The Effects of Market Differences on the Throughput of Large Container Terminals with Similar Levels of Efficiency

Abstract

Aggregate comparisons of the efficiency of container ports using techniques such as Data Envelopment Analysis usually rely on the number of Twenty-foot Equivalent Units (TEUs) moved across the quay face per year to provide the measure of output from which to derive efficiency measures. This may be due in part to the difficulty of obtaining large sample publicly available data sets at a more disaggregated level. Detailed simulation of individual container terminals using actual vessel sizes and schedules, together with detailed breakdowns of containers by size and intermodal transfer (land-sea or maritime transshipment), suggests that the nature of the market served can have a significant effect on the throughput of terminals managed and operated at similar levels of efficiency. The manufacturing centres in the Pearl River Delta are served by large, world-class terminals at Yantian and Hong Kong. These use similar technologies, operating methods and yard management systems. They operate at similar levels of efficiency at the micro-level (crane container handling rates, etc), but over the last 6 years the markets they serve have steadily diverged. This paper discusses the results of a preliminary study of the effects of exogenous market differences on the potential throughput capacity of container terminals with similar levels of efficiency as measured at the micro-level. Factors considered include the proportion of containers of differing size in the market, vessel size, the number of container moves per vessel, and the relative size of the transshipment and landed container markets. The paper concludes that the aggregate effect of these factors on throughput as measured in TEUs can be very significant and that aggregate analyses of container terminal efficiency should disaggregate the output measure into separate components. This will require larger sample sizes and more comprehensive data sets, particularly if varying returns to scale are to be assessed. It also considers the specification of frontier models of container terminal efficiency and puts forward proposals to ensure that the requirements of isotonicity and homogeneity are addressed and that the sensitivity of these models to data errors, weakly correlated inputs and outlying observations are taken into account.