The economic impact of information technology

Business Economics, Oct, 1995 by Robert B. Cohen

The economic impact of information technology has been a subject of a great deal of debate. For business economists, it is useful to identify how information technology (IT) is likely to impact the economy, because IT (defined as computer and communications technology and its applications) is likely to have a substantial impact on the economy's growth during the coming decades. The reason for this is the use of IT by nearly all industries in the economy's base, so that IT becomes a universal input to nearly all other outputs. If IT costs decline, they can create substantial economic gains for many of the industries that use IT, because money spent on IT can be invested in other inputs and improvements in production or services. Furthermore, because business relies upon IT to do a wide range of tasks and to create competitive advantage, by facilitating these tasks for end users, important gains are achieved that are difficult to measure in a classic input-output framework. In addition, IT, seen in a larger context, should have even wider impacts on the economy, because new channels of communications, such as the Internet, cellular television, and broadband applications, will provide business with new channels to reach customers and suppliers.

In the past, the economic impact of IT has been subject to much debate. The productivity paradox was first proposed by Steven Roach, the chief economist at Morgan Stanley, who found that BLS data on investments in computers had a clear negative rather than a positive impact on productivity gains in several major industries. Roach's paradox appeared to be valid because quite a few service industries had negative productivity gains between 1977 and 1984.

Some tried to explain this paradox by noting that it was difficult for workers to adjust to computers. Others noted that few computer applications made significant improvements in the amount of work most workers could do. Still additional commentators felt that the paradox was a product of poor statistical measurement.

Because this paradox was driven by the negative productivity results for several service industries, one approach was to see if the service productivity figures were accurate. One study, by Joel Popkin and Company for IBM,(1) found that the BLS productivity statistics Roach used for several service sectors had important shortcomings. Most importantly, the BLS productivity data relied on output measures that did not truly reflect the changes in the nature of work in some service industries. If these are corrected in several important service industries, two things could be shown.

1. Many of nonservice industries had a positive relationship between investments in computers and productivity.

2. When different methods were used to estimate output for several service industries, their productivity growth, instead of being negative, was slightly positive.

To extend this work, I have added investments in communications to investments in computers (and office equipment). By adding communications equipment spending, a more accurate picture is created of the relationship between computer and communications technology (IT) and productivity. In addition, looking at the overall IT investment (for computers and communications) shows a positive relationship between such investment and productivity, although the fit is much less than tight.

Figure 1 presents the results of plotting spending on communications and computers as a percent of all spending on equipment against the annual productivity growth for two data sets that include industries covered by U.S. Department of Labor statistics for the years from 1977 to 1989. Both regression lines show a positive correlation between spending on communications and computers and productivity. Interestingly, however, the regression line for the data for the years 1985 to 1989 is higher than the 1977 to 1984 regression line when the ratio of spending on communications and computer equipment was less than 50 percent of total equipment spending. The regression line for the later years' data drops below the regression line for the earlier years when spending on communications and computer equipment is greater than 50 percent.

This suggests that industries that were spending smaller amounts on communications and computer equipment in the 1984 to 1989 period were achieving greater productivity gains than in the 1977 to 1984 period. The changes between the two periods suggest that industries that have invested lower amounts of their equipment spending in communications and computers are beginning to see a greater productivity payoff from such spending. This would lead one to assume that a recognition of this fact would result in greater spending on communications and computers, including spending on storage.

If this is true, one might expect to see an increase in future years not just in spending on communications and computing but in annual average productivity gains by industries that have not spent much on communications and computing equipment historically. This is reflected in the third line drawn on the graph, which is higher than the ones fitted to the data from the two previous periods, 1977 to 1984 and 1985 to 1989. This line depicts a projection for the likely productivity gains due to spending on communications and computing during the 1990 to 1995 period. It is drawn to reflect an assumption that firms that spend lower amounts on communications and computing will achieve significant gains in productivity over the 1990 to 1995 period.