| When called on to participate in evaluation of the effects of public sector programs, accountants must apply the standards of their profession to assess the usefulness of data sources and methods. Data Envelopment Analysis (DEA) has been used to evaluate the efficiency effects of programs. Before relying on its results, it is necessary to investigate its properties as an efficiency measurement method. Important measurement criteria such as reliability and validity cannot be directly assessed in real applications. Some evidence, however, may be obtained by examining solution consistency and comparability.; The interest of this research was in analytical and empirical evaluation of DEA group efficiency determinations. First, the DEA model and past literature were reviewed, to evaluate how group decisions can be made with DEA. Problems of amalgamating information across variables and firms were shown to complicate group efficiency comparisons.; DEA solutions in one empirical context (prospective hospital reimbursement) were subjected to consistency comparisons in several ways. First, two methods of group solutions were compared. The interenvelope method was shown to be less stable over a six-year period than a single-step approach. DEA solutions over the same period, using highly correlated alternative variable sets, were found to be different in many instances. For comparison, group judgements were made using standard methods (regression and ratio analysis). As expected, when patterns could be discerned in DEA solutions, they did not generally correspond to findings of those methods.; It was concluded that further work in DEA should improve group comparison methods, and demonstrate means to obtain consistent solutions. Defending any one efficiency conclusion for hospital reimbursement given DEA solutions developed here would be difficult indeed. |
