2019 NFL Attendance Analysis

Here is a quick look at regular season home attendance over the 2002-2019 NFL regular seasons, using total league home game attendance data from profootballreference.

First, overall, total league home game attendance dropped in 2019 by 137,542 fans.  While the drop seems large, historically this total league home game attendance decline is not unusual - and compared to the previous season - is not a statistically significant drop in total league home game attendance.  For those interested in a graph of total league home game attendance, here it is.

Second, in terms of increases in team home game attendance from the previous season, the five biggest gainers were:

Team		Change
Washington Redskins		35679
Buffalo Bills		31018
Indianapolis Colts		15288
Cleveland Browns		13326
San Francisco 49ers		9253

In terms of decreases in team home game attendance from the previous season, the five biggest gainers were:

Team		Change
Jacksonville Jaguars		-47903
Oakland Raiders		-35042
Cincinnati Bengals		-28596
Kansas City Chiefs		-20057
Miami Dolphins		-19940

 Yet, overall team home game attendance is not different from the 2018 team home game attendance in a statistically significant manner.

2018 NFL Payroll and Performance

Yesterday, I blogged about competitive balance in the NFL for the 2018 season.  Today I will look at the relationship between payroll and performance in the NFL for the 2018 season.  The basic idea is that teams that pay higher salaries to their players will perform better.  As sports economists, we tend to be highly skeptical of this hypothesis, but it is a persistent topic - especially in baseball.  So let's see what happens.

To do this, first I need the data.  Payroll data is from spotrac and performance data is from profootball reference.   Once that is done, the first thing I did was to calculate the correlation coefficient between payroll and performance, and the correlation between payroll and performance is NEGATIVE!!  That is a first for me.  Now, one thing we know statistically is that you should not draw conclusions from descriptive statistics, but there are those that do.  If one was to draw a conclusion (again this is not statistically valid, then those who do this would have to conclude that lower pay results in better performing NFL teams).  I doubt you will hear much from the pay and performance crowd about the 2018 NFL season.

Running a regression between performance and payroll (with robust standard errors), I get that the coefficient on payroll is negative and statistically significant as well.  This is a problem for the payroll and performance hypothesis holders.  Additionally, just for the 2018 season, payroll "explains" only 17% of team performance, so this seems to be not only the wrong sign, but also the not a lot.

2018 NFL Competitive Balance

With the end of the NFL regular season, let's take a look at how competitive the NFL was this past season.  I grabbed the NFL standings data from profootballreference, and then calculated the Noll-Scully measure of competitive balance.  For those interested, here is a step-by-step guide to calculating the Noll-Scully Competitive Balance measure.

For this past season the Noll-Scully measure of competitive balance for the entire league was 1.445, and for the AFC the Noll-Scully was 1.364 and the NFC the Noll-Scully was 1.569.

Recent historical perspective on competitive balance in the NFL.

Competitive Balance in the NFL

With the NFL regular season over, let's take a look at how competitively balanced the NFL was since 1981.  As a reminder, I am using the Noll-Scully measure of competitive balance which statistically measures the actual performance from the ideal performance.  An ideal level of competitive balance using the Noll-Scully measure would equal 1.000, and levels above 1.000 means that the league has some competitive imbalance, with higher Noll-Scully numbers meaning more competitive imbalance.  (Here is a step-by-step guide if you want to calculate this on your own.)  By competitive balance I am looking at how well a league's standings are in relation to a league where wins and losses are determined randomly.  In order to do this, I am going to look at the NFL season statistically as a sample and as a population - mainly for comparison purposes.  The data comes from profootball-reference.com. 

As you can see below, whether I use a sample or the population, the Noll-Scully measure of competitive balance is very similar.  Additionally, the level of competitive balance has not changed much over the last few seasons, with the Noll-Scully for a sample in 2017 equal to 1.601 and for the population equal to 1.576.  So even with the Brown's win-less season, competitive balance has not declined much from last season.

Has the NFL Player Protests Led to Lower NFL Regular Season Home Game Attendance?

Has the NFL Player Protests Led to Lower NFL Regular Season Home Game Attendance?  During the 2017 NFL season, some NFL players knelt during the playing of the national anthem.  Vice President Pence left one game after the players knelt and President Trump also voiced concerns about the NFL players protesting during the national anthem.  Some NFL fans have proposed boycotting NFL games due to the players actions, although its impact is not clear.

