ABSTRACT

Container ports are a major component of internationaltrade and the global supply chain.  Hence,the improvement of port efficiency can have a significant impact on the wider maritimeeconomy.  This paper deconstructs arepresentation in the existing literature that neglects the heterogeneity ofindividual and group-specific terminal operators.  In its place, we present a hierarchical model tomake a connection between efficiency and terminal operator groupcharacteristics.  The paper develops a stochastic frontier model that controls notonly individual heterogeneity but also group-specific variations.  The model decomposes the total stochasticderivation from the frontier into inefficiency, individual heterogeneity,group-specific variations, and noise components, with the estimation beingperformed using Markov chain Monte Carlo simulations.  The validity of the model is tested with apanel of container terminal operator data from 1997-2004.  Our findings show that terminal operatorgroups are important in promoting terminal efficiency at the global level, and thatthe operators with stevedore backgrounds show a higher efficiency thancarriers.

Key Words:
Stochasticprocesses; Stochastic production frontier; Markov processes; Container terminaloperators; Port globalisation; Group-specific

1.  Introduction

1.1. Background

In recent years, operational researchmethods have gained considerable importance in econometrics.  The production and cost theories in economicsmake it possible to estimate production and cost functions empirically, andthus to investigate changes in both the productivity and technology of afirm.  The conventional stochastic frontiermethod for estimating a frontier assumes that all firms are successful inreaching the efficient frontier (and only deviate randomly).  If, however, firms are not always at thefrontier, then the conventional estimation method will not reflect theefficient production or cost frontier against which to measure efficiency.  Empirical estimations for the port productionfunction have been performed by Chang (1978) and Tongzon (1993), whereas Kimand Sachis (1986), Martínez-Budría et al. (2003), Martínez-Budría et al. (1999),and Jara-Díaz et al. (2002) estimated the cost functions of ports for bothsingle-output and multiple-output cases. Using a single frontier function, Liu (1995), Notteboom et al. (2000),and Estache et al. (2002) estimated production frontiers or cost frontiers whilerecognising that some ports may not be at the efficient frontier.