Research Commentary—Information Technology Substitution Revisited

Taking advantage of the opportunities created by the price adjusted performance improvement in information technology (IT) depends in part on the ability of IT capital to substitute for other inputs in production. Studies in the information systems literature and most economics training examining the substitution of IT capital for other inputs use the Allen elasticity of substitution (AES). We present a less-well-known measure for the elasticity of substitution, the Morishima elasticity of substitution (MES). In contrast to the AES, which is misleading when there are three or more inputs—such as non-IT capital, labor, and IT capital—the MES provides a substitution measure where the scale is meaningful, and the measure differs depending on which price is changing. This is particularly important for IT capital, because prices have been declining, and there is evidence that IT capital can substitute for non-IT capital or labor in a qualitatively different way than non-IT capital and labor substitute for each other. Methodologically, we also show the impact of imposing local regularity—for example, monotonicity of output from increases in inputs—which we do through Bayesian methods employed to estimate the underlying functions that are used to calculate various measures of substitution. We demonstrate the importance of the MES as an underrecognized measure of substitution and the impact of imposing local regularity using an economy-wide industry-level data set covering 1998–2009 at the three-digit North American Industry Classification System code level. Our MES results show that reductions in the price of IT capital increase the quantity of IT capital in use, but are unlikely to change the input share of IT capital—the value of IT capital as a proportion of the value of all inputs—in contrast to major studies using the AES. In addition, estimates for both elasticities of substitution are more stable after imposing local regularity. Both of these advances—that is, the MES and imposing local regularity—have the potential to impact future work on IT productivity, IT pricing, IT cost estimation, and any type of analysis that posits the substitution of IT capital for non-IT capital or labor.