3.3.1Returns to Information Technology

In the basic framework described in section 3.1, authors often measured the return to IT with r COMP. It can be interpreted as the output elasticity or marginal product of IT capital. If this return is significantly greater than zero, then IT is a productive resource. Findings in the previous literature vary from – 20% to +68%.

Morrison and Berndt (1991) reported a surprising negative value for the return to IT. They studied twenty two U.S. 2-digit manufacturing industries over the 1952-1986 period. Their data on information technology capital consist of the Information Processing Equipment (IPE) section of nonresidential types of equipment from the Bureau of Economic Analysis (BEA). A generalized Leontief variable cost function with non-constant returns to scale is estimated by three-stage least squares. The variable of interest is the shadow value of IT capital, revealing its marginal efficiency. This return varies across industries but is estimated on average at - 0.20%. Hence, the marginal costs of investment in this type of equipment are greater than the marginal benefits, and $1 invested in IT capital returned on average $0.80. At a disaggregated level, a study from Brynjolfsson and Hitt (1993) showed more optimistic results. Their study is based on data from Compustat and InformationWeek at the firm level for 367 business units. Using a production function framework, they estimated econometrically the output elasticity of IT capital between 54% and 68%.

Apart from estimating the value of the return to IT investment, some authors have tested whether this return was greater than that of investment in traditional equipment. Berndt and Morrison (1995) studied twenty manufacturing industries with data on investment and capital stock from BEA. They found that the returns from computer investment are not significantly different from that of other types of capital. On the other hand, a firm level analysis of Lichtenberg (1995) found that IT capital earns positive and significant returns also significantly greater than the return to traditional capital. However, the author argued that using capital stock instead of capital services overestimates returns. Moreover, the production function framework used in this study distinguishes IT and non-IT workers. Lichtenberg found that one IT worker is six times as productive as a non-IT worker. Hence, there is some evidence of excess returns to IT employment too.

Lehr and Lichtenberg (1999) built an original framework for measuring excess return to IT capital. They used firm-level computer asset and financial data for non-agricultural firms during the period 1977-1993. Their model is based on a Cobb-Douglas production function where total capital (K) is divided into computer capital (KIT) and non-computer capital (KNIT). Technical progress is embodied as follows:

Y = A [KNIT + (1- θ) KIT]α. L (1- α )(3.11)

where α measures the elasticity of output with respect to the effective capital stock, and θ is a parameter that measures the “excess productivity” of computer capital relative to non-computer capital. Re-arranging and taking logarithms leads to (details in section 4.1.1)

ln(Y) = ln(A) + α ln(K) + αθ IT% + (1- α) ln(L) (3.12)

where IT% represents the IT ratio defined as the ratio of IT capital to total capital.

Estimating 3.12, the authors found that not only did computers contribute to productivity growth but they also yielded excess returns relative to other types of capital.

Gera, Gu and Lee (1999) performed regression analysis on a pooled cross-section time-series data set consisting of 27 industries in the United States and in Canada during five sub-periods between 1971 and 1993. They regressed the annual average labor productivity growth rate of a given industry on its IT and non-IT investment rates and other variables. Their results indicate IT investments are an important source of labor productivity growth across industries. Brynjolfsson and Hitt (1995) showed that the size of the productivity impacts is similar for manufacturing and service firms.

Quinn and Baily (1994) noted that national accounts data are extremely misleading and contain major gaps. Thus, they shifted the focus away from quantifying productivity benefits of IT capital toward measuring qualitative strategic performance improvements. The authors interviewed over 100 executives in top management, finance, information, and operating positions in the service industry. Their results indicate 80% of the companies surveyed had achieved adequate to high returns on their IT investments. Because of the difficulty of measuring output in the service sector, these gains were mainly qualitative such as greater flexibility and adaptability, improved responsiveness to new product lines, enhanced quality of work life and increased predictability of operations. More specifically, David, Grabski and Kasavana (1996) estimated the gains from to the use of IT in the hotel industry. They surveyed 100 large hotel companies, and measured qualitative gains related to IT investments. They argued “the productivity paradox may be less a paradox than a conscious strategy to select improvements in guest service over increase in productivity.” Finally, Reardon, Hasty and Coe (1996) surveyed 871 retailers and found IT capital has a positive effect on the output of retail institutions. According to the authors, retailers gained relatively more output per dollar’s worth of IT input than the marginal return on other types of capital.