Quality Improvement and Learning in Productive Systems

Abstract

Recent interest in product quality suggests that effort devoted to improving the quality of manufactured products may reduce unit costs. This conjecture—that improving quality can lower costs—challenges the traditional assumption that unit costs increase with increased quality assurance activities and has significant implications for quality management. By introducing the idea of a quality-based learning curve, this paper links the previously disjoint literatures of quality control and learning curves to explain why high quality and low costs need not be inconsistent. When costs are affected by a quality-based learning curve, product quality favorably influences the rate of cost reduction due to learning. Thus, costs decline more rapidly with the experience of producing higher quality products. Two formulations of the quality-based learning phenomenon are presented. The first assumes that quality-based experience affects direct manufacturing costs. For this formulation, the optimal quality level is decreasing over time, but is always larger than the optimal base-case quality level. The optimal production quantity is constant if the interest rate is zero and increasing in time when the interest rate is positive. The second formulation assumes that quality-based experience affects quality control costs. In this case, the optimal quality level is always increasing over time. The optimal quantity behavior is qualitatively similar to the first formulation. A key feature of the second quality-based model is that it resolves the controversy between the economic conformance level model of Juran, which asserts that one should use cost-tradeoff analysis to find the optimal quality level, and the claims of Deming and Crosby, that zero defects is always the optimal quality level. For certain parameter values, the optimal quality policy in the second model conforms to the economic conformance level model but the dynamics of the model demonstrates the optimality of always pushing towards zero defects.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 321 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.