Measuring stock performance using stochastic frontier analysis model with dependent error approach

Abstract

This paper focuses on analyzing the technical efficiency of Malaysian stock performance over the period of 2013 to 2018. By utilizing the stochastic frontier analysis (SFA) production function Cobb-Douglas, the inefficiency effect of time-invariant is allowed and predicted to estimate the technical efficiency score as well as provide a ranking efficiency based on the model estimation performance. In SFA, the two main errors, random error and inefficiency error are assumed to be independent, and this assumption is not practical in a real-life situation. The assumption for random error is normally distributed and the inefficiency error is half-normal distributed. Therefore, in this paper, when the assumption of SFA is dependent on both errors, the copula is applied to capture the joint distribution of these two error components. These main findings revealed that stock efficiency estimates using copula SFA (CSFA) are appropriate because it uses more practical assumptions and among the seven models, through the AIC method, the Cot copula was selected as the best model. This paper provides new evidence on comparison ranking of technical efficiency based on the three models, yielded by copulas with SFA (CSFA-Cot copula), SFA, and DEA-CCR models. Spearman’s rank order was implemented and revealed that there was a high degree of correlation found among the rank efficiency estimates derived from the models of CSFA and SFA applied. However, the scores produced by both models are different. Accurate scores are necessary in order to make correct decisions and predictions. Therefore, the dependence error between random error and inefficiency error cannot be ignored, and the Cot copula in SFA models can be considered as an alternative suitable tool for measuring efficiency performance.