True Value in the NBA:
An Analysis of On-Court Performance and Its Effects on
Revenues
Undergraduate Honor Thesis
Harrison Li
Advisor: David Card
Department of Economics
University of California, Berkeley
May 2011
Abstract
Previous studies have investigated the relationship between player and team performance in
the National Basketball Association (NBA). Separately, others have looked into team
performance and it’s correlation with revenues. There are also studies that connect these two
relationships in order to determine the marginal revenue product (MRP) of individual players
for Major League Baseball. There is significantly less literature on this specific task in the NBA.
This paper estimates these relationships using data for several `seasons of NBA play, and then
uses the results to estimate the value of individual players to their team. This study finds that
the salaries paid to NBA stars closely match with their marginal revenue products.
Acknowledgement: I would like to thank Professor David Card, my thesis advisor, for kindly guiding me along the
research and writing process and for his willingness to help and answer my questions. Without his assistance, this
paper would not have been possible.
Introduction
“For the love of the game.” That is a popular response from athletes when answering
questions as to why they put their bodies and minds through such vigorous training year in and
year out in order to compete professionally. Despite the self-professed motives of the athletes
themselves, many analysts believe otherwise. Specifically, most commentators point to the
inordinately high salaries earned by professional athletes as the primary motive for their
extreme work ethic. For example, during the 2009-2010 NBA season, the Los Angeles Lakers’
superstar Kobe Bryant earned over 21 million dollars. Where do the owners and general
managers of the NBA come up with this amount to offer a player like Kobe Bryant? This is the
question that the following paper will answer.
Standard economic reasoning suggests that a player’s salary will be set to
(approximately) equal his expected contribution to the team’s revenues over the season – his
so-called “marginal revenue product”. From a fan’s perspective, a player’s contributions mostly
relate to the team’s win-rate: can this athlete help win the team more games and eventually
secure a championship title? However, when owners analyze this problem they actually think in
much more economic terms. They believe that the player will improve the team’s performance
and in turn will generate higher revenues which are generated from gate receipts, unshared
local television contracts,
1
and distributed national television contracts
2
.
There have been several papers that already examined the relationship between certain
variables and NBA team revenues. Some of them have looked into the belief that there are a
few star players in the league that single handedly have a significant effect on team revenues.
Hausman and Leonard (1997) proposed that certain players’ stardom was the significant
revenue driver in the NBA. They illustrated this point through their data analysis, stating that
Michael Jordan of the Chicago Bulls was responsible for 200,000 dollars of the New Jersey Nets’
revenue in a season during which the Bulls only played twice at New Jersey. On top of this,
Hausman and Leonard suggest that superstars such as Larry Bird and Michael Jordan in fact had
1
http://nbcsports.msnbc.com/id/16877177/
2
http://www.nba.com/2010/news/features/david_aldridge/02/15/morning.tip/index.html
a significant effect on team attendance
3
. This in turn positively affected the team’s revenues. It
seemed that a major factor in team revenue was not the number of wins they could accrue
over a season but how many NBA stars they could acquire in order to attract more fans.
Berri, Schmidt, and Brook (2004) extended this work and delved deeper into the
“superstar” effect. Not only did they examine certain star players like Michael Jordan, Shaquille
O’Neal and Grant Hill, but they also looked at players who did not have as much popular
recognition but were still considered All-Star caliber players by the NBA.
4
They found that these
kinds of stars did not have significant effects on gate receipts, which is one of the major
revenue sources for an NBA team. They concluded that a star’s effect on revenue was mostly
due to their effects on a team’s win-loss record and not from the popularity of that individual
player.
This paper examines the previously mentioned relationship and connects it with the
relationship between players and team wins. This paper is similar in scope to the research done
by Macdonald and Reynolds (1994) in which they found the marginal revenue product of an
average baseball player.
Data to be Employed
This paper ultimately looks at the changes in revenue due to the statistics of a certain
player. The revenue data goes back to the 1998-1999 season (the NBA season that started late
because of a lockout) and ends with the 2009-2010 season
5
.
