Competition in shared markets and Major League Baseball broadcast viewership

ABSTRACT

This work evaluates the cross-quality elasticity of related products in the context of local market Nielsen Local People Meter ratings of Major League Baseball (MLB) regular season broadcasts from 2010 through 2013 from six teams in three shared markets. We employ a fixed effects panel regression with multi-way error clustering, finding that fans exhibit nuanced behaviour related to the absolute quality and relative quality of the two local teams. Our estimates imply quality-related competition for viewership between teams in the face of large disparities in quality. However, when both teams are of high quality, viewership increased beyond what own-team success would predict alone for the competing team. The competitive effects are largely dominated by the spillover effects. These findings point to complementary effects of team success beyond own-team interest, and bring about an important nuance in the literature on market definition, competition and substitution in sport.

KEYWORDS:

• Market definition
• spillover effects
• fan substitution
• broadcast ratings
• baseball

JEL CLASSIFICATION:

• D12
• L13
• Z21

Acknowledgements

The authors would like to thank The Nielsen Company for providing the data in kind for this research. We would also like to thank, without implicating, Jason Winfree and Thomas Peeters for their helpful comments on earlier versions of this work. Any remaining errors are our own.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Rottenberg (1956) begins his demand specification by noting, ‘Attendance at baseball games, as a whole is a function of the general level of income, the price of admission to baseball games relative to the prices of recreational substitutes, and the goodness of substitutes’ (p. 246).

² Friday games are almost always at night.

³ In our data set, about 98% of all games are televised either at the local or national level.

⁴ National broadcasters are allowed at certain amount of co-existing broadcasts on certain nights of the week or at certain start times on the weekends.

⁵ While Los Angeles and Chicago are also home to two MLB teams, this research is part of a larger study of markets with multiple teams in both MLB and the NFL.

⁶ Baltimore and Washington are considered part of the same combined metropolitan statistical area (CMSA).

⁷ GB itself identifies the number of additional wins required for a team to have the same current winning percentage as the division leader. It is generally reported as half or full games, depending on the relative number of games that the team and its respective division leader have played up to that point in the season. If a team is leading the division, then they have a negative number reported for GB. We note that we also estimated our models using a zero in place of a negative value, and the results and interpretation were unchanged when this variation of the measure was used.

⁸ For example, after a single game, a team’s win percent is either 1.000 or 0.000. This provides little information about fan perceptions of team quality. For WP20 early in the season, the measure spans to the last games in the previous season.

⁹ This avoids bounding issues with our variable that would occur if games back held the quantity zero for any team in first place in the division.

¹⁰ Each team may also enter into our model as an opposing team, j, in which case there will be two observations for the given game, as there are with shared market team matchups.

¹¹ Some games in our sample appear more than once when the teams in the sample play one another, or appear on multiple networks. Therefore, we cluster the errors for these observations to account for any lack of independence in these observations. We estimated models using other error specifications, which do not change the sign or significance of our results in any meaningful way. The results of these alternative specifications are available upon request.

¹² We note that our model does not include economic variables normally included in a demand function, and therefore we characterize our estimation as a consumer choice model, with the yearly fixed effects subsuming any changing economic characteristics during the relatively short time period over which our data set takes place.

¹³ All LocalvLocal games are interleague games in our data.

¹⁴ We note that this relationship is particularly important in the context of the Baltimore/Washington market, where the Orioles receive a large portion of television from the Nationals broadcasts as stipulated by their agreement when the team moved to the area from Montreal. The incentive for Orioles ownership to prefer a higher quality Washington Nationals team seems clear in the context of expected television revenues.

¹⁵ We note that our estimations only evaluate the impact of existing competitors, and do not address the effects on an incumbent team of introducing a new competitor into the market. We leave this for future research, as the full scope of competition requires years prior and following the introduction of the new team into the market.