THE EVOLUTION OF ADVANCED LARGE SCALE INFORMATION
INFRASTRUCTURE IN THE UNITED STATES

By Shane M Greenstein, Mercedes M Lizardo, and Pablo T. Spiller
(Northwestern University, University of Illinois, Urbana/Champaign, and
University of California, Berkeley)


I. Introduction
Many recent information infrastructure policy debates have
centered on redefining "universal access" to accommodate new
information technologies. Analysts debate whether private
industry is providing equal access to new technologies or meeting
other national goals for wide dispersion of advanced information
infrastructure. Despite the prominence of this concern, not much
is known about how the recent rapid growth in digital backbone
technologies altered the distribution of information technology (IT)
capital stock across the U.S., if at all.
To address this question, we examine the distribution of
advanced large-scale backbone computing and digital
telecommunications infrastructure between 1986 and 1992, a
period of rapid growth in the backbone digital technology that
makes up information capital. Our study emphasize two findings.
First, like all infrastructure in the U.S., advanced information
infrastructure is unevenly distributed. It follows population and
locates in dense urban areas and regions with large amounts of
white collar work. Second, over time it has become less
concentrated in any specific region. Over these years advanced IT
diffused across the geographic landscape of the U.S. and access to
it increased.
We analyze changes over time in the geographic
distribution of three key ingredients of the nation's digital
infrastructure: a) fiber optic deployment by local telephone
companies, b) the number of large-scale computer users, and c) the
processing capacity of large-scale computers. Our goal in this
paper is to understand how the distribution of IT capital stock
changed as it grew in the late 1980s and early 1990s. We do not
attempt to fully explain why the IT capital stock grew as rapidly as
it did, which is an enormous research question partly addressed by
other work. Instead, we confine our attention to a measurement
question that directly relates to an important policy discussion --
did the distribution of advanced IT capital become more or less
geographically concentrated during a period of rapid growth? We
document and analyze the changes that actually occurred.
Our indices are composed of very different parts of
backbone technologies for regional digital IT networks. Both
computing and telecommunications recently experienced rapid
declines in price and rapid expansions in capital stocks. Each
serves different functions, however. The growth in fiber optic cable
at local telephone companies measures a region's ability to
transmit data in digital form, while growth in large-scale
computing capacity measures the ability of firms within the region
to store and process large databases. In both cases, we interpret the
level of capital stock as representing a variety of activities
associated with building public and private digital IT networks in a
region.
These two technologies have very different determinants.
First, the firms making the investment decisions for fiber are
regulated, while the investors in computers are largely unregulated.
Second, large-scale computers began diffusing decades ago,
although the recent competition to large-scale computing may have
altered its distribution. In contrast, the diffusion of fiber as a
market-oriented experience is a comparatively recent phenomenon.
Yet, in spite of these differences, the customer bases for digital
telecommunications and large-scale computing partially overlap.
Administrative-intensive work, such as banking, finance,
insurance, real estate, or business services, tends to make heavy
use of advanced computing and digital communications. Of
course, there are also plenty of non-overlapping industries.
Because of these different investment processes and influences, it
is not obvious whether these technologies should become more
widely or less widely dispersed during rapid growth and whether
the pattern of location of one of these technologies correlates with
the pattern exhibited by the other.
We examine the post-divestiture years for both backbone
technologies in 1986, 1989, and 1992, largely for pragmatic
reasons. These are years of rapid investment and excellent
documentation. Our study uses a variety of standard methods for
analyzing the distribution of the capital stock across regions, as
well as the distribution of access to capital.
We reach a stark conclusion. All indices we analyze show
that these IT backbone technologies have become more widely
distributed and less concentrated across the U.S. over time. This
holds for the distribution of capital stock and for access to it. It is
not possible to sustain any Cassandra-inspired argument that the
distribution has become more unequal over time. We conclude that
much of the current policy concerns about the distribution of
backbone technologies need careful refinement to reflect these
facts.