This past Monday the Sports Law Blog asked the question: Why does tanking occur in the NBA but seemingly not in other leagues?
The question basically presumes that the NBA’s losers will intentionally lose games towards the end of a season to enhance their draft position. What people wonder is why we don’t see similar behavior in other sports.
Before I offer my two cents, let me ask and answer (or at least try and answer) a few related questions.
1. Do NBA teams actually engage in this behavior?
This is what we said in The Wages of Wins:
Beck Taylor and Justin Trogdon [in an article published in 2002 in The Journal of Labor Economics] wondered how the incentive to lose altered the behavior of NBA teams. During the 1983–84 season, the year before the lottery was established, these authors found that teams eliminated from the playoffs were, relative to playoff teams, about 2.5 times more likely to lose. This result was uncovered after they controlled for team quality. In other words, non-playoff teams were found to lose more often than one would expect even bad teams to fail. When the lottery was instituted the next season, though, the increased tendency of non-playoff teams to lose vanished.
That is not the end of the story. In 1990 the NBA instituted a weighted lottery, where the odds of landing the top pick would improve the more the team lost. Once again, teams in the NBA had an incentive to lose. Once again Taylor and Trogdon report that after controlling for team quality, non-playoff teams were more likely to lose, although the size of the effect was smaller. With a weighted lottery non-playoff teams were only 2.2 times more likely to lose. Hence, as the incentives these teams faced were changed, the behavior changed as well.
So the work of Taylor and Trogdon offers pretty clear evidence that NBA teams did in the past lose games in an effort to improve their draft position. In other words, the basic premise of the Sports Law Blog seems to be on pretty solid ground.
2. How do NBA teams improve on their ability to lose?
I don’t think a team could go to its players and ask them not to try hard. For one thing, it’s possible that this behavior is technically illegal. At least, I don’t think you are legally allowed to throw a sporting contest. Certainly such behavior would make league authorities – if they could absolutely prove it was happening (and apparently regression results are not considered proof) – very unhappy. Consequently, if a player was asked to do this and he later confessed, the team’s decision-makers could be in trouble. So asking the players to throw the games is probably not a good idea.
The way to do this is to do what the Clippers and Grizzlies did last year when each needed to lose to enhance their playoff positioning. You don’t ask your players not to try, you simply don’t play your best players. And that action can be taken for supposedly legitimate reasons. Often coaches will say “We need to see how the young players perform so we can know what we are going to do for next season.” Of course, a side benefit of letting the young players perform is more losses and a greater chance at securing a better draft choice.
3. Why do NBA teams do this?
In The Wages of Wins we note the key motivation. Often the difference between the first player taken and the next is quite large. A good example is the difference between Tim Duncan and Keith Van Horn. Duncan was the first choice in the 1997 and he is on pace to have 194.2 career wins before the end of this season. Van Horn was the second pick and produced 34.4 wins before exiting the league in 2006. In sum, Duncan is one of the greatest players of all time and Van Horn, well, was not.
For a more recent example, consider LeBron James and Carmelo Anthony. This season LeBron is on pace to produce 17.1 wins and has a Wins Produced per 48 minutes [WP48] of 0.256. If we look at his first four seasons we see a player that has been paid $18.8 million and produced 65.6 wins. In sum, LeBron is really good.
In contrast, Carmelo Anthony has a WP48 of 0.084 this season (average is 0.100) and is on pace to produce about 4.1 wins. Across his first four seasons he has been paid $15.1 million but only produced 11.5 wins. In sum, Carmelo is no LeBron.
If the Nuggets had the first pick in 2003 and took LeBron, they would have seen more than 50 additional wins for only $3.7 million more in salary. Given this difference, we can see why the NBA’s losers need to keep losing.
It’s not good enough to be bad. To have the best chance at the very best talent, you have to be really bad. And if you fail to get that top choice – because you failed to be really, really bad – you can end up with a talent that really doesn’t help you very much at all.
This is essentially the point I tried to make in an NPR interview I gave to John Moe (which was broadcast on March 31st). In this interview I argued that it would be somewhat irresponsible for a team to try and win every game down the stretch of a losing season when this action reduces the likelihood a team would succeed in the long-run.
4. Should the NBA allow this behavior?
The problem for the NBA is that teams know it’s very likely Greg Oden and Kevin Durant can dramatically alter their fortunes next year. So teams have an incentive to do what they can to improve the chance that these particular talents will be acquired.
To stop this behavior, the NBA went to a lottery system in the mid 1980s. As Taylor and Trogdon noted, when the lottery was not weighted, this practice of losing to win vanished. Only when the weighted lottery was introduced did this behavior return.
Clearly the NBA can stop this behavior by re-introducing the non-weighted lottery.
