An Introduction to Incentives
Ask an economist how to solve a problem, and he’ll tell you incentives are the answer.1 He wouldn’t be wrong. Punishment and reward are fantastic tools for exploiting self-interest in the service of the common good. In football, they’re made up of red cards and penalties, trophies and relegation, and always in the interest of preserving the ‘beautiful game’.
But incentives do not always respond the way we expect them to. Take, for example, the infamous 1994 Caribbean Cup match between Barbados and Grenada. In an effort to encourage attacking play during extra-time, tournament officials decided that extra-time golden goals would be worth double for goal difference purposes. A nice idea in theory, but by the end of the match, Grenada found themselves frantically trying to score in either net while Barbados defended both goals. An incredible series of events actually made it in Barbados’ best interest to force an equalizer so that they could score an extra-time goal. You can view a video of the incident below and read about it here.
“I feel cheated […] I have never seen this happen before. In football, you are supposed to score against the opponents to win, not for them” – Grenada Manager James Clarkson.2
Even the slightest change to incentives can twist, shape, and decide games in ways we may never anticipate. This is not to say it is an easy task to predict the impact of new incentives – in fact is is often near impossible. However, it is the duty of football officials to review historical data to decide which rules have failed and need updating. In particular, the ‘three points for a win’ rule stands out as a serious offender (from here on the rule will be referred to as 3PW). Despite a growing stack of literature that shows the rule has had the opposite effect from what was intended, it has managed to almost completely fly under FIFA’s radar. It is high time to review the evidence for one of the most important laws of the game. But first let’s rewind.
“Football is not a circus”
In October 1980, Stoke City manager Alan Durban, angry at journalists’ criticisms of his tactics in a 0-0 draw against Arsenal, instructed them to “go and watch a bunch of clowns” if they were looking for entertainment. Durban, after all, was simply doing his job, and maybe not such a bad one at that. The “win at home, draw away” philosophy was popular amongst managers and defensive tactics were very much in vogue. Could Durban really take the fall for inverting the pyramid? Perhaps if rapper-cum-actor Ice-T had been present, he could have explained to the crowd of unruly reporters, “Don’t hate the player, hate the game.”
But this wasn’t the only problem. The early 1980s depression had taken its toll on England, and rising ticket prices and television exposure saw match attendance drop to nearly half of its 1950s record-setting numbers. Football fans cried foul: this was not the first time they had felt the beautiful game was under attack. A generation earlier Herbert Chapman, the legendary former Arsenal manager, remarked:
“It is no longer only necessary for a team to play well. They must get goals, no matter how, and the points. The measure of their skill is, in fact, judged by their position in the League table.”
In comes Jimmy Hill, former chairman of the Professional Footballer’s Association and legendary Coventry manager.3 Hill was not your conventional chairman – in fact, he was a bit of a maverick, famous for leading the charge to scrap the Football League’s £20 maximum wage. He was later known for engineering the Sky Blue Revolution during his tenure as Coventry manager, a club overhaul which would make Assem Allam’s efforts to rebrand Hull City Tigers look amateurish and lazy.
Hill, who “had long thought that soccer had become too defensive and dull” and was concerned that “goals had become rarer with every passing season,”4 proposed a simple revision to the rules: change the reward for winning a match from two points to three points. This would make wins more valuable and incentivize teams to not settle for draws. In 1981, less than a year after Durban’s speech, Hill convinced the FA to introduce his idea of ‘three points for a win’ or 3PW. Thirteen years later, FIFA adopted the system for the upcoming 1994 World Cup in the US, concerned that American fans would be turned off by draws. Sepp Blater hailed the move as “the most important sporting decision taken here, but it rewards attacking soccer”. In 1995, every remaining major football league switched to a three point system.
