salary per market and a typical range. A regression of the log of the top salary on log of
population yields an elasticity of 0.87 (with a standard error of 0.04). Using the midpoint
of the typical salary range, the elasticity of salary with respect to market size is 0.26 (0.02).
These size differences in salary dwarf the cost-of-living differences across markets and
therefore clearly indicate higher costs in larger markets. Using the 16 markets for which
a metro area CPI exists, the elasticity of the cost of living with respect to population is
about 0.03. The pay–market size relationship also suggests that higher-quality talent grav-
itates to larger markets and that the stations in larger markets have higher quality.
Second, there is related indirect evidence: the relationship between the number
of listeners per station and market size is strongly suggestive that costs are higher in
larger markets. A regression on the average number of listeners per local US radio
station on metro area population, using 1997 data, yields an elasticity of 0.8. Unless
ad revenue per listener fell sharply with market size—and there is no evidence that it
does—this would indicate that radio stations in larger markets have higher average
revenue. With free entry, revenues tend to provide reasonable approximations for costs,
so the fact that larger-market stations have more revenue indicates that their costs are
higher as well.
37
That larger markets have more media products provides a mechanism by which addi-
tional entry might deliver more valuable choices to consumers. That is, entry is a mech-
anism for the delivery of the basic own-group preference externality. Costs that rise with
market size seem on the face of it to mitigate the positive effect of market size on the
desirability of options for media consumers, but that depends on what gives rise to higher
costs in larger markets. For example, if larger markets had higher costs simply because of
higher input prices (land, etc.), then the deviation from linearity of entry in market size
would inhibit the welfare benefit of fellow consumers. On the other hand, if the higher
costs in larger markets reflect higher investment—and associated higher-quality
products—then the interpretation would be different.
There is a variety of reasons to see the higher costs in larger markets to be reflective of
greater investment in quality and, moreover, that media markets provide good examples
of
Sutton’s (1991) prediction that market structure need not grow fragmented as markets
get large when quality is produced through investments in fixed costs. In newspapers,
some of the direct input cost measures—page length and staff size—are directly suggestive
of quality. Moreover, some other measures of quality, such as the number of Pulitzer
Prizes per newspaper, are also higher in larger markets (
Berry and Waldfogel, 2010).
That quality rises in market size provides a second mechanism, in addition to variety
itself, whereby consumers might deliver more surplus to one another.
37
One might worry that this is driven by large markets where spectrum scarcity delivers rents to stations that
can enter, but this seems not to be the case. The elasticity is 0.65 even for markets with fewer than 1
million residents.
31
Preference Externalities in Media Markets