外文翻譯--基于dea方法的國家r&d項目多重目標績效評價比較

1、<p><b>　　中文2287字</b></p><p>　　Comparative evaluation of performance of national R&D programs with heterogeneous objectives: A DEA approach</p><p>　　Hakyeon Lee a, Yongtae P

2、ark a,*, Hoogon Choi </p><p><b>　　Abstract</b></p><p>　　The strategic importance of performance evaluation of national R&D programs is highlighted as the resource allocation draw

3、s more attention in R&D policy agenda. Due to the heterogeneity of national R&D programs’ objectives, however, it is intractably difficult to relatively evaluate multiple programs and, consequently, few studies h

4、ave been conducted on the performance comparison of the R&D programs. This study measures and compares the performance of national R&D programs using data envelopment</p><p>　　1. Introduction</p&g

5、t;<p>　　As R&D has been considered as a driving force for national competitive advantage, many countries have been raising R&D investments through various national R&D programs (Lee et al., 1996). Sinc

6、e R&D investment is one of the most decisive elements in promoting scientific and technological progress (Wang and Huang, 2007), the effective use of the limited R&D resources can be regarded as a prerequisite fo

7、r benefiting from formulation and implementation of national R&D programs. Thus, performance ev</p><p>　　Although a number of studies have been conducted to measure R&D performance at various levels,

8、 few attempts have been made at the national program-level. This is due to the heterogeneity of R&D programs in terms of policy purpose. Since each R&D program has its own primary objective such as publishing aca

9、demic papers for basic research, issuing patents and developing prototypes for applied research, and providing funds with researchers for R&D human resource development, it is intractably diffic</p><p>　

10、　Two conventional approaches to assessing R&D performance, peer review and bibliometric method do not work well for the relative evaluation of heterogeneous R&D programs. The peer review method, which is based on

11、 perceptions of well-informed experts about various quality dimensions of R&D, is inherently subjective and likely to be biased depending on interests, experience, and knowledge of the evaluators (Nederhof and van Ra

12、an, 1987; Brinnet al., 1996). The bibliometric method is considered relat</p><p>　　The tenet of this paper is data envelopment analysis (DEA) can overcome these limitations. DEA is a linear programming model

13、 for measuring the relative efficiency of decision making units (DMUs) with multiple inputs and outputs (Cooper et al., 2000). Since it can not only handle multiple outputs, but also allow each DMU to choose the optimal

14、weights of inputs and outputs which maximize its efficiency (Cherchye et al., 2007), it is capable of mirroring R&D programs’ unique characteristics by assi</p><p>　　DEA is a non-parametric approach that

15、 does not require any assumptions about the functional form of a production function and a priori information on importance of inputs and outputs. The relative efficiency of a DMU is measured by estimating the ratio of w

16、eighted outputs to weighted inputs and comparing it with other DMUs. DEA allows each DMU to choose the weights of inputs and outputs which maximize its efficiency. The DMUs that achieve 100% efficiency are considered ef

17、ficient while the other</p><p>　　The first DEA model proposed by Charnes et al. (1978) is the CCR model that assumes that production exhibits constant returns to scale. Banker et al. (1984) extended it to th

18、e BCC model for the case of variable returns to scale. When it comes to R&D returns to scale, findings from previous studies are somewhat mixed (Graves and Langowitz, 1996). It was found that R&D activity can exh

19、ibit increasing or decreasing returns to scale as well as constant returns to scale (Bound et al., 1984; Scherer, </p><p>　　2. Conclusions</p><p>　　We measured and compared the performance of th

20、e six national R&D programs with heterogeneous objectives using DEA. Every project in every program was evaluated together, and Kruskal–Wallis test with a post hoc Mann–Whitney U test was then run to compare performa

21、nce of R&D programs. The two alternative approaches to incorporating the importance of variables in reality, the AR model and output integration, were also considered. Due to the heterogeneity of national R&D pro

22、grams’ objectives, few stu</p><p>　　The DEA results are expected to provide practical implications for policy making on national R&D programs. The limited resources can be effectively allocated to severa

23、l R&D programs based on their performance rankings. R&D programs doing well (e.g. Program C and D) deserve more investments; on the other hand, poor programs (e.g. Program A and F) have to be terminated or funds