The Drawing Game
Advocates of 3PW tend to fall back on the result easiest to observe: it reduces the number of draws by increasing the incentives for breaking a draw. Indeed, according to former Football League chairman Brian Mawhinney, draws are still seen as a threat to football’s entertainment factor:
“I suggested that for drawn matches each team gets a point and then maybe the team that wins a penalty shoot-out gets an extra point […] We cannot afford to be complacent – people are always talking to be about how we can get more goals and more excitement in football.”5
Some statistics indicate that the rule switch did indeed reduce the number of draws. In the five English First Division seasons leading up to the change, there was an average of 133.0 draws per season. This was twenty more than the average of 113.4 in the first five seasons after.
Data from other countries yield similar results. Evidence from Turkish and German leagues shows a decrease in the number of draws after controlling for number of teams, games played, and cup matches.6,7
But is this a valid metric for measuring entertainment value? Does a reduction in the number of league draws indicate an increase in attacking play?
Scrutinizing England’s data may be the key to answering these questions. The graph below illustrates the number of draws per game over time in England first division football league (no cup games are counted). The value on the left indicates the percentage of matches played that resulted in a draw.
At a first glance, the number of draws per game (DPG) was already in the process of decreasing right before 3PW took effect in 1980 (and has actually been declining since 1970). In fact, it only takes five years for any perceived effects of 3PW to wear off. The reason? Between 1986 and 1988, the number of teams in the league was reduced from 22 to 20. The data indicates that any benefit of 3PW in terms of reducing DPG was negated by the formation of a more competitive league.
This reveals a fundamental problem with looking at draws: the DPG is inversely correlated to league competitiveness. Think about it: if a league is perfectly competitive, then all matches will result in a draw. Take the spike in DPG in 1968, for example, which coincides with the introduction of the substitution. The substitution rule change meant, among other things, that a team would no longer have to play with ten men if one of their players was injured on the pitch. This would naturally lead to fewer unbalanced matches, more draws, and a higher DPG. Indeed, I find a statistically significant inverse correlation between DPG and league competitiveness.
Not only did the rule switch have no noticeable long-term impact on DPG, but the reader must make a subjective judgment on whether they prefer fewer draws, or a more competitive league. If you prefer fewer draws in return for the same old winners and losers, then this rule change may be right for you. I do not however believe this to be the intention of 3PW.
There are other, better metrics, for evaluating the rule’s success. Let us consider them instead.
Regardless of the number of draws, if 3PW encourages attacking play, then it may have served its purpose after all. After all, fans do not watch games to find out the winner – there are plenty of live score feeds online – they watch to see the beautiful game unfold. If the FA is looking to increase stadium attendance, they need to make the experience worth it, most noticeably through an increase in attacking play.
As it turns out, 3PW actually incentivizes defensive play and sabotage (a punishable offence, e.g. purposefully negligent tackling). Researchers looking at card data from England, Spain, and Germany show that teams in a winning position were more likely to commit punishable offences under the 3PW system.8,9
At the core of this issue is the natural tradeoff of offensive play: by increasing your chances of scoring, you are also increasing your chances of conceding a goal. This means that, following the implementation of 3PW, if a team scores and takes the lead, then the expected payoff of playing defensively will increase relative to the expected payoff of playing offensively. In other words, the stakes are so high that a team will not risk giving up a goal. By making wins more valuable, the FA may have succeeded actually made ‘unattractive’ football more common. To quote football statisticians Chris Anderson and David Sally, “three points for a win had not rewarded attacking soccer. It had rewarded cynical soccer”.
Most damning is evidence from a 2005 study by then-University of Chicago economists Luis Garicano and Ignacio Palacios-Huerta. In a discussion paper of theirs, the two analyzed Spanish league data from the 1994-1995 season (when 2PW was last used) and compared it to the 1998-1999 season – the four year gap is so that they do not have to assume an immediate response and change in tactical development. They control league data against cup results, which should remain largely unaffected by the change, to eliminate potential external variables such as referee strictness, injuries, etc.10
Their study also provides evidence that 3PW is ineffective. They show a ~28% increase in the use of starting forwards, but also an increase in the number of defenders and a ~10% increase in both fouls and yellow cards as a result of 3PW. However, despite the increase in forwards, number of goals scored did not go up. This suggests that any attack-minded benefits of 3PW were negated by its less appealing sabotage-effect. The study found that “when ahead, teams became more conservative, increasing their defenders, scoring less goals, and allowing fewer attempts to score by their opponents”.
