A week or so ago, Robin Hanson objected to the label "network effects" on grounds that "externalities" fits better, that economists should stick standard terminology, and that "effects" is so loose as to mean nothing. Although I found that a pretty cogent argument, I remained silently unconvinced. I now offer some data supporting a contrary view.
Robin's argument left me unconvinced because I think that it courts confusion
to call call mere *pecuniary* externalities caused by network effects by a
term normally reserved for *non-pecuniary* externalities. Pecuniary
externalities represent mere "budget" effects on third parties, as when
competition leads to lower prices throughout a market. They do not at all
indicate market failure. I appreciate Robin's theoretical concern for the
purity of econo-speak, but using "network externalities" would in practice
invite policy-makers and journalists to draw the wrong conclusions about how
to react to the phenomena in question. Such non-economists would assume that
"externalities" indicates market failure. Futher study leads me to believe,
moreover, that most economists would draw the same conclusion.
I attended a conference last weekend attended by scholars of law and economics
where spoke Joseph Farrell, a professor of economics at UC Berkeley. Farrell
has, by all accounts I've heard, done pioneering work on network effects.
Noting that Farrell's talk consistently used "network effects" rather than
"network externalities," I asked him about his choice of terms. He explained
to me that economists decided in the 1920s that "externalities" ought not apply to mere pecuniary effects. He added that, while he thinks that network effects sometimes lead to conventional externalities, they often do not. Hence his preference for "network effects."