Implications of the 2011 CBA on NFL Rookie Salary Potential

McKenna Moore

Pace University


 The purpose of this paper will be to explore the estimated relationship between the introduction of the new CBA in 2011 and the role a team’s salary cap has on their ability to garner high-paying contracts for new players when they first enter the league.

The 2021-2022 NFL season averaged approximately 16.7 million views per game with the 2022 Superbowl raking in 99.18 million views and the overall season bringing in over $18 billion in revenue (Moore 2019; Super Bowl Ratings). In 2021, each team was given a salary cap of just over $208 million with 48% of that going to the players and the remaining 52% distributed amongst the owners of each team; the coaching staff is excluded (Gough 2023). Before the 2010-2011 season, the various caps were distributed 59.6% amongst the owners of the teams and 31.4% amongst the players (Rosenthal 2010). This distribution was a leading cause of the 2010-2011 postseason NFL holdout that included an inability to use practice facilities, team doctors, and an inability to sign free agents during this time. 

The impact of this salary renegotiation is best exemplified in the difference in salary of both first-round quarterback draft picks from 2009 — the last year pre-lockout with a salary cap — and 2011, the first year with the salary cap (CBA 2011). In 2009, Mark Sanchez from a Division 1A college was drafted by the New York Jets with a guaranteed five-year contract worth a total of $50.5 million. In 2011, Cam Newton, the first overall draft pick also from a D1 school, was drafted by the Arizona Cardinals with a guaranteed four-year contract worth a total of $22.02 million, a 30% decrease for an otherwise equally skilled player (Spotrac). The seemingly equitable characteristics of the players, yet clear distinction in compensation, are puzzling. The purpose of this study is to understand the relationship the new CBA has on the total compensation of first-year candidates, with a further focus on the role their collegiate athletic division has on their overall compensation. Previous studies have focused on pre-NFL performance predictors and their salary association, namely using the annual NFL combine as a measure of player potential (Kurmits 2008), and post-draft success analyzing whether teams have spent their money to get formidable results (Foltice et. al 2021), as measured by overall team performance. 

Several methods were employed to test the underlying hypothesis. A Feasible Generalized Least Squares (FGLS) model was used to look at the most influential factors on a rookie’s total contract value. An FGLS model is used to control for heteroskedasticity that makes the residuals and coefficients of a traditional Ordinary Least Squares model unreliable. This FGLS model produces the most reliable results with the presence of the CBA Binary, indicating a player was drafted between the years 2011 and 2017, with the presence of the CBA resulting in a 0.127% increase in total salary.

The rest of this paper is organized thusly: a brief literature review followed by a consideration of factors that may potentially influence signing salary. The data and methodology are presented, followed by a discussion of the results. The paper concludes with a holistic analysis of contributing factors to the rookie salaries of entering players up until 2017.

Literature Review

Much of this discussion will consist of changes during the 2011 off-season with the renegotiation of the collective bargaining agreement. This was the first time the CBA had been renegotiated since its enactment in 1993. From 1993 to 2010, there was no enforced contract length restriction, so the average contract tended to be five years long and teams were unable to have less active roster salary cap availability. Active roster salary cap availability speaks to how much a team can pay all players deemed to be active during that season. The CBA was extended until its expiration in May 2010. Subsequently, the 2010-2011 season fell under no CBA jurisdiction, so the most significant rookie salary to date, Sam Bradford came out of this draft with a total contract value of $78 million (Fitzgerald 2015). The literature emphasizes the role the 2010 CBA had in the composition and flexibility of an actual contract drafted by a team before the yearly draft and the restructuring of rookie negotiation abilities and overall team salary caps (Raj 2023).  

Draft selection is driven by what teams need going into any given season. The top players, indicated by their success in college, will be selected in the first round but the coveted first pick is determined by what position the first team drafting needs most. If the team who is drafting first, determined by who had the worst record the year prior, needs a quarterback, then the best available quarterback will be drafted first. 