But more importantly, this study shows that 3PW is actually detrimental to match attendance. They find the incentive change actually decreased attendance for teams who played more defensively and committed more sabotage. By controlling for team popularity and visiting/home factors, Garicano and Palacios-Huerta show a negative correlation between ‘team dirtiness’ and attendance, at a significance level of 1%. This means that, statistically speaking, there’s a 99% chance that a correlation between dirty play and attendance figures exists.
One point that seems to be brought up consistently is that 3PW inspires more league competition, and in particular gives lower-ranked teams a fighting chance to avoid relegation.
The idea is that by increasing points for a win, teams facing relegation at mid-season a given a fighting chance to turn everything around. If true, this would make the league more exciting for supporters of lower-ranked teams.
However, a look at the ten largest comebacks in the top flight of English football tells a different story:
It is striking that only two of the top ten comebacks happened post-1980. In particular, comparing the case of Fulham in 2010 to Ipswich in 1978 reveals the counter-productive effects of 3PW. The Whites lost only one more game during the second half the season than during the first. Their fantastic comeback was due almost entirely to their ability to convert draws into wins. Ipswich, on the other hand lost, only twice in the second half versus eleven times in the first half of the 1978 season, yet managed to advance only a similar number of spots on the league table. The Blues succeeded by converting their losses into wins. It is telling that teams pre-1981 could engineer a comeback by winning against those who beat them, while modern teams can only hope to edge out a win over teams they have already drawn.
We then look at whether or not changing the points system affects which teams are being relegated. If does, because 3PW rewards teams who win more games, teams that win/lose would benefit more than teams that survive by drawing. When the 1976-1980 season tables for the top three tiers of English football are recalculated under 3PW, we get the following:
The evidence suggests that the impact of 3PW is in fact minimal. For teams coming in last or second to last, how you count your points doesn’t change the fact that you don’t have any. Teams in third to last place may sometimes benefit from a new point system, but 80% of the time it would not have made a difference. A fourth to last place team had a 60% chance of staying relegated, but frankly, it’s the third tier of English football. Whether or not the Tranmere Rovers stay up another season is of little concern in this case. Teams that are relegated are not relegated for playing less exciting football, they just aren’t good enough.
So far, it doesn’t look like 3PW has a strong effect on teams’ comeback potential or their chances of being relegated. We still have one more metric to consider, though: league competitiveness. To calculate this metric, we consult a study penned by Kjetil K. Haugen, professor of Logistics and Sport Management at Molde University college. Haugen’s analysis demonstrates that decrease in league-competitiveness following the rule change in the United Kingdom, Norway, and Romania.
By using a tweaked version of his formula, we can calculate the competitiveness level of the bottom-6 teams under 3PW and 2PW systems. These levels are represented on a scale of 0% to 100%, where 0% represents a minimally competitive league, and 100% represents a maximally competitive league.
This graph represents the competitiveness of the bottom six teams during the second half of the season. All pre-1981 season data has been recalculated under 3PW to avoid endogenous distortions (not doing this significantly overestimates the competitiveness of pre-1981 teams). Note that the variance index was able to exceed 100% – it’s because lower-tiered teams consistently over-performed compared to their mid-season rankings. For example, in 1977, the six second-flight teams ranked last in December proceeded to win 186 combined points by June – they were only expected to get 45. This rubric is meant to give us a visualization of competitiveness, not a predictive figure.
