The Impact of Product Variety on Automobile Assembly Operations: Empirical Evidence and Simulation Analysis

Marshall L. Fisher, Christopher D. Ittner

Management Science · 1999 · 414 citations · 17 references

DOIFull text

Open access

Concepts

TL;DR

The study investigates how product variety affects automobile assembly plant performance, quantifying production losses, mechanisms, and the influence of option bundling and staffing policies. The authors analyze data from GM's Wilmington plant and conduct simulation studies of a generic auto assembly line to model the effects of product variety. Empirical and simulation analyses reveal that day‑to‑day option variability raises labor, overhead, downtime, repairs, and inventory, but with optimal buffering its effect on direct labor is negligible, and bundling options can reduce required buffer capacity while random variation remains the larger productivity threat.

Abstract

This study examines the impact of product variety on automobile assembly plant performance using data from GM's Wilmington, Delaware plant, together with simulation analyses of a more general auto assembly line. We extend prior product variety studies by providing evidence on the magnitude of variety-related production losses, the mechanisms through which variety impacts performance, and the effects of option bundling and labor staffing policies on the costs of product variety. The empirical analyses indicate that greater day-to-day variability in option content (but not mean option content per car) has a significant adverse impact on total labor hours per car produced, overhead hours per car produced, assembly line downtime, minor repair and major rework, and inventory levels, but does not have a significant short-run impact on total direct labor hours. However, workstations with higher variability in option content have greater slack direct labor resources to buffer against process time variation, introducing an additional cost of product variety. The simulation results support these findings in that once each workstation is optimally buffered against process time variation, product variety has an insignificant impact on direct assembly labor. The simulations also show that bundling options can reduce the amount of buffer capacity required, and that random variation is more pernicious to productivity than product variety, supporting the efforts of some auto makers to aggressively attack the causes of random variation.

References

17