The draft itself acts in the sense of a prisoner’s dilemma where both the teams and rookie candidates in the draft are trying to find their optimal outcome. Teams are trying to maximize their success in the coming season by getting the player that they predict will have the most success while paying not necessarily the lowest amount, but an amount that is fair enough for them to still have leftover funding to get more players. Players are looking to enter the league on the strongest team with a decent salary, so they have less negotiating power. The Pareto optimal output would arise when both the team and player come to an agreement where players have a contract with their desired bonuses, general salary, and contract length, and the teams have not exceeded their active roster salary cap and can still afford multiple excellent players.

Within each team’s salary cap, about half of the total cap goes to rookie negotiations for that season (Fitzgerald 2015, Langland). A general salary comprises a base salary, guaranteed for injury compensation, skill guarantees, cap guarantees, and other generalized bonuses, including workout bonuses and roster bonuses (Christopherson 2020) A skill guarantee is a generalized clause that prevents players from being financially penalized due to a decrease in their technical skills while their contract is in place. A cap guarantee is an additional clause that prevents players from being paid due to a team’s concern that they may exceed their salary cap. Workout and roster bonuses are contingent upon attending a specified percentage of pre-season team workouts and being officially listed on a team’s 53-man roster by a selected date. The guarantees are presented as the base salary, the bonuses are summarized in the signing bonus, and the sum of both categories is given as the total contract value (Christopherson 2020).

The structure of bonuses and guaranteed contract value was targeted through a policy change called the 25% rule during the 2011 renegotiation. Before 2011, the 25% rule stated that a rookie salary comprises a yearly base salary plus 25% of the guaranteed signing bonus. Further, each following year cannot exceed a 25% increase of the previous rookie salary cap (BLS 1993). This rule was easily worked around by including additional media and appearance bonuses before 2011, but with the new CBA renegotiation rules and regulations, became much stricter in enforcing the 25% rule, and teams who did not follow this procedure were penalized by draft pick deductions (CBA 2011; Kurmits 2008). 

Contract lengths were also a key point of conflict during the negotiation. Before 2011, rookie contracts averaged five to six years, which tended to harm teams in the long run. Players drafted with very high salaries in the first round tend to underperform when they enter the league (Christopherson 2020), but teams would have all of their funds tied up in these players so they could not afford to rebuild a stronger team. To prevent the first-round “loser effect,” a strict four-year limitation was enacted for anyone in the draft (CBA 2011). The exceptions to this rule came in two forms: 1. First-round draft picks were allowed a fifth-year extension at the team’s discretion, but the teams had to redistribute their funds to stay within their yearly cap, and 2. Undrafted free agents were restricted to a three-year maximum if they were picked up by a team (Christopherson 2020).

These changes have been instrumental in a salary average and contract length change, which is most notable when looking at the summary statistics with a before-2011 and after-2011 comparison.

Pre-Estimation Discussion

The data presented in this research comprises the salary statistics for every rookie player’s draft contract from 2005 to 2017. Additional variables include each team’s salary cap for each year in the sample, the position each candidate plays, the year a candidate was drafted, their age, the length of the initial contract, and the division each candidate played in college. Candidates who attend a D1A school are more likely to be drafted due to the assumption that D1A schools have the best players, and these schools traditionally win the College Football Championship.  This data was collected from ESPN and, a salary and contract database for every US sport. Figure 1 provides a visual summary of the college division representation and presentation of each position in the draft during the allotted time frame.

The position a candidate is associated with is imperative in their actual pick number relative to their draft position. Positions invaluable to teams, such as wide receivers, quarterbacks, and running backs, are consistently in high demand by teams. Hence, they are more likely to be drafted early in the draft and with highly enticing salaries. 

An additional factor considered as a holistic indicator of potential salary is the college division each candidate played. The D1A and D1AA divisions are the most competitive athletic divisions, and the college championship game is typically played amongst schools in these divisions. 

