The Effect Of Income Inequality In The NFL On Team Performance Author: Michael Henderson ECON 4990 Senior Seminar in Economics TR 9:3010:45 Georgia College & State University Dr. John Swinton March 12, 2015 Abstract The NFL is the highest grossing professional sports league in the world today. This is primarily due to the increased competitiveness across the league and the entertainment it provides sports fans. In my research I examine the relationship between the unequal distribution of player salaries in the NFL and team performance from the years 2006 to 2009. In this study I analyze the salaries of the top 10 highest paid players per team for each season. Franchises that distribute the majority of their salary cap to highly talented players, known as the superstareffect, leave little cap space to distribute the rest of the salaries to the remaining 53 man roster. Though my research does not show a statistically significant correlation between player salaries and team performance, there were many significant findings that do in fact have an effect on a team's ability to win.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
The Effect Of Income Inequality In The NFL On Team Performance
Author: Michael Henderson ECON 4990 Senior Seminar in Economics
TR 9:3010:45 Georgia College & State University
Dr. John Swinton March 12, 2015
Abstract
The NFL is the highest grossing professional sports league in the world today. This is
primarily due to the increased competitiveness across the league and the entertainment it
provides sports fans. In my research I examine the relationship between the unequal distribution
of player salaries in the NFL and team performance from the years 2006 to 2009. In this study I
analyze the salaries of the top 10 highest paid players per team for each season. Franchises that
distribute the majority of their salary cap to highly talented players, known as the
superstareffect, leave little cap space to distribute the rest of the salaries to the remaining 53
man roster. Though my research does not show a statistically significant correlation between
player salaries and team performance, there were many significant findings that do in fact have
an effect on a team's ability to win.
Does Income Inequality In The NFL Affect Team Performance?
I. Introduction
The relationship between an individual’s pay and their performance has always been a
major issue in economic research. When studying sports economics, specifically in the NFL,
research tends to utilize wins as a measure of efficiency. For this reason, viable and accessible
data sources measuring team performance were readily available. NFL player salary data is
abundant but is bounded by the last five years. The most recent salary data attainable ends after
the 2009 season.
The relationship between player salaries and their performance on the field is a major
issue when trying to organize a winning combination of players that agree to their wide variation
in contracts. Superstar players such as, Peyton Manning or Tom Brady are extremely talented,
and for this reason they sign multimillion dollar contracts way beyond that of their average
teammate. Rosen (1981) finds that small differences in talent have a large effect on income
inequality. He labels this phenomenon as the “superstar effect”. This occurs when team
managers distribute player salaries in favor of a set number of highly talented players in hopes
that they can utilize the abilities of the rest of the team in order to win the most games. The
problem pertaining to the question at hand deals with the large differences in pay between
players on the same team. Borghesi (2008) did a study and found that when player salary
inequality is low, individual player proficiency tended to be higher. This implies that franchises
that distribute more income to superstars, rather than evenly across the entire team, perform
worse on average. This could be caused by dissatisfaction among lower paid teammates. Lazear
(1989) in his work finds that when pay is distributed relatively evenly among employees in the
1
workforce, cooperation increases and firm efficiency improves. In the NFL, player pay is not
distributed evenly, so how could this affect their efficiency?
The NFL has the shortest season amongst all other major professional leagues in the
United States. Regardless, all teams of the National Football League have a combined brand
value of $9.1 billion with an average team profit of $53 million as of the 2013 season. Thats a
brand value of over $4.7 billion more than the MLB and $6.3 billion more than the NBA. Refer
to Figure 1. (pg.19). Over the past four years, the average player salary has held steady at around
$2.1 million. The efficient allocation of player salaries is in question. Matt Ryan, quarterback for
the Atlanta Falcons, is today’s highest paid player with a five year contract signed in June of
2013 worth $103.75 million. That’s an average annual salary of $20.75 million, $18 million
above the average player. Matthew Stafford, quarterback of the Detroit Lions, ranks second in
earnings with a three year $53 million contract extension that included a $27.5 million rookie
signing bonus. These are two of the highest paid players in the NFL today, both of which play
for teams that fail to consistently have winning seasons. Team performance may start to suffer
when income distribution is so unequally distributed. For example, Player X is only getting paid
$400,000 each season. If his teammate, Player Y, is earning $25,000,000 each season, then
Player X may feel he’s getting paid unfairly compared to his fellow teammate. If he feels as if
he’s getting paid unfairly, this could affect his performance during a game. Because he is not
performing at his expected level, his team may lose a game that should have been won, which
causes team revenue to decrease. This is devastating to any team because football is a game that
requires wins in order to continue to be a successful franchise. Every loss on the field is also a
loss in profit for the franchise, let alone the entire league. In short, such dramatic pay inequality
2
could affect team performance as a whole and therefore affect a team’s ability to maximize
profit.