Revenue Factors
Team performance variables will eventually connect a player’s statistics to changes in
revenue. The specific variables that will be included are the number of regular season wins
(current and last year’s), playoff wins from the current and previous years, as well as
championship dummy variables for last year and two years prior. The lagged regular season
3
Albeit Hausman and Leonard did state that their study on attendance effects of a superstar to be “less formal”
(pg. 609)
4
This was determined by summing up the number of All-Star votes a team had in total
5
Revenue data collected from Forbes. The revenue numbers include gate, media and stadium revenues
variables have been included because fans often have delayed reactions to the success of a
team. During a season, the success of a team will attract higher revenues through gate receipts
and possibly renegotiated TV contracts. However, the entire spectrum of effects of an improved
team will not be fulfilled within the current season. For example, after a more successful
season, season ticket sales for the next year will increase as well as demand for single game
tickets. This variable will capture increased revenues from possible “band-wagon” fans.
Along with regular season wins, playoff wins have also been included as a factor for
team revenues. Even though there are much fewer playoff games played, this factor should be
significant. Not only are playoff tickets much more expensive than regular season tickets (and
get exponentially more expensive the farther a team goes into the playoffs), but also the
deeper a team gets into the playoffs the better quality the team. This is how we can
differentiate between the truly elite teams and those that barely made it into the playoffs
6
. For
this reason, playoff dummies were not used instead of number of playoff wins, as it gave too
much credit to teams that got knocked out in the first round. On top of this factor,
championship dummies were also included to try to capture any significant increases in
revenues because of recent past championships. It is one thing to get into the playoffs, but
there could be a greater separate effect if a team wins it all.
On top of team performance, general factors such as interest in the NBA and city
variables are major contributors to revenues as well. The way this paper will address these
factors is through city dummy variables for each team, as well as dummy variables indicating
the year. The city dummy variable will pick up general economic effects in the urban area that
the team is located in, and the dummy year variable will account for effects like general interest
in the NBA and the national economic environment.
Lastly, there is a dummy variable that indicates a ‘1’ if the team did not sell out the year
before
7
. This variable is then interacted with the current season wins.
6
Presumably these teams will win only a couple playoff games and get knocked out in the first round
7
Attendance data collected from ESPN.com
All these variables were then put into a relationship with revenues and it resulted with
equation (1):
= α +


+
i = 1,2,….,354 (1)
Y = Log(revenues)
X = Revenue factors
Win Factors
The next part of this paper deals with the relationship between wins and certain
basketball statistics
8
. This will determine how the statistics a player obtains over the course of
the season affect the number of wins of his team. The manner in which this paper goes about
determining this effect is based off of the approach created by Scully (1974). His method found
the marginal product of a player by connecting his statistics to wins. Berri (2004) developed a
method where winning percentage was regressed on points per possessions and points allowed
per possession. This method is based on the concept of possessions developed by Oliver and
Hollinger (2003) who suggested that the main determinant of wins was not the absolute
statistics a player amassed during a season but how efficient they were at doing so. The data
collected for this paper takes that into consideration.
In order to accomplish this, total aggregate statistics for each team and their opponents
for that year were recorded. This was completed for both the regular and post season. These
opponent statistics are unique to each team as they all play a different mix of teams and
perform differently against each one. The reason for collecting the opponents’ statistics was to
account for players on teams with higher tempos, which score more points yet also allowed
more points, or teams that had a high-powered offense but lacked in defensive prowess were
not given a bias. In effect, this found how efficient teams were at scoring points relative to the
teams they played.
8
Basketball statistics collected from Basketball-reference.com
The end variable used in the win estimation model is the statistics of the specific team’s
season stats, divided by their specific opponents’ season statistics. All the major statistics were
included: two-point field goals made, three-point field goals made, free throws made,
turnovers, defensive rebounds, blocks, assists, personal fouls and offensive rebounds. The
offensive rebound statistic had to be relative to the number of field goals missed or else this
statistic would give a negative coefficient. This would suggest that obtaining offensive rebounds
would actually hurt your team’s chances of winning
9
. Of course this is not true, so how could
we explain this phenomenon? The answer is that teams that tend to lose more games miss
more field goals, which in turn gives the team more opportunities to collect offensive rebounds.
