Evaluating the information content of a measure of plant output: An application to high-technology manufacturing

Commonly used measures of plant output have been criticized for their inability to provide information required to manage the dynamic operations of high-technology manufacturing plants. In this paper, we propose tests to evaluate the information content of a measure of plant output that is specifically directed at these issues. These tests are based on recent developments in Data Envelopment Analysis (DEA), namely the Cone Ratio Envelopments. In this new application of DEA models, we shift the focus from Decision Making Units (DMUs) being evaluated to the DMUs that are being used to effect evaluations. The proposed tests are applied to evaluate the information content of a complexity adjusted measure of plant output, which we refer to as Total Net Die Equivalent (TNDE). Developed recently in the context of a high-technology manufacturing plant - a wafer fabrication plant of a merchant semiconductor manufacturing company - TNDE reflects the ongoing changes in product and process technologies, process flow characteristics, and volume of production. Evaluating the information content enjoint criteria of "recency" and "efficiency", the results of our tests, conducted over a 28-month period in the wafer fabrication plant, show that TNDE as a single aggregate (scalar) measure of plant output outperforms the two outputs from which it is synthesized. Thus, TNDE as a single measure of output can be used to provide an improved basis for planning a plant's future operations. En route to the development and application of the proposed tests, we illustrate how DEA concepts and models provide a rigorous and systematic basis for conducting ex post technology evaluation to guide continuous improvements in a plant's current operations.