本文根據臺灣儲蓄互助社1999年年度財務資料,以貸放社員作為產出變數,利息支出和其他營業支出為投入變數,利用資料包絡分析法,分別以投入導向和產出導向模型計算各社的總技術效率、純粹技術效率和規模效率。在投入導向模型中,本文發現324家儲互社純粹技術效率的平均值為0.454,其中13社達到變動規模報酬的效率邊界,純粹技術效率等於1,14社則遠離邊界,純粹技術效率低於0.2,表示多數儲互社都可在貸放員水準不變下,減少利息支出和其他營業支出以提高效率;規模效率的平均值為0.783,其中107社位於遞增規模報酬、15社位於固定規模報酬、202社位於遞減規模報酬。此外,以變異數分析做均值檢定,本文也發現儲互社的各項效率都與族群別無關;都市社的總技術效率和純粹技術效率都高於鄉村社,規模效率則沒有差異;大社的純粹技術效率高於小社,但規模效率低於小社,因此總技術效率和規模別無關。雖然擴大規模可以提高純粹技術效率,但也因離開固定規模報酬的效率邊界愈遠而降低規模效率,二者互相抵銷,所以擴大規模似乎無助於儲互社總技術效率的提升。以產出導向計算儲互社的總技術效率和投入導向相同,純粹技術效率平均值較高,而規模效率平均值較低,但二者之間的Spearman等級相關係數仍分別高達0.933和0.783,表示無論是純粹技術效率或規模效率,以投入導向和產出導向計算出來的效率值之排列順序並沒有太大的改變。都市社和鄉村社、原住民社和非原住民社、大社和小社的各項效率值之檢定結果都與投入導向模型一致。
Based on the data from the 1999 annual report, using loans to members as the output variable and interest expenses and other operating expenses as the Input variables in the context of the input-oriented model, we find 13 societies were located on the variable-returns-to-scale efficiency frontier while 14 societies among those that were not on the frontier had pure technical efficiency even lower than 0.2. This implies that most societies could have performed better by lowering the inputs while keeping the output unchanged. In the analysis of variance (ANOYA), we also find that the overall and pure efficiency measures were the same for aboriginal and nonaboriginal societies, urban societies performed better than rural ones in the overall technical efficiency and pure technical efficiency, and large societies demonstrated higher pure technical efficiency than small ones while the latter had higher scale efficiency than larger ones. This implies that, though an increase in scale could have raised pure technical efficiency, overall technical efficiency was unaffected because the lower scale efficiency could have offset this advantage. The same results are obtained if we use the output-oriented model instead.