II. History of the Salary Cap
Another key factor that greatly affects each team’s salary distribution is the salary cap.
While other leagues, like the MLB and the NBA, have similar salary constraints, the NFL has a
much more strict salary restriction. In the NFL, players are paid based on talent and how they use
that talent to win games. After salaries are divided between the best talent, the remaining cap
space is divvied up between the players that can assist superstars in winning.
Beginning in 1994, the NFL implemented a hard salary cap that resulted from a collective
bargaining agreement between the NFL and its players in 1993. This is a binding agreement that
places a limit on the amount of money each team can spend on player salaries per season in order
to level the playing field amongst teams across the league. Before a salary cap was ever put in
place, franchises that were still developing, especially in less populated cities, didn’t have a
substantial fan base. For this reason, certain teams couldn’t earn the kind of ‘deep pocket’ money
needed to buy many highly talented players. Without equally distributed talent throughout the
NFL, these teams were unable to compete with other wealthy teams. More established franchises
had a bigger fan base and higher earnings. This gave them the capital that allowed them to
purchase an arsenal of highly talented players that less popular teams simply could not afford.
Whenever these two types of teams came to play, fans would experience more blowouts than you
see today and this put a constraint on the level of profit the NFL could potentially earn. A salary
cap was established specifically to prevent that small group of powerhouse teams from
dominating the league year after year.
3
A salary cap is calculated by a Collective Bargaining Agreement (CBA) set by team
owners based on the average projected amount of revenue earned for football related income and
benefits for the upcoming season.
SALARY CAP=(PROJECTED REV * CBA%) LEAGUEWIDE BENEFITS NUMBER OF NFL TEAMS (32)
The CBA set the percentage of NFL revenue that can be used for player salaries at 48 percent,
which would remain through the 20122014 seasons. For the 2013 season, the salary cap was set
at $123 million per team. Each team is allotted 53 roster positions to distribute this cap,
excluding coaches salaries. Therefore, on average, each player should make around $2.3 million
if salaries are distributed equally. With a salary cap put in place, talent is distributed more evenly
across all teams. This leads to more competitive games and more competitive seasons. With
better competition comes closer and more entertaining games. This increases individual team
popularity across the NFL and increases popularity for the league as a whole. When an entire
league sees an enormous inflow of support from new fans, they see an enormous increase in
profit. Profit is the only element that matters in the NFL, hence why this is a subject worth
researching. The careful distribution of a team’s salary cap to its players is directly correlated
with a team's ability to thrive as a franchise.
A cap rollover is any extra money that can be added to a team’s salary cap that was not
used in the previous year. The salary floor is the minimum amount of the salary cap teams must
spend on players each season. The required amount of the cap that each team must spend was
lowered from 95 percent in 2012 to 89 percent in 2013. For strategic teams that want to allocate
funds to the future, they can use this cap rollover to their advantage for the next season. The
4
question is whether or not using up a majority of the cap on only a few key players gives a team
a better chance of winning than does splitting the cap more evenly amongst all the players.
III. Literature Review
The salary cap puts a restraint on how much teams can spend and forces managers to
search for the best combination of players that will be most efficient, while also attempting to
keep player salaries relatively fair amongst their players. Larsen, Fenn, and Spenner (2006) did a
study showing this effect. They showed the salary caps effect on team spending and found that
teams’ cap spending from the years 2000 to 2002 was negatively correlated with their spending
from 2004 to 2005. This result shows that the salary cap truly is effective in reducing teams from
constantly spending more than other teams year after year. Hence, the salary cap does make
competition throughout the league more balanced due to a more even pay scale and a more even
distribution of talent for all teams.
