Data Envelopment Analysis (DEA)

Data envelopment analysis (DEA) is a linear programming method that measures the relative efficiency of decision-making units, such as hospitals or branches, that convert multiple inputs into multiple outputs.

자료포락분석(DEA)은 병원이나 지점처럼 여러 투입물을 여러 산출물로 전환하는 의사결정단위들의 상대적 효율성을 측정하는 선형계획법 기반 방법이다.

How do I compare the efficiency of organizations that use several inputs to produce several outputs?

여러 투입물로 여러 산출물을 만드는 조직들의 효율성을 어떻게 비교하나요?

Data envelopment analysis builds an efficiency frontier from the best-performing units and scores each unit by its distance from that frontier, without assuming a functional form or input prices. It fits comparing comparable units with multiple inputs and outputs, using constant returns to scale (CCR) or variable returns to scale (BCC) models.

When to use Data Envelopment Analysis (DEA)

• Comparing hospitals, schools, bank branches or regions on efficiency
• Multiple inputs (staff, budget) and multiple outputs (patients, graduates)
• Identifying efficient peers and improvement targets for inefficient units

When to use something else

• You want to separate inefficiency from random noise: use stochastic frontier analysis
• You want to explain efficiency scores with external factors: follow DEA with Simar–Wilson truncated regression with bootstrap (or fractional/OLS regression) rather than a naive Tobit model

Assumptions and requirements

• Homogeneous units performing similar activities
• Non-negative inputs and outputs (positive in standard models)
• Enough units relative to the number of inputs and outputs
• A justified orientation (input or output) and returns-to-scale assumption

What the analysis reports

• Efficiency scores for each unit
• Efficient frontier and reference peers
• Input and output slacks and targets
• Scale efficiency (CCR versus BCC)
• Malmquist index for panel data

How to report it

Under the output-oriented BCC model, [n] of [total] units were fully efficient, and the mean technical efficiency was [value] (SD = [value]).

Related methods

• Stochastic Frontier Analysis
• Malmquist Productivity Index
• Tobit Regression

Run Data Envelopment Analysis (DEA) on Quorant