X hits on this document

45 views

0 shares

0 downloads

0 comments

11 / 16

154

JOURNAL OF SPORTS ECONOMICS / May 2005

TABLE 5:

Probit Regression of the Impact of Team and Market Area Racial Mix on Team Choice to Retain a Player

Explanatory Variables

Coefficient

Robust Std. Error

TWHITE : Percentage of (year 1) team who are White TWHITE : Percentage of (year 0) team who are White POPWHITE : Percentage White of (year 1) SMSA population POPWHITE : Percentage White of (year 0) SMSA population RACE: Player’s race (1 = White)

1T 0T

1T 0T

–1.372*** 0.143 0.557 1.003* 2.207**

0.548 0.580 0.581 0.573 1.038

0.119

1.244

–2.536*

1.323

0.515

1.420

–2.672**

1.331

–0.955***

0.423

TWHITE

1T

TWHITE

0T

POPWHITE

1T

* RACE

POPWHITE

0T

* RACE

  • *

    RACE

    • *

      RACE

CONSTANT

NOTE: The sample consists of all player-team affiliations during three seasons: 1996-1997, 1997-1998, and 1998-1999. To be included in the sample, the player must be observed for more than one season so that comparison data are available for both the original-period team and the subsequent-period team. Hence, a player is observed each time he has a team affiliation during the three seasons, including midseason changes in affiliations. We then observe whether the player changed affiliations or remained with the previous team. If the affiliation changed, the observation includes demographic data on both the original team (TWHITE0T) and host city (POPWHITE0T) and the new team (TWHITE1T) and host city (POPWHITE1T). If the affiliation did not change, then the observation includes Time 0 and Time 1 team and city data for the original team. The dependent variable equals 1 if the player is observed moving to a new team between Time 0 and Time 1, and it equals 0 if the team retains the player for the Time 1 season. Data sources appear in previous tables. ***Significant at .01 level; **significant at .05 level; *significant at .10 level.

The regression coefficients provide suggestive evidence that, in the last three NBA seasons in the 1990s, the performance of White players was higher in cities with larger White populations. In combination with the Table 4 results, this implies that better-performing White players have navigated toward locations that place a higher premium on their performance.16 Such a result is consistent with customer discrimination, although of a more limited type than was documented in studies using 1980s data. This “sorting” can of course only be achieved through player movement, implying that the trading of White and Black players should also be influenced by the racial composition of the team and the racial composition of its market area. Table 5 reports the results (from the 1996-1997 through 1998-1999 seasons for which we have data on player trades) of testing this proposition by mod- eling the probability that a player will be traded to a new team as a function of the percentage White on the old and new teams, the percentage White in the old and new teams’ market areas, the player’s race, and interaction between the player’s race and each of the team and market-area variables.

We find that White players are significantly more likely to be traded than are Black players. Black players are more likely to be traded from a team located in a city with a relatively large White population and are less likely to be traded to a team

Document info
Document views45
Page views45
Page last viewedTue Dec 06 19:27:44 UTC 2016
Pages16
Paragraphs357
Words7200

Comments