24、given to them should be cut down unless their performance is improved. Basically, DEA offers the way of improving efficie</p><p>　　To seek the way of enhancing performance, the reasons for poor performance s

25、hould be uncovered by examining the context in which poor programs are formulated and implemented, such as project selection procedure, operational regulation, funding systems, etc. It is obvious that the prerequisite fo

26、r this is to be able to measure and compare the performance of various R&D programs, which is the primary contribution of this study.</p><p>　　Nevertheless, this study is subject to some limitations. Fir

27、stly, since the projects that have not been finished at the time of data collection were not included, program performance was measured without them. Secondly, despite the fact that it takes several years for R&D out

28、puts to be achieved, the outputs produced only for two years after termination of projects were considered. These limitations will be overcome if the analysis is conducted again at some time in future. Thirdly, it may oc

29、cur </p><p>　　基于DEA方法的國家R&D項目多重目標績效評價比較</p><p><b>　　摘要</b></p><p>　　國家R＆D項目績效評價由于它的戰(zhàn)略重要性被高度關注，其資源分配的施政綱領吸引了更多人的注意。但是由于國家的R＆D項目'的目標的多重性，對多個項目進行評估是一件相當棘手的事，因此，很少有研

30、究對R&D項目進行績效比較。本研究利用數(shù)據(jù)包絡（DEA）衡量和比較了國家的R＆D項目績效。由于DEA允許每個決策單元的DEA選擇投入與產出的效率，最大限度地發(fā)揮其最佳的權重，通過分配高權重給各個有優(yōu)勢的規(guī)劃變量是鏡像研發(fā)項目的獨特特征。R&D項目中的每一個計劃都在DEA模型里被同時評價，然后再在其他的不同的系統(tǒng)中進行效率比較。Kruskal–Wallis和Mann- Whitney U的檢驗測試運行比較了R&D項

31、目的性能。對兩個可選擇的方法納入變量的重要性-----AR模型和輸出一體化-------也進行了介紹。該結果有望對國家研發(fā)計劃的有效制定和實施提供政策支持。</p><p>　　一．DEA方法的介紹</p><p>　　隨著R&D是一國增強競爭優(yōu)勢的驅動力的觀點的普及和被認可，許多國家開始通過各種各樣的R&D項目，對R&D增加投資。R&D投資是推動科技進步的

32、決定性要素之一，對R&D資源的有效使用被認為是從國家R&D項目的制定和執(zhí)行中獲益的前提。因此，對R&D項目進行績效評價是非常有必要的，它能使有限的R&D資源得到優(yōu)化配置，使項目實現(xiàn)優(yōu)勝劣汰。</p><p>　　盡管許多研究已對項目的績效進行了不同程度的衡量評估，但是很少有從國家的項目層次來研究過。這是因為基于國家政策目標的R&D項目有其獨有的復雜性。</p>

33、<p>　　由于每個項目都有其自己的首要目標，比如基礎研究項目的首要目標是發(fā)表學術性研究論文，應用型研究的目標是發(fā)明專利和開發(fā)模型，人類資源發(fā)展研究項目的目標則是為研究人員提供資金支持。因此在同樣的時間同樣的背景下對不同類型的研發(fā)項目進行比較和績效評價是相當困難的。</p><p>　　同行評議法和計量法是評估R&D項目的兩大傳統(tǒng)方法，但是他們對復雜多樣的R&D項目的績效評估并不能取得

34、很好的效果.同行評審的方法，基于對R&D項目各種質量方面見多識廣的專家觀點得出的，本質上是主觀的，會受評價者知識水平，經驗和利益偏向性的影響，而計量學方法被視為是比較客觀的方法，但結果高度依賴于測量方法。</p><p>　　本文的宗旨是利用數(shù)據(jù)包絡分析（DEA）來克服這些限制。DEA是一個用于處理具有多個輸入和多個輸出決策單元的多目標決策問題的線性規(guī)劃模型。由于它不僅能處理多個輸出，也能讓各個決策單元（

