Manufacturing Labor Cost Estimation With Uncertainty And Learning Effects

This study focused on building a costing system on a bus supplier company.As a starting point, the costing method was applied to the cost of which the companyhad more control, labor cost. Uncertainty and learning effects were the two maincharacteristics included in the proposed costing method. Gaussian distribution (foruncertainty) and Dejong learning model were implemented to describe the costingcharacteristics. The result showed that the existing labor cost estimation was higherthan the actual labor cost even with the inclusion of uncertainty. Besides, this modelprovided significant contributions to the studied company: cost position prediction fortender purpose, managerial focus on production line, long-term contract, andevaluation on the effects of managerial improvements. When the total labor costderived from the tender’s price was higher than the actual mean total labor cost, thenthe tender had a low risk and less managerial focus should be implemented in thatproduction line, and vice versa. When the total labor cost derived from the tender’sprice passed the left end of the uncertainty distribution graph, significantimprovements should be applied and this tender hided a high risk, and vice versa.The last two contributions were supported by the sensitivity analysis on thenumber of units produced, mean labor time learning constant, and labor time standarddeviation’s learning constant. The first sensitivity analysis suggested that at certainproduction level, the company could rely on learning curve for cost reduction, butabove that level, additional managerial improvement or sources (e.g. economy ofscale) should be applied. Both learning constants sensitivity analysis were intended todetermine the parameters needed to evaluate the performance of managerialimprovements (by a consulting firm). The result showed that both learning constantswere not good parameters to evaluate the improvements done by a consultingcompany because the cost saved by these learning constants were lower than themonthly price of a local consulting company’s service. By utilizing simple and cheapexperimental method, this model managed to provide the company’s managers quick and crucial information for cost reduction decision making, which further impacted tothe company’s survival.