In this study data envelopment analysis models were applied to evaluate the relative efficiencies of the credit departments of farmers' associations (CDFAs) in Taiwan. The findings show that the overall efficiency scores are not best and scale for CDFAs in Taiwan is relative small. It implies that the reorganization of the CDFAs may be appropriate if more efficient organization is to be pursued. Thus, this study investigated CDFAs reorganization to increase the efficiency. The proposed CDFAs reorganization alternatives have higher average efficiency scores than the current CDFAs.
This article presents a case study in which data envelopment analysis (DEA) (Charnes et al., [
In this study, we conducted DEA to distinguish between the efficient and inefficient CDFAs. The use of DEA for evaluating the relative efficiencies of CDFAs showed a good understanding of the resources utilization of each CDFA. Based on the efficiency analysis, this study investigated the reorganization of the CDFAs and proposed different alternatives for reorganizing the CDFAs to improve the overall efficiency of CDFAs.
The organization of this article is as follows. Section II presents the foundations of the DEA models and reviews the related literatures. Section III reports the results and the final section is the conclusion.
The efficiency is measured applying the ratio of the aggregated output to the aggregated input. Following Charnes et al. ([
The DEA method is a deterministic nonparametric frontier approach that adopts the concept of relative comparison. In comparing the credit departments of farmers' associations being evaluated with one another, a mathematical programming approach that originated with Farrell ([
This concept was further expanded by Charnes et al. ([
In this study, two DEA models were applied: CCR model and BCC model. In particular, the CCR model produces constant returns to scale (CRTS) efficiency frontier. The evaluated relative efficiency of the CCR model is an overall (or aggregated) efficiency score. In addition, the efficiency scores of all DMUs are set to be between 0 and 1 in the DEA models.
The BCC model produces variable returns to scale (VRTS) efficiency frontier and evaluates both the technical efficiency and the scale efficiency. Thus, the overall efficiency can be decomposed into the technical efficiency and the scale efficiency. Indeed, the value of technical efficiency times the value of scale efficiency equal to the value of overall efficiency. Therefore, a DMU is overall efficient if and only if it is both technical efficient and scale efficient. A DMU that is not overall efficient could be either technical inefficient or scale inefficient or both technical and scale inefficient. Applying BCC model can specify the major sources causing overall inefficiency.
Relevant studies on mergers among financial institutions involve cost efficiency analysis based on data after the merger has taken place (Mester, [
In terms of the empirical analysis, the evaluation of the efficiency of the frontier approach may also be divided into two further approaches, namely, the parametric and nonparametric approaches. In contrast with the parametric approach, the nonparametric approach does not determine a priori the functional form of the production frontier. For this reason, it is not limited by the functional form and also does not require the many assumptions that arise from the use of statistical methods for function estimation and efficiency measurement. Moreover, the nonparametric approach is more straightforward than the parametric approach in terms of dealing with the evaluation problems associated with many outputs and inputs. For this reason, this study adopts the nonparametric approach for the subsequent empirical analysis.
This section details an empirical study applying DEA to measure the relative efficiencies of the CDFAs. Furthermore, we discuss possible alternatives for the reorganization of the CDFAs. Following Golany and Roll ([
According to Keeney and Raiffa ([
For the validation of the developed DEA model, we examined the assumptions of the 'isotonicity' relationships between the input and output factors, i.e. an increase in any input should not result in a decrease in any output (Charnes et al., [
In this study, we applied the CCR model, with constant returns to scale, to evaluate the overall efficiencies of all DMUs. The results of the CCR model are shown in Appendix Table A1. In addition, we used the BCC model, with variable returns to scale, to evaluate the technical efficiency and the scale efficiency. The results of the BCC model are discussed in the next section. Both the dual linear programming formulations of the CCR and BCC models were run 277 times, i.e. one for every DMU.