And given that losing to win undermines the integrity of the game, a non-weighted lottery should be re-instated. Asking NBA teams to ignore their incentives is not a policy that is likely to work. In other words, NBA teams do not have a clear incentive to take action that hurts their team but might help the image of the league.
Basically the NBA should heed the primary lesson economics teaches — people respond to their incentives. If you want behavior to change you have to change the incentives people face.
5. Why do football and baseball teams not engage in this behavior?
First of all, I have not seen a study that establishes that football teams and baseball teams do not do this. Still, I did watch my Detroit Lions blow the first pick in the NFL draft by defeating the Dallas Cowboys on the last day of the 2006 NFL season. So I suspect NFL teams do not lose games in an effort to secure a better draft choice (at least if they do, someone forgot to tell the Lions).
There are a few reasons NFL teams do not do this. First of all, predicting performance in football is difficult. We are pretty sure that Greg Oden is going to be a good NBA player. As the Lions have found with Joey Harrington and Charles Rogers, top picks in the NFL draft do not always produce. Beyond the Lions, we have shown that predicting the future performance of a quarterback – even one that has been in the league awhile – is very difficult. This uncertainty reduces the incentive to do everything you can to improve your draft position.
Secondly, one player cannot easily turn an NFL team around. Even a great quarterback cannot make a team win. Teams need an offensive line, receivers, a running game, and a defense to be successful. One great quarterback (or one great wide receiver, or linebacker, etc…) can only take you so far.
Finally, NFL teams have 22 positions to fill. Even if you pick out of the top 10 it is possible to secure the best talent at your position of need. Given all this, the incentives facing NFL teams are not quite the same we see in the NBA.
As for baseball, the incentives appear to be even smaller. Certainly the uncertainty regarding future performance is even greater. In baseball you are taking a player who has to navigate the minor leagues before he can even reach your Major League roster. Odds are very long for any player to even make it to “the show”, let alone have a substantial impact on your team’s fortunes once he gets there.
Answering the question
So to answer the question, the difference between the NBA and the other sports leagues is the incentives the teams face. NBA teams are much more likely to “know” that a talent can dramatically change their future fortunes. They also know what they can do to secure that talent. Consequently, we should not be surprised when the best players on losing teams suddenly find themselves sitting a lot at the end of an NBA season.
In contrast, in both the NFL and MLB, we simply do not see the same incentives to lose games to improve draft position. Consequently we do not see much of this behavior.
Future topics
A couple of days ago I asked what people wanted me to write about in the future. Here are a few requests people made:
Analysis of the NBA draft. This is the subject of a pair of academic articles I am working on – with Stacey Brook and Aju Fenn – so there will definitely be more on this topic soon.
Analysis of player transactions this summer. This will also be something I will do. Especially since I will have more time (I hope) this summer.
Analysis of NBA history. This sounds like a good idea as well. If you have something you wish for me to look at in the past – and there have been a few requests already – I will try and look at this as well.
Thanks to everyone for taking the time to post a comment. Hopefully this summer I will be able to get to these requests.
– DJ
Wages of Wins Journal 
Jerry S
April 12, 2007
Great article on losing by non-playoff teams in the NBA. The talking heads on ESPN are having a field day with this topic right now. I cannot understand why the league does not go back to a non-weighted formula when the statistics show clearly show it is a better format.
I would love to see in the future an analysis of college players that are going to be drafted by the NBA this summer and then to follow them in the first few years to see if a correlation exists between college performance and the pros.
Andres
April 12, 2007
Either this tanking in the NBA is hitting a lot of websites or I’m reading very many like-minded writers.
I haven’t found an account that tells me with all certainty that losing games is going to help a team. In fact, the writing on this website is conflictive in this regard. There’s too many variables. Is the player actually going to perform with the way he’s rated? For ever James there’s a Carmelo, every Duncan a Van Horn. Where’s the analysis of all of the past draft picks that compares #1 to #2? Even there I don’t see a guarantee.
Then you write, well, it doesn’t take one Howard. You have to have more than that. No guarantee again.
Finally, even if you do lose, you’re only getting better odds at the pick, no guarantee.
What’s missing here is a real assessment of the GMs and the organizations at the difficult task of building good basketball teams. The sure sign of a bad organization: having to tank games to get a pick. I would suspect that getting the pick isn’t really about winning ball games; it’s about galvanizing excitement around a bad product to try to sell some season’s tickets.
On MLB, I think that there’s tanking at the end of every year. When the rosters get expanded, you always hear about the minor league prospects that come up and they wind up playing significant time for losing teams with the purpose of “talent evaluation.” Hardly playing your best players to win the game.