The salary cap has been proven to work in making the NFL a more competitive league
than in the past, but how has it affected decisions from a managerial standpoint? Kowalewski
and Leeds (1999) performed a study using gini coefficients. Gini coefficients measure statistical
dispersion, usually measuring income distribution. It’s measured on a scale from 0 to 1. The
further away the gini coefficient is from 0, or perfect equality, the more inequality there is. They
measure the variance in player salaries from 1992 to 1994. They used data from the season right
before the salary cap as well as the first season the salary cap was implemented. They found that
the salary cap created a less equal distribution of salaries per player. The Gini coefficient rose
from 0.393 in 1992 to 0.479 in 1993. This shows a significant increase in the inequality of player
contracts. They also found that superstar players were paid higher salaries on average after the
5
salary cap than before. This increase in pay for more talented players came at the expense of
lower draft picks. Higher draft picks were signing bigger contracts while lower draft picks were
signing smaller contracts than in previous years.
When measuring talent in the NFL, there is only a minimal difference between the
highest paid players and the lowest paid players. Each player has developed the skills needed to
compete against the rest of the most renowned players across the country. Quinn, Geyer, and
Berkovitz (2007) measured the relationship between income distribution and talent. In their
study they analyzed every NFL franchise’s budget from the years 2000 to 2005. They found that
small differences in talent resulted in large pay inequality because even small differences in
talent can have a significant effect on a team’s ability to win. Furthermore, they found that teams
that had a winning percentage above the league average had spent more of their cap space on
players that were their 13th through 30th picks and had spent less on their 35th through 53rd
picks than they had in the past. However, they were unable to significantly conclude that income
distribution and winning percentage were correlated.
IV. Data and Methodology
The purpose of my study is to find if, and to what degree, the managerial allocation of a
team’s salary cap could affect a team’s efficiency in regular season games. If teams choose to
purchase more superstar players in order to be efficient, than the amount of cap space left for
support players may start to be extremely skewed towards the top of the roster. This inequality in
pay could result in lessened team cooperation with fellow teammates and coaches that could
essentially alter their overall performance throughout a season. If teams choose to purchase less
superstar players, they now have more cap space to distribute salaries more evenly across their
6
roster and are able to purchase slightly more talented support players. This more evenly
distributed income in the NFL could have a significant affect on how that team performs in the
coming season.
For a franchise to be deemed efficient, player inputs (salaries) must be transformed into
productive outputs (wins). The value of a player is shown through their effectiveness on the field
and the revenue they bring in. Offensive player salaries and defensive player salaries are two
different input costs. Output on offense is determined by their ability to score. This is a measure
of points for (PF). Outputs on defense are determined by their ability to prevent scoring. This is a
measure of points allowed (PA). The overall team output is a win or a loss, more precisely
measured by margin of victory. This is the NFL’s measure of team performance. A team with a
high PF and a low PA tends to win the most games. A team that is inefficient will produce less
wins compared to a franchise that produces a higher PF and a lower PA on average using lower
offensive and defensive salaries.
I collected 1280 observations that measure player compensation, team productivity and
team earnings. Player compensation is measured using player salaries compared to the overall
team salary cap. Team productivity is measured using team performance statistics. These
statistics include offensive points for and defensive points allowed for each team during the
regular season. Other measures of performance include offensive and defensive quality based off
each team’s total points for and points allowed compared to the league average.
Using USA Today’s online database, I recorded player salaries for the top 10 highest paid
players for each team for the 2006 to 2009 seasons. Because there was no other NFL salary
database beyond 2009, I was forced to limit the number of observed seasons to four. From this
7
database I was able to collect every NFL team’s overall salary cap per season and their per
player salary cap each season. In my sample, each top 10 player made $4.78 million on average
per season while each team as a whole had an average salary cap set at $97.37 million. I
manually calculated the percent of each teams total salary cap spent on each of their top 10
players to observe the significance of income inequality across players per team. Each of top 10
players in this sample received 6.2% of their team’s total salary cap. That means each team was
using 62% of their entire cap space on 10 players while only leaving 40% to distribute between
43 other players. I used this measure to clearly see the significant pay inequality occurring in the
NFL.