Based on the CCR results, the efficiency values can be obtained for the farmers associations in each village, township and city, as listed in Appendix Table A1. The empirical results regarding the operating efficiency of the farmers' associations at the village, township and city level in Taiwan show that their average technical efficiency is low. This result appears to hide the fact that these DMUs, because of their relatively small regional spheres of operation, do not possess economies of scale, or possibly, because they have experienced problems with their internal controls, they have been unable to compete with other institutions and have also been hampered by poor quality staff. Therefore, we used BCC model to decompose the total efficiency and to evaluate the technical efficiency and the scale efficiency in the next section.
We used BCC model to evaluate the technical efficiency and the scale efficiency of the CDFAs. The results of BCC model can show the major sources of inefficiency among the 258 inefficient districts (as shown in Appendix 1 Table A1) and also provide possible improvement directions to promote the overall efficiency for each inefficient CDFA. We found that 190 of the 258 inefficient CDFAs had the technical efficiency scores higher than the corresponding scale efficiency scores. The CDFA that has the scale efficiency less than 1 is called scale inefficient. This result implies that the overall inefficiencies of the eight districts are primarily due to the scale inefficiencies. A scale inefficient DMU that exceeds the most productive scale size will present decreasing returns to scale. Alternatively, a scale inefficient DMU that is smaller than the most productive size will present increasing returns to scale. For example, as shown in Appendix Table A1, the DMU 5 and 9 CDFA have the technical efficiency scores equal to 1, but the scale efficiency scores are less than 1. These two districts are technically efficient yet scale inefficient. They can possibly increase (decrease) their operation scales to improve their overall efficiencies because they present increasing returns to scale and decreasing returns to scale as shown in Appendix Table A1. These results implied that the relative scales of these CDFAs have unbalanced status. Thus, reorganizing the existing CDFA is one way to adjust the unbalanced scales and thus improve the returns to scales.
Based on the efficiency analysis, this study investigated the reorganization of the CDFAs and proposed two reorganization alternatives for discussion. In this study, we investigated two possible reorganization alternatives to reduce the number of CFDAs and to improve the resource utilization of the CDFAs. Summaries of the two different reorganization alternatives are presented as follows.
The evaluation of efficiency is basically only a process, the ultimate objective of which is to find out where the shortcomings of an institution lie, enabling improvements and increased efficiency. The approach adopted to improve efficiency can be implemented by adjusting institution inputs or outputs, for example, through slacks variable analysis, thus increasing efficiency. This is the approach already explained by Charnes et al. ([
Based on the work of Chen ([
Table 1. Operating efficiency of farmers' associations at the county- and city-levels after reorganization
DMUs 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Overall efficiency 1.00 0.77 1.00 0.83 0.85 0.91 0.82 0.83 0.81 0.73 0.84 0.78 0.81 0.91 0.80 0.88 0.84
The average efficiency scores of the new CDFAs from the two reorganization alternatives were between 0.83 and 0.85. The average efficiency score of the existing districts was 0.73. Thus, the average efficiency scores from the first and second reorganization alternatives are better than the current values. In particular, the second alternative is the best-proposed reorganization alternative because it has the highest average efficiency scores of the reorganized CDFAs. This result implies that combining adjacent inefficient districts that present increasing returns to scale and combining an inefficient CDFA with efficient CDFA can possibly increase the overall efficiencies. Although Taiwan has not reorganized the CDFAs, this study proposed the directions for reorganizing the CDFAs and provided feasible alternatives.
In this article, DEA models were applied to evaluate the relative efficiencies of the CDFAs in Taiwan. The empirical results regarding the operating efficiency of the farmers' associations at the village, township and city level in Taiwan show that their average technical efficiency is low. This result appears to hide the fact that these DMUs, because of their relatively small regional spheres of operation, do not possess economies of scale, or possibly, because they have experienced problems with their internal controls, they have been unable to compete with other institutions and have also been hampered by poor quality staff. Thus, this study also investigated reorganizing CDFAs by combining inefficient CDFAs to improve the current unbalanced status. The proposed reorganization alternatives can reduce the number of CDFAs from 277 to 96 or 17. The reorganized CDFAs of the proposed alternatives had better average efficiency scores than the current CCDFAs.
