SERVICE MODEL ESTIMATION FOR TECHNICAL EFFICIENCY OF INTERNET SERVICE PROVIDERS IN WESTERN NIGERIA

Abstract

This work focuses on the provision of internet services to the consumers by the internet service providers in Western Nigeria. The technology of the internet service providers is assumed to be specified by the Stochastic Translog frontier model as a relationship among certain network characteristics taken in as inputs in the day-to-day provision of services. Consequently, an 'Internet Service Model Estimation Questionnaire' was designed and administered to a total of 203 randomly sampled internet service providers in Western Nigeria, in order to collect data about these network characteristics. The collected data were analysed using the Statistical Package for Social Scientists (SPSS)for Windows software and the Stochastic Frontier software. Two models were fitted from the analysed data using the two softwares and sample estimations were made with the models. Finally, a comparative analysis of estimations from the two models was carried out. The hypotheses that the Cobb-Douglas Model adequately represents the data were accepted in both methods. However, it was found that the Frontier estimates perform better because of the capability of the software to isolate the inefficiency errors in the prediction and the generation of firm-level technical efficiency for each of the service providers. It was found that although the mean technical efficiency is high (about 91.2%), a wide variation of about 64.9%exists between the lowest and the highest technically efficient service providers. Furthermore, network characteristics such as transmission link and transmission bandwidth have the most effect on the technical efficiency of a service provider in Western Nigeria.