In order to combat this effect, this study has created another variable that tries to find the
value of another offensive rebound given the amount of field goals missed.
These ratio statistics were put together into equation (2):
= α +


+
i = 1,2,….,10 (2)
Z = Team’s total season statistics
Opponents’ season statistics
W = Number of wins
Estimation of the Models
The following is equation (1) which was used to model team revenues on certain factors:
= α +


+
i = 1,2,….,354
Below is Table 1 that shows the results for this linear regression. Out of the top 7
variables in the table only four of them are significant in affecting revenue: number of current
season wins, number of lagged wins, number of playoff wins and the interactive sell out
variable. It is clear that out of these seven team performance variables that current season wins
has the greatest effect on revenues. The coefficient of .0036, in this context, means that about
9
In the book “The Wages of wins: taking measure of the many myths in modern sport” by Berri, Schmidt and Brook
(2006) they found that offensive rebounds had a coefficient of -.2
every extra win during the season will bring in .3% more revenue to a team. To put this into
context, each individual win for a team like the Los Angeles Lakers, who gained about 214
million dollars in revenues this past year, brought in about 642,000 dollars. And that the 57
wins that the Lakers accrued last year contributed to 57*.0036431 ≈ 20.8% of the revenues that
were generated that year. Finally, since the Lakers did sell out
10
on average this past year the
positive effect of wins was not offset by the interactive term.
TABLE 1: Estimated coefficients for equation (1)
11
Variable
Coefficient
t Statistic
P Value
Wins
0.0036431
5.21
0.000
Wins(-1)
0.0029043
4.76
0.000
Championship(-1)
-0.018641
-0.50
0.620
Championship(-2)
0.0043962
0.13
0.894
Playoff Wins
0.0043694
2.37
0.019
Playoff Wins (-1)
0.0007206
0.36
0.722
No98
-0.0015006
-4.28
0.000
Observations = 354
R
2
.9436
Adjusted R
2
.9343
Wins = Current regular season wins; Wins(-1) = Lagged regular season wins; Championship(-1) = Dummy
variable for lagged championship won; Championship (-2) = Dummy variable for two year lagged
championship won; Playoff Wins = Number of playoff wins this season; Playoff Wins(-1) = lagged
number of playoff wins; No98 = Interactive variable with Wins, this variable had a value if the team did
not sell out (using 98% capacity as the cut off for selling out) on average the year prior and had a ‘0’ if
they did sell out the year before.
The negative effect on the interactive term (No98) came as a surprise. The initial
intuition behind this variable was that if a team had sold out on average the previous year
10
Sell out cut off was filling up 98% of full capacity on average.
11
There are 43 more dummy variables which encompass yearly effects and general city effects. The cities with the
largest coefficients are the large sports markets with rich NBA history (listed in order of coefficient size): New York
Knicks, Chicago Bulls and the Los Angeles Lakers.
(dummy variable would equal ‘0’), the effect of wins would be diminished. This belief emerged
from the thought that once a team had a strong fan base, the additional revenues to be gained
would be tougher to obtain than the initial gains, because the team would then have to look for
more creative ways to increase revenues such as better advertising deals and local television
contracts which take a longer time to take effect than gate receipts. If this hypothesis had been
correct, then the interactive variable should have a positive coefficient.
Because this interactive variable has a significantly negative coefficient
12
, it must be re-
evaluated. There are several reasons why this variable should be negative. Firstly, if a team
could not sell out last year there is a good chance that it is not a top caliber team which would
hurt future season ticket sales. On the topic of ticket sales, if a team was not able to sell out the
previous year, then they will not be able to raise ticket prices in the following year because they
know that demand will not be high enough to warrant a rise in prices. Lastly, a team that isn’t
able to win a lot of games and bring in a lot of fans will not be attractive to corporate sponsors.
