تعیین معیار هدف هدایت شده برای بهره وری سازمانی

In this paper, we extend the standard data envelopment analysis (DEA) model to include longer term top management goals. This extension is in recognition of the fact that benchmarking for decision making units (DMUs) is more than a purely monitoring process, and includes a component of future planning. The new model uses a goal programming structure to find points on the efficient frontier which are realistically achievable by DMUs, but at the same time achieving a closer approach to long term organizational goals (as distinct from the local performance of individual DMUs). Consideration is given to the possibility of adjusting constraints on the DMU by investment in extended inputs or new technologies, in which case minimization of associated investment costs becomes an additional management objective.

Data envelopment analysis (DEA) has become a widely recognized tool for the evaluation of organizational efficiency, and a number of extensions and applications have been reported [e.g. [3], [15], [6] and [2]]. Results from DEA include evaluation of the efficiency of “decision making units” (DMUs) in the data base, a non-parametric estimate of the best practice frontier of the production possibility set (PPS) relating outputs produced to inputs consumed, and a consequent provision of benchmark performance levels on the efficient frontier of the PPS for the inefficient DMUs. A number of writers [e.g. [10], [16], [1] and [5]] have commented on the links between DEA and multiple criteria decision analysis (MCDA), while others in a similar sense have discussed means of incorporating judgemental management goals in DEA [e.g. [4] and [15]. Cooper [5] warned, however, that a superficial mathematical similarity between methods of DEA and MCDA should not obscure fundamental organizational differences between management monitoring and control on the one hand, and management planning on the other hand. The assessment of the historical efficiency of a DMU is part of monitoring and control, and the thrust of DEA is to be as objectively fair as possible in making such assessments. The concern of MCDA with values and goals is essentially prospective, and relates to planning, i.e. the process of moving from where we are to where we want to be. Nevertheless, one of the standard outputs of a DEA analysis is the establishment of “benchmarks” for each inefficient DMU, with the implication that these may serve as targets towards which the DMU should aspire. The point of departure for the present paper is that such targets move from pure monitoring and control to planning, and as such should include value judgements from group management as to what is desirable in addition to what is achievable technically. In fact, future targets can meaningfully be set even for efficient DMUs; they may be efficient in terms of their current inputs and outputs, but there may still be room for improvement in the sense of moving closer to overall management objectives. The theme of the present paper is thus to propose means of specifying benchmarks within a framework similar to that of DEA, but incorporating future management goals. In the next section we introduce some basic notation and review standard DEA models. The basic form of the proposed new model is described in Section 3, and is extended in Section 4 to include costs of new investments that may be needed. The approach is illustrated by a numerical example in Section 5. In a concluding section we indicate further extensions that could easily be incorporated.

We have introduced an enriched form of data envelopment analysis linked to future planning goals. The outputs still partition DMUs into efficient and inefficient units in the same was as before. However, the resulting benchmark performance levels are modified from the conventional DEA benchmarks such as obtained from the input- and output-oriented analyses. The benchmarks developed here then bridge the transition from monitoring and control (performance measurement in the conventional DEA sense), to management planning (the domain of application of multicriteria decision analysis). A number of extensions to the basic models proposed here can certainly be envisaged. • The production trade-off and weight restriction model of Podinovski [14] should easily be incorporated into the above models, in order to include additional planning options. The terms View the MathML source∑j=1nλjxij and View the MathML source∑j=1nλjyrj would need to be extended in (5) and (6) to View the MathML source∑j=1nλjxij+∑t=1Tπtpit and View the MathML source∑j=1nλjyrj+∑t=1Tπtqit, where for each potential new technology t, (p1t,…,pmt,q1t,…,qstp1t,…,pmt,q1t,…,qst) is a set of technologically achievable changes. The multipliers πtπt then represent the proportions of each new technology incorporated into the establishment of the benchmarks for the DMU under consideration. Now, as with the simple adjustment to inputs discussed in Section 4, introduction of new technologies would incur investment costs. If these could be deemed proportional to the πtπt, then the resultant costs could be added to the costs of modifying inputs, i.e. the ciξikciξik terms in (6). • Deviations from goals could possibly also form a basis for a goal-directed rank-ordering of DMUs (providing a method for generating a single rank ordering incorporating both efficient and inefficient units). These potential extensions will form the basis for further research.