Author : Viroj Pimmanrojnagool

Summary : Yield prediction models with the use of a system of equations model was the primary subject of this study. The study explored the simultaneous equations model for yield prediction and compared it with the ordinary least squares method. It also attemptedto develop a yield prediction model for Pinus kesiya plantations in Chiengmai, Thailand. The model presented in this study was composed of two equations: stand basal area and yield equations. Based on two hundred forty-seven 10 meter radius circular plots located and measured in four Pinus kesiya plantations in the study site, the model used the two stage least squares and ordinary least squares method. With the management regime and genetic variation held constant, yield was shown to be adequately explained by stand age, site index, original stand spacing, and stand basal area. The results obtained in deriving the model using both the methods of two-stage least squares and ordinary least squares were similar in some respects. First, the component equations of the model were found to be highly significant. The basal area and yield equations provided very high values of R (superscript 2). Second, both methods provided similar yield curve which conformed to the properties of a theoretical yield curve (being S-shaped) and satisfying the differentiation properties. Third, using the chi-square test of accuracy, both the methods of two-stage least squares and ordinary least squares gave the same level of accuracy, which was within 11.5 percent of the true value at .05 significance level. Theory, however, tells us that with a system of equations like the one developed in this study, the method of ordinary least squares could give inconsistent estimators while the two-stage least squares method could provide consistent and asymptotically efficient estimators of the system parameters. As such, the results obtained by the method of two-stage least squares were used to specify the yield prediction model for the Pinus kesiya plantations in Chiengmai, Thailand. A system of equations model without original stand spacing and a single equation model were also developed in this study for purposes for comparison. Based on the coefficient of determination, the properties of the theoretical yield curve, and level of accuracy, the two-stage least squares model gave slightly better results. The level of accuracy was higher by 0.9 percent for the proposed model than the model without original stand spacing, and higher by about 1.2 percent than the single equation model.