These three reasons are epitomized in the Boston Celtics. Before the 2007-2008 season,
the Celtics had been consistently recording a below .500 record. What changed for the Celtics
during the offseason? They acquired two All-Stars in Ray Allen and Kevin Garnett. The first year
these two were acquired, the Boston Celtics went on to win 42 more games than they did the
previous year and secured an NBA championship. With this success, the Boston Celtics were
able to sell out their games that season which situated them for higher profitability in the
future. Even before the season ended, Boston Celtics executives were planning to raise ticket
prices 10-15% for the next season. On top of this, they wanted to sign additional corporate
sponsors, which were estimated to be worth 5 – 10 million dollars
13
.
Along the same lines, the lagged wins variable has a strong effect on a team’s revenues.
The intuition behind this variable is very similar to the intuition that was just explained using
12
Negative for all sell out percentages from the cut off of being a “sell-out” of 100% all the way down to 90% of full
capacity.
13
These two factors: gate receipts and sponsorship deals were specifically important to Boston Celtics team
revenue.
http://www.boston.com/sports/basketball/celtics/articles/2008/06/05/ticket_prices_sponsors_on_rise_for_the_g
reen/?page=2
the Boston Celtics as an example. In addition, this lagged effect is greater than the effect gained
from the wins during the current season when a team did not sell out their games the year
before.
Is there a similar relationship with playoff wins and lagged playoff wins? The number of
playoff wins positively affects the revenue. Even though there are relatively very few playoff
games to gather revenues, these ticket prices can be sold at much higher prices. Unlike regular
season wins, the lagged playoff victories are not significant.
Total Team MRPs
There are generally two different types of teams that obtain success in the NBA. The
first type follows the route of the Boston Celtics who signs or trades for multiple stars in order
to improve the quality of their team. In doing so they are able to quickly obtain more wins than
they did in the previous season and therefore quickly grow their revenues. In the instance of
the Boston Celtics, the increased number of wins can be clearly attributed to the acquisitions of
the two superstars and we can see the effect these wins had on Boston’s revenues:
% Effect on Current revenues (.0036431 - .0015006) * 42 = .089985 => 9%
% Effect on Current revenues (.0043693) * 16 = .06991 => 7%
% Effect on future season revenues (.0029043) * 42 = .12198 => 12.20%
The above numbers are the percentages that the increase in wins by the Boston Celtics
affected Boston’s revenues. During the 2006-2007 season, the Boston Celtics obtained 24 wins
and 58 losses. The following season they obtained 66 wins and won the NBA Championship. In
the above equations, the number 42 is calculated from the difference in wins between the
2007-2008 season and the 2006-2007 season. This is the increased revenue obtained from the
extra wins. The 16 is calculated in the same way that the 42 was calculated. The total revenue
effect from the improved team performance can be calculated in the following manner:
Boston Celtics 2007-2008 Revenue: $ 149 Million dollars
Boston Celtics 2006-2007 Revenue: $ 144 Million dollars
Team Performance MRP:
9% * 149 = $ 13.41 Million Dollars
-> The immediate dollar effect of the increase in wins over last season.
12.20% * 144 = $ 17.57 Million Dollars
-> The delayed effect that this season’s increase in wins over last season will have on the
following year’s revenues.
7% * 149 = $ 10.43 Million Dollars
-> The immediate effect the increased playoff wins had on team revenues.
Total team MRP = $ 41.41 Million Dollars
Team Salary during 2006-2007 Season = $ 53.62 Million Dollars
Team Salary during the 2007-2008 Season = $ 73.81 Million Dollars
Total increase in salary = $ 20.19 Million Dollars
Luxury taxes paid in 2007-2008 Season = $ 8.32 Million Dollars
Total player cost = 20.19 + 8.32 = $ 28.51 Million dollars
On the other hand, there are more teams that take another route. They obtain younger
players who they believe will develop into stars and sign cheaper supporting players to help
their developing stars. This strategy keeps a team’s roster relatively constant throughout the
years. This can be demonstrated through the Atlanta Hawks team. Between the 2006-2007 and
the 2009-2010 seasons, the team had a slow and steady increase in wins from 30 to 52 wins.
Using the same techniques we used for the Boston Celtics, we find that these 22 more wins
from these two periods should account for about a 14 Million dollar increase between the two
periods. In fact their team revenues jumped up by 20 million dollars, and the team improved
team performance was matched by a 23 million dollar increase in team salary. This salary
increase was mostly for players that were originally on the team but got larger contracts for
their improved performance.