To further the salary discussion, it is essential to note that a significant component of how much a player gets paid is primarily based on how much a team can spend on its active roster, known as the active roster cap. Figure 2 depicts the trend of the average team salary cap for their active roster from 2005-2017.

Since 2005, there has been a continuous positive, upward trend in the average team salary cap with a couple of notable distinctions. At the end of the 2009-2010 season, the 1993 CBA expired after multiple extensions. The expiration of this CBA resulted in a year of tense negotiations, and, as a result, the 2010-2011 season had no enacted CBA, and there was no formal agreement between the NFL and the players union about how much each team could spend. After implementing the new salary guidelines, teams decreased their spending for two seasons (from 2011 to 2013) to abide by the new regulations. Once this correction was made, there has been a consistent upward trend.

Another aspect that could influence a contract’s total value is the actual length of the total contract. The pre-2011 CBA had no hard limit on the length of a contract, so the average contract length was between five and six years and a candidate’s signing salary was distributed throughout the specified amount of time. The new CBA placed a hard limit on the total length of a contract with all drafted players being allowed to sign a four-year contract, with an optional fifth-year extension presented by individual teams to first-round draft picks only. 

The 2011 CBA was enacted on July 27. The most prominent clause regarding salary stated that 60% of the total team salary was now devoted to players instead of the team owners, a 12% increase from the 1993 CBA. This resulted in much higher salary caps in the long term as depicted in Figure 2. Table 1 provides the summary statistics for all variables of interest and control variables that are being applied in the model. Based on the notion that a team’s total salary can be spent on a player has generally increased following the 2011 CBA, it can be assumed that this introduction and time variable positively affected a presented draft salary as new candidates entered the NFL.


 A standard OLS model tends to fall subject to heteroskedasticity because the data fails to meet the assumption that it has a constant variance, and subsequently, the residuals are calculated with a constant variance assumption. This produces biased standard errors as well as biased variable coefficients. In this model, heteroskedasticity stems from the individual candidates because there is a lack of consistency in the individual salaries over time which in turn produces a lack of constant variance. 

Heteroskedasticity can stem from variations in team salary cap over time, as well as variations in position and round drafted. To confirm that heteroskedasticity was present in the model, a Breusch-Pagan/Cook-Weisberg Test was performed to reject the null hypothesis, with a null stating there is homoscedasticity, or the residuals do not vary. A feasible generalized least squares (FGLS) model is preferred to control for heteroskedasticity when the source cannot be identified to reduce potential heteroskedasticity. 

The presented functional form is a double-logarithmic model with the dependent variable being the log of Total Contract Value and the primary variables of interest being the log of Team Cap and the presence of the 2011 CBA. Added controls include the round a candidate was drafted, the position a candidate plays, the year they were drafted, the length of their contract, their age, and the division the candidate’s college is associated with. The final two models also include an interaction between the CBA and the round a candidate was drafted in. 

The round a candidate was drafted in is significant with the historical knowledge that first-round draft picks get paid more and have the option to extend their contract an additional year under the advice of the team they sign with. Position is included to see if there is any statistical backing to the assumption that offensive positions, such as quarterbacks and wide receivers, get paid more than other positions. 

The age a candidate was drafted is included due to the inference that older players will not play as well and be paid less. This payment assumption is proved by the parabolic relationship between age and total contract value, which is why age is then squared in the model. However, it would be intriguing to see if the age a player was initially drafted affected their contract value.

The contract length is added as a control, as this was a key point of discussion when introducing the new CBA, seeing as the policy placed a hard limit of four years on contract length and previously this limit did not exist. College division is included to identify if players from D1A universities — who are typically more desired due to their collegiate success — get a monetary return on their college investment in the form of higher salaries. 