35、DMU）來選擇輸入和輸出的最優(yōu)權重，多個輸入（輸出越小越好）和多個輸出（輸入越大越好）。通過分配高權重給各個有優(yōu)勢的規(guī)劃變量是鏡像研發(fā)項目的獨特特征，這項研究衡量和比較了六個國家在韓國的R&D項目上使用的DEA效率性能。</p><p>　　DEA是一種非參數(shù)方法，不需要對生產函數(shù)的函數(shù)形式和投入與產出的重要性先驗信息做出任何假設。一個決策單元的相對效率是通過估算加權投入產出比來衡量，并與其他決策單元做出

36、比較， DEA允許每個決策單元的選擇使其投入與產出比率最大化的權重。決策單元組達到100％的效率被認為是有效的，而低于100％的效率則被認為是無效的。</p><p>　　第一個DEA模型是由Charnes等人于1978年提出的CCR模型，假定生產規(guī)模收益不變。1984年，Banker等人將它延伸到變量返回的規(guī)模下的BBC模型。當涉及到研發(fā)規(guī)模收益時，從以往的研究結果來看好壞參半（Graves and Lango

37、witz，1996）。結果發(fā)現(xiàn)，R＆D活動可以顯著增加或減少規(guī)模報酬，以及規(guī)模報酬不變(Bound et al., 1984; Scherer, 1983).,，因此，本研究便運用了BCC模型。著名的DEA模型也是一個模型的目標：最大限度地提高輸出（輸出型）或盡量減少投入（投入導向）。這是含蓄地假定R＆D的目標是基于增加產出，而不是減少投入。因此，本研究采用產出導向模式。</p><p><b>　　二

38、．結論</b></p><p>　　我們利用DEA對六個國家的R&D項目的多目標績效進行了比較。每個項目中的每個計劃進行了同時評估，Kruskal–Wallis和Mann- Whitney U的檢驗測試運行比較了R&D項目的性能。對兩個可選擇的方法納入變量的重要性-----AR模型和輸出一體化-------我們也同樣考慮在內。由于國家的R＆D項目'的目標的多重性，很少有研究對R

39、&D項目進行績效比較。本研究有助于填補該領域運用DEA對國家R&D項目績效評價的空白。 DEA方法，特別是不同系統(tǒng)之間的效率比較模型，被證明在多重目標的R&D項目的績效比較方面是比較有效的。</p><p>　　在DEA結果被期望為國家R&D項目決策的制定提供的實際的應用。有限的資源能夠依據(jù)幾個項目的績效排名來有效地進行分配。做得好的R&D項目（例如項目C和D）值得更多的投

40、資，另一方面，績效差的項目（如項目A和F）應該被終止或應減少給予的資金，除非其績效能夠得到提高。基本上，DEA方法提供了單位個體間優(yōu)勝劣汰的方法，盡管它沒有在本文中提出明確處理這個問題的方式。一系列高效的項目為每個低效項目提供了一套參考基準，它又反過來導致各個項目績效的提升。然而，DEA告訴我們的是提高效率的方法是有多少產出應增加以達到100％的效率，而不是在當前設置增加實際產出的方法。</p><p>　　為了

41、尋求提高績效方式，表現(xiàn)不佳的項目應通過檢查低效方案的制定和實施，如項目的遴選程序，運行規(guī)則，資金系統(tǒng)等發(fā)現(xiàn)不足。各種各樣的R&D項目的績效是通過先決條件來衡量和比較，這是顯而易見的，而且這也算本研究的主要貢獻。</p><p>　　不過，這項研究受到一些限制。首先，由于在數(shù)據(jù)收集時未完成的項目不包括在內，項目績效考核時并不包括他們。其次，盡管事實上R＆D的產出需要數(shù)年來實現(xiàn)，但是產出的程序在只在終止后兩年

42、進行審議。假如在未來的某個時間這些分析能夠再次進行，這些限制將被克服。第三，一個研發(fā)計劃被認為是高的表演者，即使他們沒有達到自己的目標，但在另一個領域成就卓越的成果，這種情況也是有可能發(fā)生的。雖然它沒有在這項研究中發(fā)現(xiàn)，但是在這種情況下，判斷可能引起爭議。這些問題應在今后的研究中處理。未來研究的另一個途徑是采用擴展的DEA模型來比較不同的結果。另一種模式將帶領我們尋求一個評估和比較多重目標的國家R&D項目績效更好的辦法。<