At a first glance, it doesn’t look like there is any correlation between 3PW and bottom six. Top-flight competitiveness was already increasing before the change, and data from the other flights doesn’t reveal anything either. A subsequent statistical test shows two things: First, 3PW is a poor indicator of a team’s performance during the second half of a season. Second, on average, a bottom-six team’s performance during the first half of a season is not a great indicator of its performance during the second half.
These results show that, in fact, 3PW doesn’t give losing teams a second wind.
Jimmy Hill’s role in launching 3PW ultimately won him the Contribution to League Football Award at the 2009 Football League Awards. Perhaps if the FLA administrators had done their research, they would have given it to someone else. In most regards, 3PW has been ineffective in accomplishing its goals and, as several studies report, has actually encouraged sabotage and decreased stadium attendance. Further analysis shows that leagues actually become less competitive due to the rule change. This is not to say 3PW affects a league’s inherent quality, but rather that it makes teams’ differences more noticeable in table rankings. If the FA really wants to avoid the same winners and losers each year, then they’ll need to reconsider 3PW.
Points of Interest
- It is worth nothing that if 3PW encourages team to shed the “win at home, draw away” mentality, then it may have value. Indeed, analysis of Portuguese league data reveals a reduction in home field advantage, albeit at the cost of league competitiveness12. Similar results were obtained by looking at German league data.11, 13 It’s not a bad metric, but if home field advantage is how we gauge entertainment, then Barcelona is the most boring team in the world.14
- My analysis, despite its strong results, sometimes comes from a relatively small sample size, ranging from 30 to over 1000 observations depending on the calculations. It would be worth reviewing my conclusions from a larger database, preferably of leagues from outside the UK.
- All my data is organized and available for download here. Part of the reason I decided to start a blog was to make football data access easier for the general public. It took me months to download everything and organize it, so please don’t waste your time doing the same thing. I highly encourage anyone who found this post interesting to check out my data and see what they can do with it.
- You can find a more in-depth explanation for my methodology here, along with some cool graphs to consider.
- Please let me know if you find any mistakes in my approach or if you feel additional data would be of particular value in analyzing the impact of 3PW. You can reach me at email@example.com
- Special thanks to Jonathan Wilson of the Guardian for inspiring this post.
4 Chris Anderson, David Sally, The Numbers Game: Why Everything You Know About Soccer Is Wrong, 2013
6 Bas¸ levent, C., & Tunal (2001). Incentives and outcomes in football: The effect of the three-points system and home advantage on outcomes. Retrieved Febuary 21, 2008, from http://smye2002.univ-paris1.fr/program/paper/e5_bas.doc.
7 Dilger, A.,&Geyer, H. (2009). Are three points for a win really better than two? A comparison of german soccer league and cup games. Journal of Sports Economics, 10, 305-318.
8 Julio del Corral & Juan Prieto-Rodriguez & Rob Simmons, 2010. “The Effect of Incentives on Sabotage: The Case of Spanish Football,” Journal of Sports Economics, , vol. 11(3), pages 243-260, June.
9 Dilger, A.,&Geyer, H. (2009). Are three points for a win really better than two? A comparison of german soccer league and cup games. Journal of Sports Economics, 10, 305-318.
10 Garicano, Luis and Palacios-Huerta, Ignacio, Sabotage in Tournaments: Making the Beautiful Game a Bit Less Beautiful (September 2005). CEPR Discussion Paper No. 5231. Available at SSRN: http://ssrn.com/abstract=831964
11 Dewenter, R. (2003). Raising the scores? Empirical evidence on the introduction of the three-point rule in Portuguese football. Discussion Paper, Institute of Economic Policy, University of the Federal Armed Forces, Hamburg.
12 Guedes, J. C., & Machado, F. S. (2002). Changing rewards in contests: Has the three-point-rule brought more offense to soccer? Empirical Economics, 27, 607-630.
13 Amann, E., Dewenter, R., & Namini, J. E. (2004). The Home-Bias Paradox in Football. Discussion Paper, University of Duisburg-Essen, Essen. Dilger, Geyer / A Comparison of German Soccer League and Cup Games