Estimation of the Win Model
In order to complete our estimation of a specific player’s worth to a team, we must model
number of wins against the specific statistics of each team. Referring back to equation (2):
= α +


+
i = 1,2,….,10
Z = Team’s total season statistics
Opponents’ season statistics
Table 2: Estimated values for equation (2)
Variable
Coefficient
TwoFG
24.4153
FT
3.2402
ThreeFG
17.2092
DRB
112.6733
ORB
3.217
AST
1.4918
STL
5.7035
BLK
2.0905
TOV
-56.0667
PF
-12.7199
Number of observations = 354
R
2
.7136
Adjusted R
2
.7053
All variables are ratios of specific team’s total season statistics over each specific opponents’ total
season statistics
TwoFG = two-point field goals made; FT = free throws made; ThreeFG = three-point field goals made;
DRB = defensive rebounds; ORB = offensive rebounds over field goals missed; AST = number of assists;
STL = steals; BLK = blocks; TOV = turnovers; PF = personal fouls
As we can see in Table 2, the largest contributing factor to wins is the number of
defensive rebounds a team gets compared to their opponents. Every defensive rebound gives a
team another chance to score and does not allow the opponent a second chance. The largest
negative factor towards number of wins is the number of turnovers compared to one’s
opponents. The next two important factors to winning were number of two point field goals
made and three point field goals made. There was only one more negative factor and that was
number of personal fouls, which can be explained by the fact that a personal foul not only
usually leads to opponent making free throws but also leads to limited playing time. And for an
impactful player to have limited playing time lowers their chances at winning.
TABLE 3: Estimated values for equation (2) applied to playoff statistics
Variable
Coefficient
TwoFG
10.5314
FT
1.1225
ThreeFG
4.8997
DRB
25.7317
ORB
0.6448
AST
-1.0281
STL
-3.5152
BLK
0.902
TOV
-12.4182
PF
-4.9546
Number of observations = 192
R
2
.4910
Adjusted R
2
.4629
Table 3 displays the playoff statistics; they were not as strong of a fit as the regular wins
(the R
2
was less than .5). The statistics were all normalized by finding per minute statistics. The
worse fit can be explained by less observations and games being decided by a lot smaller
differentials on average, as teams are closer in quality to one another. The fewer amounts of
observations may explain the abnormal results regarding the ratios of assists and steals to
number of playoff wins. These turned out to be slightly negative, but the rest of the coefficients
were very similar in scale to the regular season results. A possible explanation for the negative
value on assists is that playoffs are a time where star players tend to take over the game, and
their value is really accentuated. This can result in less passing, as fans of the Lakers are aware
of, and therefore less assists. At the same time though, the teams that seem to have the
stronger star players seem to win more often than not in these close contests.
In the case of steals, the smaller number of games emphasizes any differences including
number of possessions. A good explanation for the apparent negative relationship between
playoff wins and number of steals could be that getting more steals is a result from having
fewer possessions than the opposing team. This usually means that the team has fewer chances
of scoring, and obviously fewer chances of scoring will lead to less wins.