Now that the regular season is over, let's take a look at the numbers and see if there is a statistically significant change in NFL regular season home game attendance.  First, the numbers.  Using the NFL attendance data from Pro Football Reference, I looked at the 2002 to 2017 NFL regular seasons.  This is a time period where the number of teams remained constant at 32 (even though some teams moved cities or stadium capacity changed).  Given the relative stability during this time period, it is easier to make season to season comparisons.  As you can see below, 2016 had the highest total regular season home game attendance, followed by a rather large decrease in 2017.  Also notice, that NFL regular season total home game attendance increases and decreases from season to season.



Let's look at that dip from 2016 to 2017.  That represents a decline of 535246 total fans for the 2017 season as compared to the leagues high water mark in 2016.  Over half of the decline is due to the Los Angeles Chargers playing in a much smaller stadium that in the past.  The Los Angeles Rams, the Cincinnati Bengals and the Washington Redskins also had sharp declines in total home game attendance.  Of the 32 teams, 22 had declines in total home game attendance, 9 had increases in attendance and 1 team (Philadelphia) had no change in total home game attendance as compared to 2016.  In percentage terms, attendance is down 3.1% from 2016 overall, and taking out the LA Chargers, total home game attendance is down 1.6%.  There is no doubt that NFL regular season total home game attendance is down from 2016.

But other seasons have seen declines in home game attendance, even without NFL player national anthem protests.  So a decline in attendance from one season to another is not unusual.  What I want to know is if this decline is unusual as compared to the ups and downs of NFL regular season team home attendance, and the way to determine that is if the team's home attendance is statistically different from last season to this season.

To do that i will use a statistical technique called a t-test.  Specifically, I am using a paired, two-tailed t-test; paired since the same teams have been in the NFL since 2002 and two-tailed since attendance can go up or down for each team from one season to the next.  A t-test is a statistical technique that compares the means of two variables and tells us if they are different from each other, and lets us know if they are significantly different as opposed to them happening by chance.  Statistical significant is the way to judge whether the results are meaningful.  In general we use an error rate less than 5% (p value <=0.05) to judge whether two variables (like NFL team home game attendance from one season to the next) are different from each other.  We choose a low error rate to avoid Type I error, which is to incorrectly conclude the existence of something that does not exist.

So what are the results?  First, from 2002 to 2016, none of the paired, two-tailed t-tests for NFL team home attendance are statistically significant.  Even the "great recession" of 2007 to 2009 did not yield a statistically significant difference in NFL team regular season home game attendance, even though we see attendance declining during that time period.  If something that large did not impact the NFL team regular season home game attendance, would a fan boycott?

Using the same paired, two-tailed t-test, I find that team regular season home game attendance was not statistically significant, and as such I cannot conclude that the NFL players protests during the national anthem lead to a decline in attendance as opposed to NFL team regular season home game attendance decreased due to random chance.

2014 NFL Income Inequality by Team

Last week I blogged about income inequality for the 2014 NFL season and showed that the NFL is more unequal in terms of income distribution than the US as a whole was in 2010 (closest year to this past season).  So I thought that I would follow up on this by looking at income inequality among the 32 NFL teams in 2014, again using the active roster data.  I am using the Gini coefficient (step-by-step guide to make this calculation) which is a measure of income inequality.

Below are the teams regular season winning percent, points scored and points surrendered, their payroll and relative payroll and lastly the team Gini coefficient for 2014.   San Diego was the least unequal in terms of income inequality and the Baltimore Ravens were the most unequal for 2014.

Not surprisingly, there is no statistically significant relationship between income inequality (Gini coefficient) and regular season winning percent.

Team	WinPct	PF	PA	Payroll	Relative Payroll	Gini
San Diego Chargers	0.563	348	348	87162147	0.8372	0.5255
Philadelphia Eagles	0.625	474	400	119105381	1.1440	0.5280
Indianapolis Colts	0.688	458	369	92940781	0.8927	0.5319
New York Jets	0.25	283	401	94454268	0.9073	0.5381
Washington Redskins	0.25	301	438	91755213	0.8813	0.5504
Cincinnati Bengals	0.656	365	344	107297822	1.0306	0.5512
Buffalo Bills	0.563	343	289	116442414	1.1185	0.5534
Detroit Lions	0.688	321	282	111226441	1.0684	0.5607
Dallas Cowboys	0.75	467	352	90952114	0.8736	0.5638
Miami Dolphins	0.5	388	373	105571107	1.0140	0.5641
Tennessee Titans	0.125	254	438	95962330	0.9217	0.5660
Jacksonville Jaguars	0.188	249	412	85737794	0.8235	0.5729
San Francisco 49ers	0.5	306	340	99854068	0.9591	0.5751
Carolina Panthers	0.469	339	374	89034415	0.8552	0.5779
Green Bay Packers	0.75	486	348	129562715	1.2445	0.5821
Chicago Bears	0.313	319	442	86877698	0.8345	0.5839
Pittsburgh Steelers	0.688	436	368	124378270	1.1947	0.5858
New England Patriots	0.75	468	313	110887895	1.0651	0.5858
Oakland Raiders	0.188	253	452	92244836	0.8860	0.5901
Houston Texans	0.563	372	307	94874026	0.9113	0.5981
Minnesota Vikings	0.438	325	343	94282411	0.9056	0.6028
Arizona Cardinals	0.688	310	299	96202750	0.9241	0.6031
St. Louis Rams	0.375	324	354	106781395	1.0257	0.6038
New York Giants	0.375	380	400	90112983	0.8656	0.6043
Kansas City Chiefs	0.563	353	281	110797441	1.0642	0.6048
Seattle Seahawks	0.75	394	254	119423389	1.1471	0.6130
New Orleans Saints	0.438	401	424	96329917	0.9253	0.6169
Denver Broncos	0.75	482	354	130851372	1.2569	0.6218
Tampa Bay Buccaneers	0.125	277	410	99527806	0.9560	0.6283
Atlanta Falcons	0.375	381	417	131809795	1.2661	0.6317
Cleveland Browns	0.438	299	337	105183505	1.0103	0.6389
Baltimore Ravens	0.625	409	302	123867781	1.1898	0.6577