CBA and draft rounds interact to look at the monetary effect the new CBA had on players that were specifically drafted in the first round, seeing as how the new CBA specifically targeted first-round draft picks in their salary limitations and contract length restrictions. Summary statistics for numeric variables are presented in Table 2.

The following model represents the empirical approach:

Reference Model: 

Log_TotalValue = β1(log_TeamCap)i + β2(CBA)i + μ                   (1)

The key independent value is log_TotalValue which is the log of a candidate’s Total Contract Value. Log_TeamCap and CBABinary are serving as our two variables of interest. 

Controls model with RSE: 

Log_TotalValue = β1(log_TeamCap)i + β2(CBABinary)i x (RoundDrafted)i +β3(Role)i + β4(Year Drafted)iβ5(Age) + β6(Length) + β7(Division)i + μ                   (2)

Where, RoundDrafted, Role, YearDrafted, Age, Length, and Division are controls across all models and μ represents the error term. An additional control was implemented for heteroskedasticity by applying robust standard errors. All other terms are held constant.

FLGS Model: 

Log_TotalValue = β1(log_TeamCap)i + β2(CBABinary)i x (RoundDrafted)i +β3(Role)i + β4(Year Drafted)iβ5(Age) + β6(Length) + β7(Division)i + μ                               (3)

Heteroskedasticity is controlled for by performing the squared inverse of a standard regression between the predicted standard baseline model and the absolute value of the standard errors associated with that model. This model is called a Feasible Generalized Least Squares (FGLS) model and this controls for unknown heteroskedasticity when a source cannot be pinpointed amongst the presented variables.

Results and Discussion

The results from all regressions (baseline regression, OLS with robust standard errors FGLS) are reported in Table 3 with a data visualization in Figure 3. As noted in these regressions, the log_TeamCap, the percentage change in a team’s salary cap yearly, and the presence of the 2011 CBA both are statistically significant at the 0.99% confidence level. 

After performing an OLS estimation with robust standard errors, a Breusch-Pagan/Cook-Weisberg Test, as well as a White Test, were conducted to identify if heteroskedasticity was present in the OLS control model. Once the tests were conducted, a feasible generalized least squares model was implemented by calculating the weights of the standard errors from the OLS model residuals to create homoscedastic standard errors.

Determined by FGLS regression, the presence of the CBA Binary, a player being drafted between 2011 and 2017 has a 0.127% increase in their total contract value, consistent with the previous studies of the trend in total team salary cap. A t-statistic of 12.401 for the FGLS model, compared to the 1.962 t-statistic from the OLS model, shows that the effect of the CBA greatly increased in the FGLS model, which further validates the effect that the CBA has on total contract value. 

The t-statistics for all the Round Drafted indicators also increased with the FGLS model. Being drafted in round 2 using the OLS model produced a coefficient of -1.542 with a t-statistic of -61.278. The FGLS model produced a coefficient of -1.485 and a t-statistic of -244.487. This indicates that being drafted in round 2 decreases salary by 1.485% and the higher t-statistic with the FGLS model increases the magnitude of this statistic.

The interaction between the presence of the CBA and the round a player was drafted explains the additional marginal effect of being drafted in any round that is not the first round after the implementation of the new CBA. Using the FGLS model, being drafted post-2011 in round 2 increased a player’s salary by an additional 0.621% relative to the 1.485% decrease that just being drafted in round 2 has on salary. This indicates that the overall salary effect of someone being drafted in round 2 post-2011 is a 0.864% decrease in salary. 

Decreasing the negative effect that not being drafted in the first round has on salary following the introduction of the CBA further emphasizes that the implementation of the new salary guideline had a positive effect on all player salaries no matter which round you were drafted in. 

All models showed positive effects on the CBA variable but the controls and FGLS model have a negative coefficient associated with the log_TeamCap variable that contradicted the study’s expectations.

Using this final FGLS regression, predictions can be made about any candidate entering the NFL draft so long as the team’s active roster salary cap is available. A player looking to maximize their earning potential when entering the draft would want to be drafted after the CBA implementation and ideally, these players would be drafted in the first round with a fifth-year extension available. 