Table 4: Accuracy of Win Model
Team
Actual Wins
Predicted Wins
Error
Atlanta Hawks
30
28.5
(1.5)
Boston Celtics
24
28.9
4.9
Charlotte Bobcats
33
30.7
(2.3)
Chicago Bulls
49
53.5
4.5
Cleveland Cavaliers
50
51.8
1.8
Dallas Mavericks
67
61.7
(5.3)
Denver Nuggets
45
43.2
(1.8)
Detroit Pistons
53
52.0
(1.0)
Golden State Warriors
42
37.6
(4.4)
Houston Rockets
52
51.9
(0.1)
Indiana Pacers
35
35.2
0.2
Los Angeles Clippers
40
36.3
(3.7)
Los Angeles Lakers
42
37.2
(4.8)
Memphis Grizzlies
22
25.9
3.9
Miami Heat
44
42.2
(1.8)
Milwaukee Bucks
28
27.6
(0.4)
Minnesota Timberwolves
32
29.8
(2.2)
New Jersey Nets
41
38.7
(2.3)
New Orleans Hornets
39
34.9
(4.1)
New York Knicks
33
31.5
(1.5)
Oklahoma City Thunder
40
40.1
0.1
Orlando Magic
35
32.8
(2.2)
Philadelphia 76ers
61
54.9
(6.1)
Phoenix Suns
32
28.1
(3.9)
Portland Trail Blazers
33
33.5
0.5
Sacramento Kings
58
62.8
4.8
San Antonio Spurs
31
31.4
0.4
Toronto Raptors
47
40.3
(6.7)
Utah Jazz
51
47.8
(3.2)
Washington Wizards
41
38.9
(2.1)
The above table demonstrates the accuracy of the estimated win model from the 2006-
2007 season. The average error in this season was 2.8 wins in absolute terms. This accuracy
suggests that the following estimations of contributed wins will be accurate estimations.
What does this all mean?
Rewind back to the last time the Los Angele Lakers won a championship with Kobe
Byrant and Shaquille O’Neal. That was the last successful run of the tumultuous duo. During
their time with the team, they would get along and then have conflicts. And of course each one
of them thought that they were more important to the team than the other. Using equation (2)
applied to both regular season and playoff statistics, we can calculate the number of wins that
each single player contributed to the team’s total.
Players 2001 - 2002
Kobe Bryant
10 wins – Regular Season Wins
4.2 wins – Playoff wins
Shaquille O’Neal
16 wins – Regular Season wins
4.6 wins – Playoff Wins
Maybe the long debate to who was more crucial to the team can come to the end with
these results. The main difference in these results seems to be the amount of defensive
rebounds grabbed by each player. O’Neal nabbed a total of 151 defensive rebounds more than
Bryant that year. This accounts for the difference in wins as this difference accounts for 6.7
wins during the regular season. Bryant fares better during the playoffs, yet still isn’t as crucial to
the team as O’Neal. This is true because defensive rebounds are relatively less important
compared to three point field goals in the playoffs.
What do these differences mean for the owners of the team? Plugging in each of these values
for O’Neal into the revenue equation we get:
Shaquille O’Neal’s MRP:
% of 2001-2002 Revenue = (.0036431)*16 ≈ 5.8%
% of 2001-2002 Revenue from Playoffs = (.0043694)*4.6 ≈ 2%
% of 2002-2003 Revenue = (.0029043)*16 ≈ 4.6%
Total MRP of O’Neal = 152 * 5.8% + 152*2% + 149* 4.6% = $ 18.71 Million dollars
14
For Kobe Bryant, doing the same calculations we get that his MRP was 11.06 Million
dollars
15
. In conclusion, O’Neal was a total of about 6 wins more valuable than Bryant with
respect to the team’s on-court performance and about 7.65 million dollars more valuable with
respect to the team’s revenues.
How is Kobe faring nowadays? Following the same calculations but with his 2009-2010
statistics, Kobe contributed about 14 wins during the regular season and 4.4 playoff wins.
Kobe’s MRP is 23.5 million dollars, assuming team revenues go up by $ 5 million dollars next
year (the amount it increased from the 2008-2009 season to the 2009-2010 season). His actual
salary is 21.26 million dollars. The salary paid closely matches Kobe’s MRP. His playoff
performance has seemed to stay pretty consistent over the years, other than a slight .2 game
increase. However, his regular season contribution has increased by 4 wins and this can be
attributed to the fact that Kobe hit three times as many three point field goals this season than
he did in the 2001-2002 season and had less personal fouls. Even with this increase, Kobe is still
not the most valuable player to the Los Angeles Lakers. Pau Gasol has contributed about 20
wins to the Lakers during the regular season and 5.8 playoff wins (making even more valuable
than Shaquille O’Neal was during the 2001-2002 season). Pau Gasol’s MRP is 33.74 million
14
Shaquille O’Neal’s actual salary during the 2001-2002 season was $ 21.42 Million dollars (salary data from USA
Today)
15
Kobe Bryant’s actual salary during the 2001-2002 season was $ 11.25 Million dollars(salary data from USA Today)
dollars. (assuming team revenues go up by $ 5 million dollars). His actual salary last year was
only 15.10 million dollars. Gasol seems to be underpaid according to these estimates. Kobe
Bryant won the Finals MVP award. Maybe Gasol actually earned the MVP award with earning
1.4 more wins during the playoffs.
