Measuring Spillovers From Information Technology Investments

Intra-industry spillovers from information technology investments have been cited as a potentially important driver of productivity growth. Using firm-level data to measure the sizes of these spillovers, however, can be challenging because of biases caused by 1) measurement error and 2) the difficulty in separating the effects of spillovers from the effects of shared technological opportunity. In this analysis, we employ two separate approaches to develop more accurate estimates of the contributions of information technology driven spillovers to economic output. First, we use an instrumental variables approach to correct biases caused by measurement errors in IT capital, and second, we use technological variation from establishment level data to create richer models of knowledge spillover pools that are less vulnerable to the econometric problems identified above. We report estimates from both of these approaches and compare them with estimates from methods relying on conventional firm-level models of spillover. Our results suggest that existing estimates of within-industry spillovers from information technology investments may be considerably overstated. The estimates produced by our different approaches are internally consistent, and suggest that the contribution of within-industry information technology spillovers may actually be smaller than previously reported.