Proactive data envelopment analysis: Effective production and capacity expansion in stochastic environments

Chia Yen Lee, Andrew L. Johnson*

*Corresponding author for this work

Research output: Contribution to journalArticleScientificpeer-review

28 Citations (Scopus)


Demand fluctuations that cause variations in output levels will affect a firm's technical inefficiency. To assess this demand effect, a demand-truncated production function is developed and an "effectiveness" measure is proposed. Often a firm can adjust some input resources influencing the output level in an attempt to match demand. We propose a short-run capacity planning method, termed proactive data envelopment analysis, which quantifies the effectiveness of a firm's production system under demand uncertainty. Using a stochastic programming DEA approach, we improve upon short-run capacity expansion planning models by accounting for the decreasing marginal benefit of inputs and estimating the expected value of effectiveness, given demand. The law of diminishing marginal returns is an important property of production function; however, constant marginal productivity is usually assumed for capacity expansion problems resulting in biased capacity estimates. Applying the proposed model in an empirical study of convenience stores in Japan demonstrates the actionable advice the model provides about the levels of variable inputs in uncertain demand environments. We conclude that the method is most suitable for characterizing production systems with perishable goods or service systems that cannot store inventories.

Original languageEnglish
Pages (from-to)537-548
Number of pages12
JournalEuropean Journal of Operational Research
Issue number3
Publication statusPublished - 1 Feb 2014
MoE publication typeA1 Journal article-refereed


  • Data envelopment analysis
  • Demand uncertainty
  • Marginal product
  • Short-run capacity expansion
  • Stochastic programming


Dive into the research topics of 'Proactive data envelopment analysis: Effective production and capacity expansion in stochastic environments'. Together they form a unique fingerprint.

Cite this