I decided to run a multiple linear regression on the data, hypothesizing that we could predict teams' media rights using
their market size, on-the-field performance, and the amount of TV coverage offered. Using that Sponsor chart, Baseball Reference for the 1960 season, and this population database, I compiled the following variables:
- 1960 winning percentage
- Top WAR in 1960
- Number of Hall-of-Famers on the 1960 roster
- Number of players who made the 1960 All-Star Game
- Number of managers in 1960 (quality teams fire their managers less often; others, like the Red Sox, had 3[!])
- 1960 attendance
- Population size
- Two teams (a dummy variable: 1 if there were two teams sharing the market, 0 otherwise)
- New team (a dummy variable: 1 if the team had been founded in the last 10 years, 0 otherwise)
- AL or NL (a dummy variable: 1 if AL, 0 if NL)
- Number of 1961 games played, since the AL and NL played different numbers of games that year
- Number of televised games
- Number of TV station feeds
- Color broadcasts (a dummy variable: 1 for yes, 0 for no)
Before I get into the results, there's one more important factor: 1961 was the first year of the Expansion Era, as the Los Angeles and Minneapolis/St. Paul markets joined the American League. More specifically, the Angels were founded as an expansion team, the original Washington Senators moved to Minnesota and became the Twins, and the second Washington Senators team set up shop in D.C. Obviously, not all of these variables would therefore be available for all teams. The Twins inherited the on-field related variables from the 1960 Senators (e.g., winning percentage and Hall-of-Famers), but the Senators inherited market-related variables (especially 1960 attendance).
I was a little surprised to find that market size, and market size alone, contributed to the best model. Using R, I was able to produce the following model:
rights = .0002822*(1960 attendance) + .02356*(1960 population in 000s) + 128.8
This model has an adjusted R2 value of .7933, and both coefficients are significant with p < 0.01. For this model, I included only those 16 teams that had an attendance figure for 1960, so the Angels and Twins were out of luck.
I'm still a little disappointed that there is no direct correlation between on-field performance and the value of a team's media rights, though obviously winning teams draw better. This makes it seem as if the link between winning and revenue was tenuous in the age before free agency, contrary to the models produced by people like J.C. Bradbury to explain the differences in modern franchise values. Add it to the list of things to investigate, though it should take some actual work to get franchise valuations from 1960.
Lastly, here is a table describing the difference between each team's predicted and actual media rights.
|CITY||TEAM||LG||1960 POP (000s)||1960 ATTEND||PREDICTED (000s)||1961 RIGHTS (000s)|