Tomorrow, I hope to blog about income inequality by player position for 2014.  If not, I will get to this within the coming days.

2014 NFL Income Inequality

In July I blogged about MLS income inequality and one reader asked me about the NFL.  My response, was basically - cool idea; I don't know; have not done it.  Now I have.  I am using the Gini coefficient (step-by-step guide to make this calculation) which is a measure of income inequality.  The Gini coefficient is bound between zero (complete equality) and one (complete inequality).  For comparison purposes, the Gini coefficient for the US in 2010 is 0.411 (or 41.1 for the Gini Index, which is the Gini coefficient multiplied by 100).

Specifically, what I have done is taken every active player pay (not the amount attributed to the salary cap) listed on the website for the 2014 season and measured the amount of income inequality.  I have not included players on injured reserve or players that were given a contract but then dropped from the team.  Here is the link for Arizona in 2014, follow the dropdown menu for the other teams.  Once all the players have been included in the dataset, I have a total of 1688 NFL players in 2014.  Total payroll among those players was $3.331 billion.  For all the active players, the Gini coefficient was 0.5942, which is quite a bit more unequal than the US in 2010.

UPDATE:  (09/16/2015:  US Gini coefficient for 2014 is 0.480.)

2014 NFL Payroll and Performance

With the NFL regular season in the books, let me turn my attention to the relationship between team payrolls and team performance.  As we discuss in our book The Wages of Wins, while there tends to be a positive and statistically significant relationship between relative payroll and team performance, the relationship is not really very large enough to conclude that spending by teams is related to on-field performance.  While the book was published in 2006 (and updated in 2007), I like to try to keep up to date on these topics and thus the reason for this blog post.

So, I downloaded the NFL regular season team standings data and NFL team payroll data to run this analysis for just the 2014 season.  Before I present the results, I want to discuss the payroll data.  Using the link above you can see that I am using "cash" and not there salary cap option.  The reason is that I want to focus on actual spending by the team on the players as opposed to how money is attributed to payroll regulations.  For those interested in doing this on there own, here is a step-by-step guide to payroll and performance analysis in Microsoft Excel.

So for the 2014 NFL regular season team payroll and performance is positive and statistically significant, but the amount that payroll explains team performance is about 25%.  Meaning that there is 75% unexplained.  If I wanted to know why teams wins, using only payroll seems to be a limited resource.

NFL Competitive Balance 2014

Now that the NFL regular season is wrapped up, it is time to take a look at how competitively balanced the NFL was last season.  In order to do this, I am going to look at the NFL season statistically as a sample and as a population - mainly for comparison purposes.  Additionally, I will take a look at the conferences separately in terms of competitive balance.  The data comes from profootball-reference.com.

There are a lot of ways to measure competitive balance, and I have chosen to use a simple measure called the Noll-Scully Competitive balance measure.  By competitive balance we are looking at how well a league's standings are in relation to a league where wins and losses are determined randomly.  Here is a step-by-step guide if you want to calculate this on your own.  As a reference, here is the Noll-Scully using a sample or a population for the NFL from 1981 to 2012.

For the NFL in 2014 the Noll-Scully was 1.562 (population) and 1.587 (sample).  If you click on the link in the last sentence above, you will notice that the NFL's competitive balance for this past season is very similar to recent seasons. 

Since the 2000 NFL season, competitive balance in the NFL has been fairly stable as shown below: