In terms of the nature of production process, all university and college are organizations that consists of production process with multiple inputs, multiple outputs and heterogeneous sub-production processes. The output of a particular sub-process can be the input of another sub-process, therefore becomes an intermediate product.
The university or college mainly consists of three major sectors: 1) the academic sector, 2) the administrative sector, and 3) the library. The academic sector is the major sector that produces the final product. The function of administrative sector is to provide support for the teaching and research activities of the teachers and students. The mission of library is to collect, organize and deliver the information needed by the academic sector. There is a significant heterogeneity in the nature of production process among these three sectors.
However, the previous research on university or college efficiency evaluation using Data Envelopment Analysis (DEA) has been conducted mostly basing on the hypothesis that there exists a set of similar peer Decision Making Units (DMU).
In addition, there has been no literature discussing the role of intermediate product in the overall production process within the university and college. In this dissertation, the technical analysis of efficiency evaluation will emphasize the role of intermediate product in the overall production process of the university and college. This setting helps enhance the accuracy in evaluating the production efficiency of the university and college.
The research objectives of this dissertation are listed as follows:
1.Propose a new efficiency evaluation model that reflects the real organization structures of the university and college. The modified DEA model encompasses multiple inputs, multiple outputs, and multiple heterogeneous sub-production processes with emphasis on the role of intermediate product within the organization structure. The calculated efficiency index from this proposed modified DEA model was then compared to that attained from the conventional DEA model.
2.Explore the correlation between the overall efficiency and individual sector efficiencies as well as the correlation among various individual sector efficiencies.
3.Utilize regression analysis to explore the factors influencing the individual sector efficiencies and overall efficiency.
The conclusions are summarized below:
1.The results of overall efficiency evaluation can vary, depending on the models used (Fare 1991, modified Fare model and Charnes, Cooper and Rodnes Model 1978), the degree of input variables aggregation and the measurement of intermediate products (the optimal quantity and actual quantity).
2.There exist significant linear correlation between the overall efficiency of the university or college and the efficiency of individual academic sector, library sector and administrative sector. The academic sector efficiency has the highest Spearman correlation coefficient with the overall efficiency, with the library being second and the administrative sector a distant third. Comparing the correlation coefficients among the overall efficiency and each individual sector efficiency from different evaluation models reveals that CCR model underestimates the importance of the academic sector and overestimates the importance of administrative and library sector.
3.There are no linear correlation among the individual efficiency of academic, administrative, and library sectors.
4.The results from regression analysis of individual sector efficiency are summarized as follows:
1) The regression analysis on academic sector was analyzed according to two sets of variables. The analysis of the first set of variables reveals that the academic sector efficiency correlates positively with the science orientation (SIC) and student--teacher ratio (STR). The analysis of the second set of variables indicates that the academic sector efficiency correlates positively with the diversity (DIV) and average class size (CLA).
2) The regression analysis on administrative sector was discussed according to two sets of variables. The analysis of the first set of variables reveals that the administrative sector efficiency correlates positively with the science orientation (SIC) and average class size (CLA). The analysis of the second set of variables indicates that the administrative sector efficiency correlates positively with the size of school (SIZE) but inversely with the diversity (DIV).
3) The regression analysis on library sector was discussed according to two sets of variables. The analysis of the first set of variables reveals that the library sector efficiency correlates positively with the science orientation (SIC) and student--teacher ratio (STR). The analysis of the second set of variables indicates that the library sector efficiency correlates positively with the student-teacher ratio (STR). The correlation with diversity (DIV) is inconclusive.
5.The results from regression analysis of overall efficiency are summarized as follows:
1) After adjusting with environmental variables, the academic sector efficiency (RAC) and the library sector efficiency (RLB) have a positive and significant influence on the overall efficiency. However, the administrative sector efficiency (RAD) does not have a stable correlation with the overall efficiency. In addition, most of the administrative sector efficiency (RAD) coefficients are not significant. The academic sector efficiency has the highest correlation coefficient, with library the second and the administrative sector the last. Furthermore, the effect of academic sector on the overall efficiency is higher in the Fare (1991) model and the modified Fare model than in the CCR model.
2) In terms of environmental variables, the coefficient of property ownership variable (OWN) is negative, which suggests that the private school is more efficient than the public school, and those of school history (RHIST) are positive therefore the more established school show higher efficiency.
For strategy variables, the coefficients of the science orientation (SCI) and school history (RHIST) are positive, whereas the coefficients of the part-time to full-time faculty ratio (RPFR) are negative. In all the school sampling, only the CCR model indicates that diversity (RDIV) has a significant negative correlation with the overall efficiency, which suggests that lowering the diversity helps elevate the overall school efficiency.
6.To further explore the correlation between the overall efficiency and the diversity (RDIV), the samples were separated into two sets according to the science orientation (SIC). In the samples where the science and engineering are the majority (SIC greater than 0.5), the coefficient of diversity (RDIV) is positive, which indicates the higher the diversity, the higher the overall school efficiency. In schools where liberal arts and social science are the majority (SIC lower than 0.5), the coefficient of diversity (RDIV) is negative, which indicates the higher the diversity, the lower the overall school efficiency.
Based on the findings in this dissertation, we recommend that the environmental and strategy variables used in the regression analysis and their correlation with efficiency can be good reference tools for the education and school authorities when distributing education resources and adopting education and management policies. The environmental variables include the school history and school property ownership. According to the nature, the strategy variables can be categorized as: developmental characteristics (diversification of departments and science orientation), school member structure (part-time to full--time teacher ratio and student-to-teacher ratio), and the size of school (total number of students, average class size and student-to-teacher ratio).