Concluding Observations
How does a player’s performance on the court affect the owner’s off the court? Some
literature looked into whether or not superstars had an effect on revenues, and other literature
investigated what players were most valuable to a team with respect to team wins. This paper
goes against the approach of Hausman and Leonard (1997) and more closely follows the lead of
Berri (2004) when determining what affects revenues the most. This study puts into
consideration all team performances from current and past seasons and account for general
city and yearly effects. The second part of this paper is to connect the value of this team
performance with the statistics that a certain player obtains over the season.
The results find that ultimately the big men in the NBA rule the game. The strong effect
defensive rebounds have on wins shows that rebounders are the most important part of the
game. As each shot is missed by an opponent, it is crucial to the opposing team to get the
rebound. With each defensive rebound, the team gets a new chance at scoring points. If a team
lacks players that specialize in rebounds then not only do they miss more opportunities to
score, but they also allow their opponents more chances to score as well.
And what does this mean for revenues? It means if there were two players who were
the same quality relative to their position, that an owner would be smart to choose the center
or power forward instead of the guard. This move will gain more wins and therefore more
revenues for his team. The last finding of this paper is that for how large these contracts are for
the NBA players, they are mostly align with their marginal revenue product (if not lower than
how much they bring to the team). Even superstars like O’Neal do not get much salary premium
over their MRP if any. And younger big men like Gasol seem to be outperforming their current
salaries giving the owners a healthy margin.
To extend this research, one might want to incorporate an accurate clutch variable.
There are many games over the course of the season decided in the last minute. This makes
some of the player statistics less differentiated. This causes a problem when trying to value
certain statistics and estimate wins. An addition of a clutch variable, that recorded within the
last 3 minutes of the game how many field goals or free throws a certain player made, would
add greater value to star players or certain “clutch” players.
References
Berri, David J., Martin B. Schmidt, and Stacey L. Brook. "Stars at the Gate The Impact of Star Power on
NBA Gate Revenues." Journal of Sports Economics 5.1 (2004): 33-50. Print.
Macdonald, Don N., and Morgan O. Reynolds. "Are Baseball Players Paid Their Marginal
Products?" MANAGERIAL AND DECISION ECONOMICS 15 (1994): 443-57. Print.
Hausman, Jerry A. "Superstars in the National Basketball Association: Economic Value and
Policy."Journal of Labor Economics 15.4 (1997): 586-624. Print.
Hollinger, John. Pro Basketball Prospectus: 2003–04. Washington , DC : Brassey's, 2003.
Oliver, Dean. Basketball on Paper. Washington , DC : Brassey's, 2003.
Berri, David J., and Anthony C. Krautmann. "Shirking on the Court: Testing for the Incentive Effects of
Guaranteed Pay." Economic Inquiry 44.3 (2006): 536-46. WIley Online Library. Web. Feb.-Mar. 2011.
<http://onlinelibrary.wiley.com/doi/10.1093/ei/cbj033/full>.
Berri, David J. "Who Is ‘Most Valuable’? Measuring the Player’s Production of Wins in the National
Basketball Association." MANAGERIAL AND DECISION ECONOMICS 20 (1999): 411-27. Online Wiley
Library. Web. Feb.-Mar. 2011. <http://onlinelibrary.wiley.com/doi/10.1002/1099-
1468(199912)20:8%3C411::AID-MDE957%3E3.0.CO;2-G/pdf>.
Schmidt, Martin B., and Stacey L. Brook. "Modeling Team Wins in the NBA." The Wages of Wins: Taking
Measure of the Many Myths in Modern Sport. By David J. Berri. Stanford: Stanford UP, 2006. 93-94.
Print.
Leeds, Michael, and Peter Von Allmen. The Economics of Sports. 2nd ed. Pearson Education, 2005. Print.