Robustness of the EOQ Model

By Praf Joglekar La Salle University, Philadelphia, PSA 19141 (joglekar@lasalle.edu) Most Operations Management textbooks point out that the EOQ model is rather insensitive to the parameter values. However, they often stop at showing something like when the annual demand of a product doubles, the EOQ rises only by approximately 41%. In what follows, I show the true magnitude of the insensitivity of the EOQ model. Consider a situation where D = Annual demand is 5,000 units per year S = Ordering cost per order is $100/order H = Holding cost per unit per year is $1.00/unit/year In this case, EOQ = Q = (2DS/H)1/2 = 1,000 units per order And the optimal total annual cost of ordering and carrying TC = (D/Q)S + (Q/2)H = 500 + 500 = $1,000/year. Now, suppose that we had misestimated the annual demand to be D = 6,000 units per year. Note that the % error in the estimation of the demand is = 100 [(6,000 - 5,000) / 5,000] = 20%. Here, we would have calculated our EOQ as EOQ = Q = [(2DS)/H]1/2 = [(2)(6000)(100) / (1)]1/2 = 1,095.44 units per order. Note that the % error in EOQ = 100 [(1,095.44 - 1,000) / 1,000] = 9.54%. However, as we implement our incorrect EOQ policy through the year, we will discover that the true demand is 5,000 units per year. Hence, the total cost we will experience must still be based on the true demand, D = 5,000. Thus, TC = (Q/2)H + (D/Q)S = (1,095.44/2)(1) + (5,000/1,095.44)(100) = 547.72 + 456.44 = $1,004.16 per year Hence, the % increase in TC = 100[(1,004.16 - 1,000) / 1,000] = 0.4%. That is, 4/10th of 1%. In short, a 20% error in D leads to a 9.5% error in Q and only a 4/10 th of 1% error in TC! Conclusion: The EOQ model’s robustness (or insensitivity to estimation errors) is truly dramatic! Note: You can similarly verify EOQ model’s robustness with respect to estimation errors in H and S.

Sign up to vote on this title
UsefulNot useful