Category: Entertainment

Presentation Description

Liked by Godwin Thiruchelvam


Presentation Transcript

Show Me the Money: An Analysis of the Impact of Free Agency on NFL Player Performance:

Show Me the Money: An Analysis of the Impact of Free Agency on NFL Player Performance


Overview Contract year phenomenon -conventional wisdom is that players try harder and therefore play better in the final years of their contract (year before becoming free agents ) Corollary-players exert less effort in the year after signing a multiyear contract in free agency What is the truth behind this conventional wisdom?


Contents Game structure differences between MLB, NBA, and NFL Contractual differences between MLB, NBA, and NFL Economic model Data Methodology Analysis Conclusions

MLB and NBA background:

MLB and NBA background In MLB, Paul Sommers (1993) found evidence of a decline in performance following signing of multiyear contracts In NBA, Kevin J. Stiroh (2007) found both a very significant increase (5% level) in a player’s composite rating (+.381) pre-free agency and a very significant decrease (-.325) in the year after a multiyear signing Stiroh also finds that a contract shift from the contract year to the first year of a multiyear contract was associated with 4.5 fewer team wins Bottom line: many studies in MLB, inconsistent but significant evidence for contract year phenomenon; one very comprehensive and well-executed study in the NBA that finds very significant evidence of the contract year phenomenon

Game Structure Differences:

Game Structure Differences Number of players: NFL-11, MLB-9, NBA-5 Player specialization: NFL-3 types (offense defense special teams), MLB-pitchers, batters, NBA-none Greater interaction of teammates: NFL>NBA>MLB Games played: NFL-16, NBA-82, MLB-162 Bottom line: probably much more difficult for an individual player to affect individual or team outcomes by trying harder in NFL

Contractual Differences:

Contractual Differences Contractual features of MLB, NBA, and the NFL and their expected influence on free agent incentives (incentives to perform harder in free agency year rated greatest, 1, to least, 3, of three leagues) Free agency eligibility and types Salary cap Revenue sharing (per team) Average player salary % contract money guaranteed MLB All UFA None 31% 2.82 Most Incentives 1 1 Unclear, 2 2 1 NBA UFA and RFA Soft Roughly 20-25% 5.585 Most Incentives 2 2 Unclear, 3 1 1 NFL UFA, RFA, tags Hard Approximately 70-75% 1.5 Little (mainly some bonuses) Incentives 3 3 Unclear, 1 3 3

Economic Model:

Economic Model Player performance (P) is affected by factors within a players control and factors outside of a player’s control P = β1(player performance outside of a player’s control) + β2(intrinsic player performance) + ε = β3(team quality) + β4(coaching) + β5(player-team fit) + β6(playing time) + β7(years of experience) – β8(years of experience squared) + β9(natural ability) + β10(effort) + ε Motivated more by career Motivated more by short-term incentives


Data Only quarterbacks (QB), running backs (RB), tight ends (TE) and wide receivers (WR) from 1998-2007 (10 years) NFLPA data on free agency Player performance data from footballoutsiders.com Traditional metrics: yards, yards/touch, QB rating Innovative metrics: DYAR, DVOA Overall: 1090 players, 3594 player observations (average career of 3.29 years, NFL average is ~3.5) Independent variables: FA, YA, switching teams, years of NFL experience (age), playing time (touches) Main hypothesis: there will be positive FA effects and negative YA effects, but they will be less significant than NBA and MLB


Methodology FA=1 if a player is going to be a free agent at the end of the year YA=1 if a player was an FA last year and is not an FA this year Use Stata to conduct OLS regressions on dependent vars (performance) using independent vars (FA, YA, yearsexp, etc) First, run regressions on all positions as a whole Second, break down data in specialized categories to see if the best players show different effects from the rest Within this, break players down into two categories: those who are regulars (type 1) and those who are not (type 2)


Analysis Regressing just FA on performance; result: huge negative correlation FA QB throw for 533 fewer yards (1% level significance)—WHY? Importance of controlling for fixed effects Still, FA QB throw for 429 fewer yards (1% level significance) YA shows significant positive correlation with performance Yet Yards/touch shows no significant changes This is the opposite of what’s expected—What’s going on?

PowerPoint Presentation:

Potential problem with data—players can be cut in NFL before their contracts are up So, players who are cut and players who finish their contracts are both coded as FA FA who are cut do not face a priori incentives and will not resign with previous team Switching teams (dummy) one way to control for this, although it narrows the sample considerably Also, free agency is likely correlated with age, and players get worse as they age When both switching teams and age (yearsexp) are included in regressions, FA results are insignificant Same thing for YA when yearsexp is added

PowerPoint Presentation:

Yards/touch regressions using newteam, FA, yearsexp, yearsexp^2 as independent variables Newteam FA Yearsexp Yearsexp^2 QB1 -.185 (.117) -.074 (.117) .108*** (.040) -.008*** (.002) RB1 -.337*** (.095) -.042 (.086) .002 (.040) -.003 (.003) TE1 -.742*** (.249) .158 (.210) .012 (.099) -.009 (.008) WR1 -.404** (.165) .212 (.172) -.080 (.066) -.000 (.005)

Summary of results:

Summary of results Two specialized regressions: 1) dividing players into performance thirds 2) dividing players into categories based on how many times they have been FA given their yearsexp Neither shows significant results, although age is still very significantly negative Overall, important findings: Below-average players make up the majority of the free agent pool No significant FA or YA effects All players get very significantly worse with age QB show unique aging effects


Conclusions NFL shows no evidence of contract year phenomenon while MLB and NBA do Two main reasons for this: Players less able to alter performance by trying harder in the NFL Players in NFL don’t have guaranteed contracts Some questions remain: Why is it that free agents have less playing time without significant declines in performance? Why is it that QB show a different aging pattern?

Final Questions/Thoughts:

Final Questions/Thoughts Do free agents who improve do themselves a disservice? Compile a list of player salary and see if players who improve are compensated for it How do free agent effects differ before and after the implementation of the NFL salary cap in 1994? NFLPA should distinguish between two types of free agents Need better estimates of teamwork, coaching, other NFL intangibles (known-unknowns) that affect performance



authorStream Live Help