As budgets are increasingly constrained, the technical efficiency of the organizations that produce public goods and services is vital to policy makers. Organizational and public policy scholars have measured the efficiency of organizations with several different tools, but one of the most popular is Data Envelopment Analysis (DEA). DEA has many advantages as a tool for efficiency measurement, but a major underlying assumption in DEA often not considered by policy scholars using DEA. Specifically, DEA requires that all inputs and outputs be substitutes. Failing to test and adjust DEA for the failure of this assumption can adversely impact the precision of DEA scores, and thus the policy resulting from them.

                DEA was developed by Charnes, Cooper, and Rhodes (1978) as a mathematical linear program designed to measure organizational efficiency non-parametrically. DEA combines multiple weighted input/output ratios to compute an efficiency score. The organizations’ scores are enveloped by their most efficient peers, and scores computed as a percentage of the most efficient organizations. DEA has many advantages as a tool for efficiency measurement, particularly over parametric methods, like regressions. Regressions are constrained by a singular input (output) measure for use as a dependent variable. DEA can also combine multiple input and output ratios such that a single efficiency score is calculated.

                As a result, it is often used by policy researchers to measure the efficiency of organizations in every public sector. A foundational component of the economic theory underlying DEA, however, is that it requires that inputs and output are substitutable. When this assumption is violated, the precision of resulting efficiency sores can be compromised. Using data from a sample of 77 public high schools in Chicago, this paper shows that imprecise organizational efficiency measurements can be obtained with DEA when inputs and outputs are not substitutable. This paper then gives steps to test and adjust DEA models such that precise efficiency scores can be obtained.              

                The results highlight implications for public policy. DEA is often used to highlight the most efficient organizations so that further inquiry can highlight best practices. This is particularly the case in educational policy. If efficiency is measured imprecisely, policy makers run the risk of identifying inefficient organizations as efficient. This can lead to inaccurate identification of organizational factors leading to best practice and best policy, and serve as a suboptimal basis for public and organizational policy.