My performance data was collected from www.profootballreference.com which
originates from www.sportsreference.com and works with USA Today to provide the most
accurate uptodate statistical figures. From here I was able to collect my dependent variable of
team performance measured by margin of victory. Average leaguewide team performance
appeared as 0.004. This number is very close to zero because teams must lose by the same
amount their opponent wins by so all teams have a combined average margin of victory equal to
zero. This measure of performance differentiates itself from the normal measure used in recent
studies. Many researchers in previous studies were using number of wins rather than how much
teams were winning by. With a different measure of performance I was able to observe different
results. Other variables I was able to collect through this database included each team’s number
of wins per season, points for, points allowed, strength of schedule, and offensive and defensive
quality. Because teams give up the same amount another team scores, the number of points for
and points against across the entire league was the same at 343 points. Overall, teams had less
The dependent variable TEAM_PERF is a measure of margin of victory calculated using average
per team point differential. My key independent variable CAP_SPENTPP is the percent of a
team’s total cap space spent per player for each team. This puts a percent measure on player
salaries to show the precise value of each player compared to total team cap space. This helps to
show the true significance of player income inequality. PLAYER_CAP is a team’s allotted dollar
value from the total team cap that is preassigned to each player during the off season. This
amount is not the true salary of each player because teams can choose to spend more or less than
their preset salary limit based on circumstantial decisions once the season begins. Operating
income is represented by OP_INCOME to measure profit before interest and taxes. The variable
REVENUE measures each franchise’s total profit at the end of each season. VAL_CHANGE
represents the percent of a team’s change in total franchise value from one year to the next.
Using this variable I am able to see the percent increase or decrease of individual franchise
values compared to other teams across the league and its relationship to team performance for
that season.
11
I used multiple dummy variables that included COACH, CONFERENCE, and
POSITION. Any first year coach for a team that season was labeled as 1 and for any coach that
had been part of that team for at least one year or more was labeled as 0. Because there are only
two conferences in the NFL, the AFC received a value of 1 while the NFC received a value of 0.
Players were labeled based on what side of the ball they played on. Special teams players receive
a value of 0, offensive players receive a value of 1, and defensive players receive a value of 2. If
you refer to Table 1 (pg.17). you can see that the mean position is 1.68 signifying that a majority
of the top 10 highest paid players were defensive players. With this dummy variable I was able
to capture the effect playing on different sides of the ball had on team performance.
The variable TEAM was given a value of 1 through 32 for the number of teams in the
league while PLAYER was valued 1 through 1280 based on the number of player observations.
There were four seasons recorded in this sample accounted for using the variable YEAR.
OFF_QUALITY and DEF_QUALITY as well as PF and PA describe perteam offensive and
defensive performance and their ability to score as well as prevent scoring. This is used in
determining an accurate measure of the strength of a team’s opponent labeled as
SCHED_STRENGTH. My last two variables, WINS and TIES are a more broad measure of
performance. Including ties throughout a season shows what affect that gap between winning and
losing would have on overall team performance.
Though my data includes many observations and key variables, there were other lurking
variables I was unable to account for. Race is a key variable that may have a small effect on team
performance. Dufur and Feinberg (2008) through field research and semistructured interviews
found that minority players in the NFL experience discrimination during the hiring process.
12
While minority and white players described much of their labor experience as similar, minority
athletes identify far more negative repercussions.
VI. Results
Refer to Table 2. (pg.18) for regression results. There was not a statistically significant
correlation between the percent of a team’s total cap spent per player and team performance in
my study. Overall, this study was effective in showing significant effects from other key
variables on team performance. Nine variables showed to be significant at the 99% confidence
level while three variables were significant at the 95% confidence level. My regression had an
adjusted R2 of 0.99 meaning that my independent variables explain 99% of the variation in team
performance.
