Decision Making Unit

A Decision Making Unit (DMU) is the unit of analysis in Data Envelopment Analysis (DEA) literature — a similar entity (one of N homogeneous organisations) being benchmarked, that consumes multiple inputs to produce multiple outputs. Typical examples: 500 bank branches of one bank; 200 public hospitals; 1,000 elementary schools in a province; 80 agricultural cooperatives; 81 provinces in a province-by-province efficiency comparison. The **homogeneity** condition is critical: same type of operation, comparable scale range, the same defined input-output structure; heterogeneous DMUs make the DEA frontier meaningless. The critical **sample-size heuristic**: N ≥ 3 × (number of inputs + number of outputs) — Cook-Seiford (2009) — otherwise frontier estimation is unreliable and almost every DMU automatically comes out efficient (statistically meaningless). Banker-Charnes-Cooper (1984) extended the DMU concept from CCR to BCC (acknowledging scale variation via variable returns to scale). Canonical sources: Banker-Charnes-Cooper (1984) + Cook-Seiford (2009).