以資料包絡分析法評估地方環保機關污染防治績效分析之適用性

Abstract

本研究欲針對我國所有縣市之地方環保機關在污染防治業務方面，縣市之間的相對績效良窳排序，鑒於多投入多產出的同儕之間相對績效的分析情境，選擇以資料包絡分析法(Data envelopment analysis, DEA)，作為本研究之主要分析方法，藉之以分析2008年至2012年5個會計年度間所出版之《中華民國環境保護統計年報》中，所揭露記載2007年至2011年5年間，有關空氣污染、水污染、廢棄物管理、環境衛生與毒化物管理等污染防治業務相關數據，此前先輔以皮爾森相關分析法，先行過濾剃除DEA效率分析所需之投入項與產出項之間，非呈正相關之數據項目。\n\n 分析結果發現，不同的污染防治情境有不同的特色：空氣污染無論是在空污防治效率或是空氣品質兩者排名，都顯示出鄰近縣市之間，或有相互影響排名的可能性，而其中最具影響性之環境因素之一乃每(萬)人所分配之汽車輛數；水污染防治效率排名前段班的縣市，大多在人口、工廠、車輛等環境負荷項目數量或密度上，低於全國平均，但數據也顯示環境負荷量低的縣市不一定就有能在水污染防治效率排名上擁有優勢；廢棄物管理效率排名，北部縣市居中，中部與南部區域內排名呈現M型化；環衛毒化物管理的效率排序，只有離島地區有區域性排名不佳的問題；整合上述四種污染防治效率排序後，臺灣本島東半部名次優於西半部，西半部的中部地區及其鄰近縣市的污染防治效率有普遍性不佳的趨勢，但污染防治效率的排序前後與環境品質的排名的相關性相當微弱，因此在污染防治的整體排名上的軒輊，完全不意味著當地環境品質現況之良窳。\n\n 在選擇投入產出項目過程中，本研究根據唐先楠(1995)與黃旭男(1996、1999)的作法，將產出根據行政機關可控制程度，粗分為兩類，可控制程度較高者，本研究稱之為「行政產出」，反之稱為「行政效果」，「行政效果」與投入資源的關係，在污染防治方面的數據上，大多呈現顯著的負相關，因此無法列入一般DEA效率分析；在DEA效率中所計之投入資源，建議採「歲出決算」方面數據，方能符合DEA投入產出項之邏輯觀－投入影響產出，反之，有鑑於預算於行政機關之編列程序，不建議採計「預算」方面的數據，恐不符合投入影響產出之因果關係，影響DEA效率分析之品質與解釋力。

The aim of this paper is to rank the pollution prevention performance of the local environmental protection authorities in Taiwan, and ranking of the Data envelopment analysis, the solution for measure productive efficiency of decision making units (DMUs) with multi-input and multi-output. The Ranking and analysis database were 《Yearbook of Environmental Protection Statistics, Republic of China》, published by Environmental Protection Administration, Executive Yuan, published in 2008-2012, recording about pollution prevention and control business related data of air pollution, water pollution, solid waste management, environmental sanitation, toxic chemicals and so on during 2007-2011.\n\n After those analyses, these results are showing some trend in different sort pollution prevention performance. About air pollution, both of the ranking for efficiency of air pollution prevention and for air quality are showing that the ranking close between neighboring counties, and one of the most probable factors is that the number of automobile vehicles been allocated for per 10,000 people. About the water pollution efficiency ranking, most of those top class cities and counties are having lower than the national average volume or density in population, factories, vehicles and other environmental load. While integration of all 4 kinds rank of efficiency of pollution prevention, the score of Eastern Taiwan is better than Western Taiwan, and those most worst ranking concentrated in Central Western Taiwan, but the rank of efficiency of pollution prevention does not mean those cities and counties’ environmental quality are good or bad.\n\n In this paper, output is distinguished from the degree of control by administrative organ: if the degree is high, the output has been sorted as “Administrative Outputs”, on the contrary, the output has been sorted as “Administrative Effect”. Most of the relationship between “Administrative Effect” and input are in negative correlation obviously, so, if the output been sorted as “Administrative effect”, Most of them could not be included in the general DEA efficiency analysis. In the other hand, the input item about money, must be adopt the Final Annual Expenditure Accounts, not Budget, that make sure the causal relationship between input and output is consistent with the DEA’s assumption.