The overall team cap space (OVERALL_CAP) and operating income (OP_INCOME)
were negatively correlated with team performance at the 99% confidence level. An additional
$1,000,000 in total cap space is correlated with a 0.001 point reduction in team performance
while an additional $1,000,000 earned in operating income was correlated with a 0.0003 point
reduction in team performance that season. This seems strange at first, but teams that have more
cap space may purchase more offensive or defensive players that year which could weaken the
overall ability of the opposite side of the ball. Moreover, teams that forecast a bad season may
put that excess cap space aside to use for the next season which would hurt them in the current
season. Operating income consists of advertising and merchandise sales, so if certain superstar
players are being heavily advertised before a game or throughout a season, that may cause them
to attempt perform at a level unattainable for them in order to please the fans. This may in turn
cause them to make irrational decisions that could potentially lose them a game that should
13
otherwise have been won. Year (YEAR) was also significant at the 99% confidence level. For
every new season, team performance increased by 0.008 points. New talent is constantly being
cycled through the NFL. Weaker teams draft younger more talented players than previous years
in order to stay competitive. As weaker teams become stronger, overall competitiveness
increases and team performance goes up. Offensive quality (OFF_QUALITY), defensive quality
(DEF_QUALITY), points for (PF), points against (PA), and strength of schedule
(SCHED_STRENGTH) were all significant at the 99% confidence level. This is most likely due
to the fact that team performance is based on margin of victory derived points for and points
against. For every one point increase in offensive and defensive quality, team performance
increased by 0.89 points. These numbers are very close to the same because, as stated earlier,
these are calculated based on points for and points against compared to the league average and a
team must give up as many points as another team scores. Similarly, for every 1 point increase in
points scored, team performance increased by 0.007 points and for every 1 point decrease in
points allowed, team performance went down by 0.007 points. Strength of schedule was
measured using each team opponent’s average offensive quality plus defensive quality. For every
1 point increase in a team's schedule strength, overall team performance decreased by 0.89
points. These variables are highly correlated because they are calculated based off of overall
team performance rather than just offensive or just defensive performance.
The percent change in a team’s overall value from one season to the next
(VAL_CHANGE) was significant at the 95% confidence level. As team value increased by 1%,
team performance also increased by 0.0007 points. Team value may increase due to gained
popularity for a franchise. This newly gained popularity may help motivate players to perform
14
slightly better than in recent seasons. Whether or not you were a new coach for a team (COACH)
was inversely related to team performance. If a team gained a newly hired coach that season,
team performance decreased by 0.008 points. New coaches may need a season or two to adjust to
new players and a new staff, so this makes sense that a team’s performance would suffer.
Position (POSITION) was not significantly correlated with team performance. This was
surprising, because my summary statistics showed that out of the top 10 highest paid players,
teams hired a majority of defensive players. I would have expected this overcompensation on
defense to have a significant effect on team performance.
VII. Conclusion
The salary cap puts a constraint on manager choices for player employment. I attempt to
find how these different compensation strategies affect team performance. Through my research,
I found that team performance is not significantly affected by the percent of a team’s cap space
spent on the top 10 highest paid players. This shows that the superstar approach may be an
effective strategy in the NFL. Although my main question was insignificant, I found that teams
who gain substantial amounts of cap space tend to perform worse than in previous seasons. With
new money, teams either go after fresh players out of college, or veterans in free agency. The
Associated Press from New York Times wrote, “there are always reasons these players are
available in the first place, be it consistency, character, chemistry or simply cost”. Teams recruit
past superstars that may be too old or arrogant to work well with a new team.
After coaches are fired, my study shows that teams with newly hired coaches tend to be
less effective. Lack of communication, understanding, and mutual respect between veteran
players and new coaches can inhibit a team’s ability to consistently play well together. Teams
15
may be more efficient when hiring coaches that are currently and have been part of the franchise,
such as an offensive or defensive coordinator, in order to prevent coaches from having to enter a
completely new team environment.
With a greater sample size and additional variables, future research may show to be
significant when relating player salaries to team performance. Variables such as race, weather,
and number of years in the league are key variables I was unable to collect. Later studies may
gather all needed variables to arrive at a more accurate conclusion. The perfect allocation of team
resources is still in question, but further analysis may help determine the best possible approach
when hiring players for the upcoming seasons.
16
Table 1. Summary Statistics (Obs 1280)
Variable Mean Std. Dev. Min Max TEAM_PERF CAP_SPENTPP OVERALL_CAP PLAYER_CAP OP_INCOME REVENUE VAL_CHANGE TEAM YEAR COACH* CONFERENCE* PLAYER* POSITION OFF_QUALITY DEF_QUALITY PF PA SCHED_STRENGTH WINS TIES