Comments

Thomas J. Holmes: This paper is the third installment of the "Whom Benefits Whom?" series by Joel Waldfogel and his coauthors.²⁵ It is now well understood that when scale economies and product differentiation are important, individuals can benefit from having more people live nearby—the increased demand can make it economically feasible to make locally available a large variety of products. The big idea of Waldfogel's research project is that when individuals have different preferences, it matters who these additional neighbors are. If additional neighbors to individual x have different preferences than individual x, the additional neighbors may not help the individual get access to more of the kinds that he or she likes. In fact, adding more people with different preferences might even hurt individual x. Waldfogel focuses on media markets, but the idea is more general. If a small town has a minor league baseball team, and if a bunch of people who like hockey move in, the increased scale of demand may make it possible for the town to get a hockey team. If the hockey team takes some of the baseball team's market away, leading to the exit of the baseball team, the addition of the hockey fans has hurt the baseball fans.

The earlier papers in the series examined how the sizes of racial and ethnic groups affected the newspaper market.²⁶ These are very interesting papers. They make a compelling case that different demographic groups have quite different tastes and that product variety targeted at a particular group is closely linked to size of that demographic group in the market. The work also shows that consumption behavior is affected, suggesting that this increased variety has welfare effects. For example, a black person living in an area with a large black population is more likely to listen to the radio. The presumption is that this change in behavior results from the larger variety of black-targeted radio found in areas with larger black populations.

The current installment of the series conducts the same kind of exercises with local television markets. Given the size of the television industry, it is an important one to look at, and the paper makes a useful addition to the series. My discussion just raises one concern about a data limitation that arises with this project, which was not as much of an issue in the earlier work. This limitation makes it difficult to cleanly measure the link between variety and market size.

Data Limitation

To examine the relationship between the number of products targeted at a particular ethnic group and its market size, one needs some way of classifying the degree to which a product is targeted at a particular ethnic group. In Waldfogel's article "Preference Externalities,"information comes as part of the raw data. Radio stations are classified in the data as, for example, "black-gospel," "black-talk," and so on.²⁷ It is then straightforward to examine the link between the number of black-gospel stations in an area and the size of the black population.

The data on television viewing considered are rich in many ways. For each half-hour time slot for each station, there is demographic information about who is watching the show. However, there is no information about what the program is that people are watching. There is no information that the show is Baywatch or Friends. There is no analog to the classification of a program as "black-gospel" in the radio data. To get around this issue, Waldfogel attempts to infer something about the degree of targeting of a program, based on who is watching the program. A difficulty is that markets with larger black populations will naturally have larger black audiences, even if the programming remains the same.

Some formal notation helps make the point. Let i index a half-hour slot for a particular local station. Let t_i ∈ [-1,1] represent the degree...