Although this study focused on the operation aspects of the CDFAs, external factors that are not controllable nor operational may affect the efficiency scores of the CDFAs. For example, NAN-TOU (DMU 131–143) that had the lowest average efficiency score covered the largest mountainous area. Thus, it was inherently more costly for NAN-TOU CDFAs to provide the same service as a compact urban CDFAs did. Nevertheless, NAN-TOU CDFAs has reorganized other CDFAs by merging several small CDFAs with a big CDFA to reduce the input resources to improve the efficiency.
The author would like to thank the Council of Agriculture of the Republic of China, Taiwan for financially supporting this research under Contract No. COA-91AS-1.6.1-FS-#1(
Table A1. Operating efficiency of credit departments of farmers' associations
DMUs Overall efficiency Technical efficiency Scale efficiency Returns to scale DMUs Overall efficiency Technical efficiency Scale efficiency Returns to scale 1 0.855 0.894 0.956 DRTS 34 0.626 0.903 0.693 IRTS 2 0.770 0.886 0.869 DRTS 35 0.782 0.886 0.883 DRTS 3 0.880 0.905 0.972 DRTS 36 0.665 1.000 0.665 IRTS 4 0.854 0.894 0.955 IRTS 37 0.673 1.000 0.673 IRTS 5 0.744 1.000 0.744 DRTS 38 0.775 0.905 0.856 IRTS 6 0.785 0.881 0.891 IRTS 39 0.671 0.904 0.742 IRTS 7 0.758 0.906 0.837 IRTS 40 0.783 0.906 0.864 IRTS 8 0.669 0.810 0.826 DRTS 41 0.905 0.906 0.999 DRTS 9 0.800 1.000 0.800 IRTS 42 0.774 0.904 0.856 IRTS 10 1.000 1.000 1.000 CRTS 43 0.782 1.000 0.782 IRTS 11 0.800 1.000 0.800 IRTS 44 0.788 0.868 0.908 DRTS 12 0.916 0.954 0.960 DRTS 45 0.901 0.902 0.999 DRTS 13 0.750 1.000 0.750 IRTS 46 0.674 0.902 0.747 DRTS 14 0.786 0.890 0.883 DRTS 47 0.672 0.846 0.794 DRTS 15 0.804 0.852 0.943 DRTS 48 0.689 1.000 0.689 IRTS 16 0.939 0.985 0.953 DRTS 49 0.706 0.854 0.827 DRTS 17 0.809 0.858 0.943 IRTS 50 0.669 0.907 0.738 IRTS 18 0.807 0.858 0.941 DRTS 51 0.921 0.999 0.922 DRTS 19 0.818 0.871 0.939 DRTS 52 0.695 0.907 0.766 IRTS 20 0.631 1.000 0.631 IRTS 53 0.942 0.998 0.944 IRTS 21 1.000 1.000 1.000 CRTS 54 0.786 0.905 0.869 IRTS 22 0.904 0.947 0.955 DRTS 55 0.811 0.860 0.942 IRTS 23 0.906 0.962 0.942 DRTS 56 0.783 0.904 0.866 IRTS 24 0.786 0.860 0.914 DRTS 57 0.805 0.903 0.891 IRTS 25 0.567 0.888 0.639 DRTS 58 0.802 0.903 0.888 DRTS 26 0.895 0.906 0.988 IRTS 59 0.642 0.894 0.718 DRTS 27 0.429 0.674 0.637 IRTS 