The values of these coefficients could be overestimated due to omitted variable bias. There are a multitude of factors that go into deciding a contract’s total value. As team expectations and procedures evolve throughout the NFL, policy changes will likely influence signing bonuses and yearly salary caps, two significant subcomponents of a contract’s total value. 


Although there has been a significant number of studies about players’ success or lack thereof once they enter the NFL and how this success subsequently affects their salary, there has been no study about their initial salary and what factors have the most significant impact on this initial salary. This paper uses web-scraped data from and ESPN to analyze what factors could most significantly impact a candidate’s signing salary while examining the impact of the 2011 CBA implementation on player starting salaries.

The FGLS model with residual control of the standard OLS output provided the most reliable results due to heteroskedastic standard errors which showed that the presence of the 2011 CBA has a 0.127% increase in a player’s total contract value when they first enter the league. This CBA impact further had the most significant impact on candidates drafted in the first round. This result was statistically significant at the 99% level. This is because the CBA limits the length of the contract for all players which, in turn, increases yearly compensation and the CBA decreases the negative effect of not being drafted in the first round.

Considering the limited public information that is provided as to how each NFL team decides the value of a contract, further information is needed to draw more conclusive results about what impact each new CBA has on initial contract values as well as the precise role that yearly changes in a team’s active salary cap as to how much they can and are willing to spend on rookie candidates as the draft approaches. 

Further analysis can be done on the changes made during the 2020 CBA renegotiation to identify the effect of updated salary guidelines. Additional studies can also be done to track players’ statistics in the NFL about the expansion of their salary as they become more of a veteran in the NFL.

[see pdf for full figures and formula]

Works Cited

Collective Bargaining Agreement, August 4, 2011. Retrieved March 27, 2023, from

BLS Contract Collection – Metadata Header This contract is provided by Martin P. Catherwood Library, ILR School, Cornell. (1993, March 29). U.S. Department of Labor. Retrieved March 27, 2023, from

Christopherson, C. (2020, May 21). The Art of NFL Contracts Part 2: The Examples. Sports Illustrated. Retrieved March 25, 2023, from

Fitzgerald, J. (2015, March 26). Explaining the NFL’s Rookie Salary Cap. Over the Cap. Retrieved March 27, 2023, from

Foltice, B., & Markus, J. (2021, November 22). Finding Value in NFL Draft Picks.

Gough, C. (2023, March 1). NFL average TV viewership per game 2022. Statista. Retrieved March 25, 2023, from 

Kurmits, F. E., & Adams, A. J. (2008, November). The NFL Combine: Does it Predict Performance in the National Football League. Journal of Strength and Conditioning Research, 22(6), 1721-1727. 10.1519/JSC.0b013e318185f09d

Langland, T. (n.d.). How Rookie Wage Scale Has Changed the Way NFL Teams Draft. Bleacher Report. Retrieved March 27, 2023, from

Moore, T. (2019, April 11). NFL Revenue ($18 Billion) Keeps Going Up, But Quality Of Play (3-3 Teams) Keeps Getting Sacked. Forbes. Retrieved March 25, 2023, from

NFL CBA: Cap History. (n.d.). Retrieved March 25, 2023, from

Raj, A. (2023, March 21). What is the 25 percent rule in NFL rookie contracts? First Sportz. Retrieved March 27, 2023, from

Rosenthal, G. (2010, March 3). 20 questions about the uncapped year – ProFootballTalk. ProFootballTalk. Retrieved March 25, 2023, from

Salary Signing Data 2005-2017. (n.d.). Sports Contracts, Salaries, Caps, Bonuses, & Transactions. Retrieved March, 2023, from

Super Bowl Ratings Chart, All-Time. (n.d.). Sports Media Watch. Retrieved March 25, 2023, from

Join the Discussion