Tools › Operations Research › Decision Analysis › Decision Analysis › Example

Heizer, Render and Munson: Thompson Lumber plant-size decision

ToolDecision Analysis
SourceRender, B., Heizer, J., and Munson, C. (2017). Operations Management: Sustainability and Supply Chain Management, 12th ed. Pearson, Boston. Module A, Decision-Making Tools, the Thompson Lumber Company example.
LevelUndergraduate

Problem

Thompson Lumber can build a large plant (200,000 if the market is favourable, −180,000 if not), a small plant (100,000 / −20,000) or do nothing. With no information on the market, which plant each criterion recommends; with equal priors, what is the expected value of perfect information; and with a market survey costing 10,000 (positive reading with probability 0.70 if favourable and 0.20 if unfavourable), what is the survey worth?

Inputs to Enter

  • Payoff table: Favorable market, Unfavorable market / Construct large plant 200,000 -180,000 / Construct small plant 100,000 -20,000 / Do nothing 0 0 (no prior line = equal priors)
  • Likelihood table: Positive survey, Negative survey / Favorable market 0.70 0.30 / Unfavorable market 0.20 0.80
  • Test cost 10,000; Hurwicz α 0.8
Open the Decision Analysis tool

Answers

  • Maximax large plant (200,000), maximin do nothing (0), equally likely small plant (40,000), criterion of realism α = 0.8 large plant (124,000), minimax regret small plant (100,000)
  • EMV: small plant 40,000 (best); EV with perfect information 100,000, EVPI = 60,000
  • P(positive) = 0.45; P(favourable | positive) = 0.78, P(favourable | negative) = 0.27 (book, rounded)
  • Positive → large plant, negative → small plant; EVSI = 19,000 exact (book 19,200 with the rounded posteriors), net 9,000 → survey
  • Efficiency 19,000/60,000 = 32 %

Notes

Type the two tables in place of the Goferbroke defaults; a payoff table without a probability line is read with equal priors. The book's EVSI of 19,200 comes from posteriors rounded to two decimals; the tool's exact value is 19,000. Edition and author order of the citation are not stated in the test script.

More Examples for This Tool