60 0.908 0.996 0.912 IRTS 28 0.767 0.807 0.950 IRTS 61 1.000 1.000 1.000 CRTS 29 1.000 1.000 1.000 CRTS 62 0.936 0.986 0.949 IRTS 30 0.660 0.900 0.734 IRTS 63 0.672 0.900 0.747 IRTS 31 0.678 0.821 0.826 IRTS 64 0.659 0.898 0.734 IRTS 32 0.656 0.903 0.726 IRTS 65 0.896 0.928 0.966 IRTS 33 0.659 1.000 0.659 IRTS 66 0.662 0.868 0.763 IRTS 67 0.937 1.000 0.937 IRTS 100 0.716 0.905 0.791 DRTS 68 0.665 0.906 0.734 IRTS 101 0.667 0.897 0.744 DRTS 69 0.942 0.997 0.945 IRTS 102 0.656 0.894 0.734 DRTS 70 0.663 0.907 0.731 IRTS 103 0.716 0.866 0.827 DRTS 71 0.588 0.903 0.651 IRTS 104 0.653 0.890 0.733 IRTS 72 0.649 0.900 0.721 IRTS 105 0.904 0.983 0.920 DRTS 73 0.647 0.903 0.716 IRTS 106 0.892 0.906 0.985 IRTS 74 0.907 0.993 0.913 DRTS 107 0.630 0.907 0.695 DRTS 75 0.647 0.902 0.717 IRTS 108 1.000 1.000 1.000 CRTS 76 0.645 0.882 0.731 IRTS 109 0.940 0.998 0.942 DRTS 77 0.612 0.726 0.843 IRTS 110 0.544 0.905 0.601 IRTS 78 0.676 0.886 0.763 IRTS 111 0.568 0.907 0.626 IRTS 79 0.773 1.000 0.773 IRTS 112 0.522 0.906 0.576 DRTS 80 0.753 0.872 0.863 IRTS 113 0.567 0.899 0.631 IRTS 81 0.771 0.900 0.856 IRTS 114 0.565 0.904 0.625 IRTS 82 0.717 0.870 0.824 IRTS 115 0.926 0.998 0.928 DRTS 83 0.747 0.873 0.855 IRTS 116 0.945 0.993 0.952 DRTS 84 1.000 1.000 1.000 CRTS 117 0.941 0.988 0.952 DRTS 85 0.653 0.881 0.741 DRTS 118 0.650 0.889 0.731 IRTS 86 0.798 0.899 0.888 DRTS 119 0.692 0.817 0.847 IRTS 87 1.000 1.000 1.000 CRTS 120 0.660 0.896 0.737 IRTS 88 0.644 0.893 0.721 DRTS 121 0.771 0.904 0.853 IRTS 89 0.650 0.902 0.720 DRTS 122 0.949 0.998 0.951 DRTS 90 0.898 0.907 0.990 IRTS 123 0.661 0.903 0.732 IRTS 91 0.679 0.900 0.754 DRTS 124 0.613 0.895 0.685 IRTS 92 0.629 0.903 0.696 IRTS 125 0.706 0.906 0.779 IRTS 93 0.689 0.903 0.763 DRTS 126 0.676 0.892 0.758 IRTS 94 0.651 0.902 0.722 IRTS 127 0.667 0.904 0.738 IRTS 95 0.748 0.905 0.827 IRTS 128 0.919 0.985 0.933 IRTS 96 0.653 0.726 0.899 IRTS 129 0.434 0.871 0.498 IRTS 97 1.000 1.000 1.000 CRTS 130 0.667 0.900 0.741 IRTS 98 1.000 1.000 1.000 CRTS 131 0.653 1.000 0.653 IRTS 99 0.704 0.906 0.777 IRTS 132 0.665 0.808 0.823 IRTS 133 0.703 0.871 0.807 DRTS 166 0.895 0.906 0.988 IRTS 134 0.584 0.726 0.804 IRTS 167 0.665 1.000 0.665 IRTS 135 0.632 0.897 0.705 IRTS 168 0.686 0.901 0.761 IRTS 136 0.704 0.907 0.776 IRTS 169 0.674 1.000 0.674 IRTS 137 0.627 0.726 0.863 IRTS 170 0.662 