It is obvious that modem telecommunications networks possess at least
some of the attributes of what we commonly call 'infrastructure'. They
are essential to modern commerce and, indeed, to modern living. It is
essential that all subscribers be able to connect to most other subsc
ribers, but perhaps not through the same network but through interconn
ected networks. These networks may even have natural monopoly characte
ristics and may generate substantial externalities. Nevertheless, the
speed of technological change in electronics and communications provid
es a strong argument against government programs to develop or build m
odern telecommunications infrastructure. The choice of any network arc
hitecture is fraught with risk and could require the investment of $10
00 or more per subscriber that could easily be obsolete as soon as it
is completed. Rather than providing grants of monopoly and relying on
the ensuing regulation of network architecture, governments should ope
n their telecommunications sectors to competition, allowing private fi
rms to shoulder the risk of building these expensive new networks. Ind
eed, the United States has recently moved strongly in this direction a
fter a quarter century of selective telecommunications liberalization
that has seen private networks flourish. The evidence on the effect of
new telecommunications infrastructure on economic growth is too weak
to justify a conclusion that this infrastructure has already created l
arge externalities. As networks proliferate, with private and public s
ystems competing with each other and complementing each other, it will
be even more difficult to demonstrate these externalities. Thus, the
case for government support and direction of telecommunications infras
tructure investment remains very weak. (C) 1997 Elsevier Science B.V.