Estimation of Cobb-Douglas production function parameter through a robust partial least squares

AThis research aimed to develop a new model for estimation problems of Cobb-Douglas production function. Specifically, the researcher focused on the problems of outliers and multicollinearity, and on solutions that were proposed to deal with these issues simultaneously. The Cobb-Douglas production function was usually fitted by first linearizing the models through logarithmic transformation and then applying the method of least squares. However the log transformation process did not get rid of all the outliers and multicollinearity problems. These problems appeared to have wide repercussions. Yet, these problems had not been known for their written solutions, and no serious research had been conducted in this field. The researcher attempted to develop the best possible method for providing estimates in the context of the Libyan agricultural sector by using the robust partial least squares path modeling (RPLS­PM). The results from RPLS-PM analysis showed that each block consisted of strong latent variables and one or more indicator variables. Furthermore, the results revealed that the standard measures of reliability, validity and model fit represented their respective latent constructs adequately. In addition, an overall assessment of the structural model was carried out based on the goodness of fit (GoF) index. The GoF value obtained indicated that the model fit the data. In other words, the findings from the model correspond to very good results and provide important new insights. This supports the thesis that this approach has the potential for solving problems in the agricultural sector as well as in other sectors. Also, the researcher noted that the model performed better than the least squares method when the multicollinearity and outlier problems existed in the data.