0.906 0.731 IRTS 138 0.684 0.906 0.755 IRTS 171 0.626 0.726 0.862 IRTS 139 0.652 0.903 0.722 IRTS 172 0.698 0.906 0.771 IRTS 140 0.662 0.905 0.732 IRTS 173 0.764 0.905 0.844 IRTS 141 0.644 0.905 0.712 IRTS 174 0.765 0.862 0.887 IRTS 142 0.634 0.726 0.873 IRTS 175 0.765 0.906 0.844 IRTS 143 0.626 0.873 0.717 IRTS 176 0.772 0.896 0.862 DRTS 144 0.699 0.817 0.856 DRTS 177 0.764 1.000 0.764 IRTS 145 0.703 0.895 0.785 DRTS 178 0.787 0.906 0.869 IRTS 146 0.708 0.889 0.797 DRTS 179 0.764 0.898 0.851 IRTS 147 0.666 0.887 0.751 DRTS 180 0.728 0.880 0.827 IRTS 148 0.698 0.906 0.771 IRTS 181 0.551 0.833 0.661 IRTS 149 1.000 1.000 1.000 CRTS 182 0.676 1.000 0.676 IRTS 150 0.716 0.817 0.876 DRTS 183 0.670 0.904 0.741 IRTS 151 0.653 0.905 0.722 IRTS 184 0.696 1.000 0.696 IRTS 152 0.605 0.904 0.669 IRTS 185 0.928 0.971 0.956 DRTS 153 0.651 0.772 0.844 IRTS 186 0.668 0.907 0.737 IRTS 154 1.000 1.000 1.000 CRTS 187 0.690 1.000 0.690 IRTS 155 1.000 1.000 1.000 CRTS 188 0.678 0.906 0.748 IRTS 156 0.660 0.905 0.729 IRTS 189 0.925 0.998 0.927 IRTS 157 0.647 0.900 0.719 IRTS 190 0.940 0.999 0.941 IRTS 158 0.729 0.903 0.807 IRTS 191 0.671 0.900 0.745 IRTS 159 1.000 1.000 1.000 CRTS 192 0.633 0.907 0.698 DRTS 160 0.687 0.904 0.760 IRTS 193 1.000 1.000 1.000 CRTS 161 0.686 0.817 0.840 DRTS 194 0.804 0.889 0.905 DRTS 162 0.727 1.000 0.727 IRTS 195 0.817 0.906 0.902 DRTS 163 0.659 0.905 0.728 IRTS 196 0.787 0.898 0.877 IRTS 164 0.677 0.817 0.829 DRTS 197 0.677 0.904 0.749 IRTS 165 0.663 0.907 0.731 IRTS 198 0.797 0.904 0.882 IRTS 199 0.668 0.905 0.738 IRTS 232 0.693 1.000 0.693 IRTS 200 0.626 0.895 0.699 IRTS 233 0.671 0.901 0.744 DRTS 201 0.791 1.000 0.791 IRTS 234 0.807 0.894 0.903 IRTS 202 0.665 0.772 0.862 IRTS 235 0.666 0.898 0.742 IRTS 203 0.837 0.897 0.933 DRTS 236 0.696 0.901 0.772 DRTS 204 0.665 0.772 0.862 IRTS 237 0.819 0.870 0.942 DRTS 205 0.669 0.897 0.746 IRTS 238 0.695 0.901 0.771 DRTS 206 0.683 0.904 0.755 IRTS 239 0.671 0.901 0.744 IRTS 207 0.656 0.894 0.734 IRTS 240 0.714 1.000 0.714 IRTS 208 1.000 1.000 1.000 CRTS 241 1.000 1.000 1.000 CRTS 209 0.724 0.817 0.886 IRTS 242 0.604 0.852 0.709 IRTS 210 0.683 0.903 0.756 IRTS 243 0.657 0.873 0.752 IRTS 211 0.664 0.883 0.752 IRTS 244 0.695 0.902 0.770 DRTS 212 0.645 0.895 0.721 DRTS 245 0.682 1.000 0.682 IRTS 213 0.780 0.822 0.948 DRTS 246 0.660 0.817 0.808 IRTS 214 