A Research Forum for Technology Commercialization and Economic Logistics

Rating the financial performance of high tech companies.  Data envelopment analysis (DEA) is coming into general use as a tool to measure the relative cost effectiveness of industrial corporations. We discuss below how DEA may be further extended to measure the nature and extent of the high tech revolution and to investigate the stock "fundamentals" of high tech corporations each year, and to rank the companies in relation to each other.

A series of studies are described, analyzing the efficiency of all publicly held US computer companies during the course of a decade, tracking the position of each company relative to the moving efficiency frontier, and its economies of scale. In the common fashion, DEA enables the analyst to gauge the performance of each company relative to the "leading edge" in the industry.

A few companies, like Apple and Compaq, manufacturing products with long and sustained life cycles, were consistently located at the efficiency frontier. But most companies, spending heavily to bring on line a stream of innovative products, were inefficient. The results indicate that the computer industry was polarized into two camps: large aging corporations with decreasing returns to scale, and swarms of small upstart companies with advanced technology exhibiting increasing returns to scale. 

Several of the papers listed below are available for free downloading. The three first papers (1994, 1996a and 1996b) report on the DEA calculations in some detail. The next paper (by Abad, Thore and Laffarga) report on a new two-stage DEA procedure. The two last papers are mainly devoted to interpretation.

For a brief introduction, see S. Thore, "Cost Effectiveness and Competitiveness in the Computer Industry: A New Metric", Technology Knowledge Activities, Vol 1, No.2, Fall 1993, pp. 1 - 10 which can be downloaded by clicking here.

Abstract.  DEA (data envelopment analysis) is a technique for determining the efficiency "frontier" (the envelope) to the inputs and outputs of a collection of individual corporations or other productive units. DEA is here employed to estimate the intertemporal productive efficiency of U.S. computer manufacturers, using financial data brought from earnings statements and balance sheets. The results indicate that a few corporations, including Apple Computer Inc, Compaq Computer Corp, and Seagate Technology were able to stay at the productivity efficiency frontier throughout the time period investigated. But not all "successful" corporations did; sometimes subefficiency (= disequilibrium) actually goes together with very rapid growth. A new Malmquist type productivity index is calculated for each corporation, measuring shifts of the estimated intertemporal efficiency frontier.

To download a pdf version of this paper, click here

From the Statement of Scope and Purpose:  

"Whereas the financial results of a corporation are easy to measure, its cost effectiveness or - - using the more technical term preferred in the subsequent - - its efficiency has only recently become amenable to routine calculation, using the method of so called Data Envelopment Analysis (DEA). We present such calculations for the U.S. computer industry during the decade 1981-90. In order to interpret the results, we discuss the efficiency ratings of a computer company during the course of a typical product cycle. Both theoretical considerations and empirical observations support the proposition that the cost efficiency of a company will tend to be low during the introduction and the upswing of the product cycle, but gradually rise as the product reaches maturation and eventual saturation (an efficiency cycle). The overriding aim of high technology firms is to generate a sustained flow of successful new market introductions. Cost efficiency is certainly desirable, but many companies try to steer a middle course, sacrificing efficiency in order to gain organizational slack, innovative potential, and growth."

Abstract. The method of Data Envelopment Analysis (DEA) is applied to rank the efficiency of US computer companies during a 10-yr period. To reflect the dynamic setting of the computer industry, the inputs include investment in real capital ande expenditures on R&D, the outputs are sales revenues, profits and market capitalization. We develop a procedure for studying the time path of the observed DEA ratings of a high tech company over its product cycles. The empirical observations confirm the key relationship between efficiency and the product cycle. Since computer companies differ greatly in their success in managing their product cycles, they will also show quite different efficiency results. A few companies, like Apple and Compaq, manufacturing products with long and sustained cycles, were consistently located at the efficiency frontier. But most companies, spending heavily to bring on line a stream of innovative products, were inefficient.

Abstract: Up to recently, economists have had no good tools to measure the returns to scale of individual corporations in an industry. Data envelopment analysis (DEA) is a linear programming technique for determining the efficiency frontier (the envelope) to the inputs and outputs of a collection of individual corporations or other productive units. While DEA offers an avenue for calculating the returns to scale of individual corporations, the approach has been riddled by mathematical complications arising from the possibility of alternate optima. The present paper develops theory for calculating the entire range of these alternate optima. Furthermore, in a quite ambitious empirical application, DEA is employed to determine the time path of returns to scale of all publicly held U.S. computer companies over the time period 1980-1991. For the great majority of companies, a unique time path is obtained; only in less than 4 percent of the linear programming calculations is an entire range of alternate optima obtained. The results indicate that the computer industry was polarized into two camps: large aging corporations with decreasing returns to scale, and swarms of small upstart companies with advanced technology exhibiting increasing returns to scale.

From the Introduction:  

A characteristic feature of fundamental analysis is that it searches for an explanation of stock price and market value via an un-observed underlying causal factor: future earnings. Precisely because it is un-observed, fundamental analysis searches deeper, down to the financial fundamentals of the stock. The last step of fundamental analysis (associating future earnings with market value) therefore can never stand on its own. It needs the preceding first step as a prerequisite (associating standard financial indicators of the stock with future earnings). To represent this cascading causal mechanism mathematically, we propose a novel format of the so-called two-stage DEA. We construct two successive DEA frontiers fitted to the statistical observations, with revenues being an output variable of the first frontier, and an input variable into the second frontier. To be more precise: for each stock, the idealized and unobserved revenues calculated from the first frontier are fed as inputs into the second frontier.

To download a pdf version of this paper, click here.

The final two papers in this series explore further the nature of network externalities and economies of scale in the digital industry:

From the Introduction:

"For some time economists have realized that many of the laws of the market economy are being rewritten by a growing reliance on software and knowledge-based products. Toward the end of the 20^th century, a new kind of economic goods and services made its appearance in the marketplace which has the potential of revolutionizing the workings of the entire capitalist economy. These are the digital products - - communication, education and entertainment in digital form. The appearance of the digital economy represents the current high water mark of this development of a knowledge-based economy, with knowledge encoded in digital form."

"In the text to follow, I shall argue that a new kind of economic analysis is needed to study the knowledge-based economy. I shall sketch the outlines of a new economic paradigm that traces the rapid transformation of the market economy in the information age...

Whereas conventional economics assumes optimizing behavior, equilibrium and the stability of equilibrium, the dominant mode of operation of the new knowledge-based corporations is sub-optimal behavior and disequilibrium. To back up these claims, I shall report on an extensive empirical. study of the US computer industry."

From the Introduction: "This chapter argues that increasing returns to scale are common in the new digital industry. As new knowledge encoded in digital form is diffused throughout the economy, network externalities appear, driving many new digital firms onto a path of rapid growth - hypergrowth... Such firms enjoy increasing returns to scale during the upswing of their product life cycles.

There is a wealth of circumstantial evidence of hypergrowth and disequilibrium in the digital industry... In order to reconcile such observations with our theoretical understanding, we need to both develop a theoretical format of increasing returns to scale and undertake empirical measurements."