Ryan J. Parker
April 12, 2007
It’s fairly easy to see Carmelo is not as good as LeBron, but is LeBron’s Wins Produced higher simply because of the lack of talent around him when he was drafted? Put Carmelo on the Cavs and I think he’s got more Wins Produced, while LeBron would have less on the Nuggets.
dberri
April 12, 2007
I feel like if I respond to one comment I should respond to all. But I don’t have time to respond to everything (I do read them all, though).
Okay, that being said….
Ryan your question made me go look at the difference between LeBron and Melo’s teammates (I just happened to have the spreadsheet open). LeBron’s teammates have produced 30.9 wins (not counting tonight). Melo’s teammates have produced 37.6 wins. So the difference is less than seven wins. That is not enough to make much difference in each player’s overall productivity.
Jason
April 13, 2007
If tanking is a problem *and* the NBA acknowledges that draft order exists in some way to try to give lousy teams a way to become competitive, one thought I’ve had is to change the weighting to give better odds not to the team that finishes poorest in any season, but by the number of years out of the playoffs (with invariable ties split by franchise record since then). I suspect that this would have similar effect to a non-weighted formula in terms of incentive to lose.
Tanking 10-15 games seems plausible, but tanking successive seasons for a shot at that kid who just finished the 7th grade? Less so.
Paulo
April 13, 2007
I think Jason has a great idea. I remember back then in ’97 that it was pretty much unfair for the Spurs to get the number 1 pick when they had David Robinson and Sean Elliot pretty much out for the year (was that the year they desperately signed Dominique Wilkins?) . Same with the Magic with two consecutive #1’s (Shaq and Webber — traded for Penny).
However, it would still reward a perennially inept and mis-managed franchise (think Clippers pre-Sam Cassel) that would inevitably trade away/let go of all its young talent. I read Bill Simmons’ article on this, and he pointed out that out of all the past “premiere” lottery picks, only Tim Duncan was the one able to deliver a championship to the team that drafted him.
I’m beginning to think the Spurs deliberately tanked that year too. Hmmm…
Owen
April 13, 2007
very interesting proposal Jchan, signed up for the fantasy league, should be fun…
Ryan J. Parker
April 13, 2007
Thanks for the reply. Yeah, 7 wins isn’t that big of a difference.
dberri
April 13, 2007
Jason Eshleman has an interesting proposal. You only get the better shot at the better talent if you are bad for multiple seasons. That might be a way to go as well.
And thanks Owen for noting JChan’s fantasy league. Hopefully everyone goes to NBA Babble and Win Score and signs up. Sounds like fun.
Ryan… before looking I would have thought what you suggested might have had some role to play. But I think LeBron ultimately offers more than Carmelo because LeBron is simply a better player.
TK
April 13, 2007
One other difference in the incentive to lose in the NFL and the NBA is the way that rookie contracts are established. In the NBA, they’re basically set in stone on a sliding scale, and at an artificially low level (well below what the market would bear if the rookie could negotiate). In the NFL, they’re negotiated within a large pool of money, and top picks command enormous guaranteed money and long-term deals.
So not only is NFL evaluation more difficult, but the financial risks are much greater. When the Lions were wrong about Harrington, the mistake occupied a huge chunk of their salary cap and was a real anchor on the team, which floundered and fired its coach. But just across town, when the Pistons were wrong about Darko, it cost them the opportunity of adding a very good player. But it didn’t tear apart their finances. They traded for ‘Sheed, won a championship, resigned him to a huge deal… and moved on.
TK
Andrew Ritchie
April 14, 2007
While those are good suggestions for future analysis I think the most interesting avenue you could venture down is the effect players have upon each other. So questions like: Are there players that improve their teammates? Is there one position on the court that is more important to be strong at than the other four? Is it better to assemble a team with one dominant player or to try and balance positive contributions from many players?
BadgerBucco
April 15, 2007
The payment system for draft picks certainly changes things, especially in the NFL. Because the top picks take up so much of the rookie salary cap and because talent evaluation is less precise in football than basketball (but more precise than in baseball – partly due to pitcher injury), the top picks are not the best value. In fact, when you consider salary, picks in the low 20s are the best value. There is a study to this effect but I can’t remember the authors off the top of my head. If the best value happens later in the first round, no team that really understands this will intentionally tank.
dberri
April 15, 2007
BadgerBucco and TK,
Good observations on the cost of being wrong in the NFL. The study of the NFL draft was offered by Massey and Thaler. I think I commented on this study last April.
John
April 16, 2007
James and Carmelo are not similar players. James is one of the league’s best players and Carmelo is below average. (see above)
doug
April 17, 2007
tim duncan and david robinson were drafted high by san antonio and got a championship but all players drafted high by a team in somewhat close years does not guarantee a title. the hawks drafted marvin williams, the next year sheldon williams, the next year josh childress and in the first round josh smith , and this year these players together have the fourth worst record.