFA) and standard DEA. The results show that while the efficiency scores obtained from the three methods provide useful and consistent measures of the ports' efficiency, they differ significantly. Furthermore, while the efficiency scores produced by bootstrapped DEA are consistent, unbiased, and not sensitive to the sample size, standard DEA and SFA yield efficiency scores that are much larger than bootstrapped DEA. In addition, bootstrapped DEA provides the confidence intervals for efficiency scores and allows for hypothesis tests of port performance.