0.784 0.886 0.885 DRTS 247 0.685 0.904 0.758 IRTS 215 0.797 0.906 0.880 IRTS 248 0.609 0.896 0.680 IRTS 216 0.803 0.862 0.931 IRTS 249 0.750 1.000 0.750 IRTS 217 0.632 0.902 0.700 IRTS 250 0.620 1.000 0.620 IRTS 218 0.707 0.861 0.821 DRTS 251 1.000 1.000 1.000 CRTS 219 0.686 0.906 0.757 IRTS 252 0.638 0.900 0.709 IRTS 220 0.909 1.000 0.909 IRTS 253 0.701 0.904 0.775 DRTS 221 0.910 0.999 0.911 IRTS 254 0.725 0.904 0.802 DRTS 222 0.689 0.904 0.762 DRTS 255 0.633 0.896 0.707 IRTS 223 0.703 0.907 0.775 IRTS 256 0.583 0.905 0.644 IRTS 224 1.000 1.000 1.000 CRTS 257 0.747 0.905 0.825 DRTS 225 0.777 0.899 0.865 IRTS 258 0.686 0.868 0.791 IRTS 226 0.792 0.904 0.876 DRTS 259 0.681 0.824 0.826 IRTS 227 0.668 0.817 0.818 DRTS 260 0.588 0.880 0.669 IRTS 228 0.671 1.000 0.671 IRTS 261 0.660 1.000 0.660 IRTS 229 0.673 0.904 0.744 IRTS 262 0.721 0.905 0.797 DRTS 230 0.689 0.902 0.764 IRTS 263 0.943 0.991 0.952 IRTS 231 0.685 0.900 0.762 IRTS 264 0.630 0.887 0.710 IRTS 265 0.642 0.905 0.709 IRTS 266 0.793 0.839 0.945 IRTS 267 0.805 0.900 0.895 IRTS 268 0.709 0.900 0.787 IRTS 269 0.629 1.000 0.629 IRTS 270 0.644 0.902 0.714 IRTS 271 0.674 0.901 0.748 IRTS 272 0.650 0.901 0.721 IRTS 273 0.645 0.905 0.713 IRTS 274 1.000 1.000 1.000 CRTS 275 0.595 0.900 0.661 IRTS 276 0.667 0.908 0.735 IRTS 277 0.626 0.890 0.703 IRTS
Table A2. Operating efficiency of farmers' associations at the county- and city-level after partial reorganization
DMUs Overall efficiency DMUs Overall efficiency DMUs Overall efficiency 1 0.893 34 1 67 0.906 2 1 35 0.765 68 0.883 3 0.838 36 0.742 69 0.785 4 0.808 37 1 70 0.763 5 1 38 0.911 71 0.796 6 0.858 39 0.607 72 1 7 1 40 0.893 73 0.830 8 0.898 41 0.837 74 0.772 9 1 42 0.872 75 0.928 10 0.693 43 0.726 76 0.842 11 0.890 44 0.829 77 1 12 0.730 45 0.740 78 0.876 13 0.760 46 0.741 79 0.821 14 0.777 47 0.704 80 0.745 15 0.892 48 0.720 81 0.806 16 0.866 49 0.723 82 0.853 17 0.758 50 0.702 83 0.859 18 0.838 51 0.774 84 0.912 19 0.931 52 0.867 85 0.746 20 0.876 53 0.724 86 0.716 21 1 54 1 87 0.869 22 0.831 55 0.747 88 0.8118 23 0.857 56 0.870 89 0.772 24 1 57 0.762 90 0.769 25 0.697 58 0.819 91 1 26 0.807 59 0.742 92 0.757 27 0.709 60 0.728 93 0.893 28 0.842 61 0.841 94 0.766 29 0.820 62 0.914 95 1 30 0.899 63 0.749 96 0.742 31 0.841 64 0.749 32 0.809 65 0.838 33 0.765 66 1
By Chun-Chu Liu
Reported by Author