Tuesday, June 17, 2008

NBA teams win 7 games more when their coach is a former All-Star

I'm shocked by the results of this NBA study: "Why Do Leaders Matter? The Role of Expert Knowledge." It's by Amanda H. Goodall, Lawrence M. Kahn, and Andrew J. Oswald.

Basically, the study tries to figure out if having been a successful NBA player would make you a more successful coach. The answer: yes, and hugely.

After controlling for a bunch of possible confounding variables, it turns out that:

-- for every year the coach played in the NBA, his team will win an extra 0.7 games per season [over a coach who never played].

-- if the coach was *ever* an NBA All-Star, his team will win an extra 7 games per season.

That last number is not a misprint. I'll repeat it in bold:

A coach who was an NBA All-Star at least once is associated with a 7-game increase in team wins.

The result is statistically significant, at 2.5 SDs above zero.

What could be causing this?

My first reaction was that teams who spend a lot of money on talent might also willing to spend to hire a high-profile coach. But the study controlled for team payroll, so that can't be it.

So what is it? I can't figure it out, but I'd bet a lot of money that it's not (as the study thinks it is) that All-Star coaches are somehow "experts" in getting the most out of their teams. I see no reason why it should be assumed that better players would make better coaches (and, in fact, I have heard plausible arguments the other way, that guys with natural talent have no idea how they do it, and so can't teach others to do the same).

For the record, here are the other variables controlled for: payroll, race of coach, age and age squared, NBA head coaching experience (and experience squared), college head coaching experience, "other pro" head coaching experience, and number of years as an NBA assistant coach.

(In the study, I am looking at the regression with no team fixed effects. That's Table 1, bottom panel, second from the right. There were 16 All-Star coaches in the sample, for 68 seasons. The authors don't list them.)

Any ideas? I feel like I must be missing something ... coaches can't be contributing seven wins to their team just because they were better players. Could they?

(Hat tip:
Andrew Leigh via Justin Wolfers.)


At Tuesday, June 17, 2008 5:26:00 PM, Blogger Don Coffin said...

I agree; their findings suggest coaches have an extraordinary effect on team performance. (My calculations, incidentally, indicate a range of estimated effects of being an All-Star, ranging from 5.3 to 9.3 wins per season. Also, the efects of NBA experience are of similar magnitude. A coach with 10 years playing experience apparently increases wins by between about 5 and about 7 wins.)

Juts being an All-Star may not be the best measure of player quality though. Let's see. 30 teams, 15 (?) man rosters, so about 450 (?) players. All-Star rosters are about 20, right? So nearly 10% of the players make an All-Star team each year. Over a period of a few years, it's likely larger than that. I personally think the length of playing career may be a better measure of quality than ever making an All-Star team.

But it's still the case that a long playing career matters.

I wish they had tried to provide some causal mechanism by which this effect operates. But they do seem to have controlled for all the things it seems to make sense to control for.

But try this. If, in fact, length of career is a better measure of overall player quality, then, to some extent, we're looking at players who could hold a job even as their physical skills began to decline. So they may, in fact, have been better "students of the game," better able to adapt.

I dunno; I'm making this up as I go.

At Tuesday, June 17, 2008 5:50:00 PM, Blogger Phil Birnbaum said...

The only causal (correlational?) mechanism I can think of ... does the NBA have draft choices who are artificially cheap because they're not yet eligible for arbitration? Maybe the teams who hire all-star coaches are the same teams who depend on the draft. That would explain why they win seven extra games after adjusting for payroll.

A young superstar might be worth 7 games, no?

But there are 68 seasons in the study, and the average of 68 teams having one extra underpriced superstar seems unlikely.

At Thursday, June 19, 2008 2:39:00 AM, Blogger Unknown said...

It's possible that teams are racist. The majority of basketball players are black. The majority of coaches without NBA playing experience are white. If teams kept black coaches on a much shorter leash than white coaches (i.e. were less likely to tolerate losing for extended periods of time), and black coaches were a lot more likely to be all-stars than white coaches, it could explain the observed effect in the data. Of course, that's a lot of racism.

It's also possible that veteran teams are A) better and B) more likely to hire a 'player's coach', and player's coaches are more likely to be former players. Young teams hire disciplinarian coaches, who are more likely to have been college coaches, are are less likely to be all-stars.

At Thursday, June 19, 2008 2:50:00 AM, Blogger Phil Birnbaum said...

Hmmm, interesting ideas.

But if black coaches were kept on a shorter leash, it wouldn't explain their better records unless they really were better.

And if veteran teams are better, shouldn't that show up in higher payroll?

At Wednesday, June 25, 2008 10:38:00 PM, Anonymous Anonymous said...

Maybe I miss the point, but it doesn't look like the authors controlled for team quality (payroll helps, but confounds many explanations).

In some other work, co-authors and I control for team quality extensively and find coaches are fired according to their marginal contributions, regardless of race.

Could just be that all-star coaches are hired to coach better teams; it's not the coaching margin but the player quality margin in the first place.

At Wednesday, June 25, 2008 10:43:00 PM, Blogger Phil Birnbaum said...


Yes, that could be it. I agree that payroll gives only a rough estimate of team quality.

How do you isolate team quality from the actual results of the season? For instance, suppose a coach is fired after going 38-44. How can you tell how team was actually better (or worse) quality than its 38-44 record?

At Thursday, June 26, 2008 11:15:00 AM, Anonymous Anonymous said...

This exactly what "technical efficiency" determines. [Related to your earlier post, below.]

Figure out the contribution that statistics make to winning. Then figure out the statistics of the actual players on a coaches roster. You could then determine the players' contribution to winning. But you would not have isolated the coach's part.

So, instead, you can use production function estimation to find the marginal contribution to winning of the player roster and the coaches' characteristics, separtely.

This of course misses other elements of coaching. But at least part of the story is untangled.

Finally, relative to other coaches with similar talent over history, you can compare their "efficiency" given the talent they have to work with.

So, sometimes, more complicated estmiates are worth the trouble.

At Tuesday, July 01, 2008 12:17:00 AM, Blogger DQuinn1575 said...

The authors state their time period is 1996-2004. During that time period Jerry Sloan's W-L was 459-247, or .650. Larry Bird was 147-67 - combined 606-314 or .659

68 seasons at 82 games at .533 is 2972-2603. Take away 2 extremes and you get 2388-2289, or .508, or 42 wins.

It seems to me that their sample is extremely weighted by 2 outlayers; a good and a great players who made great coaches.

At Tuesday, July 01, 2008 12:25:00 AM, Blogger Phil Birnbaum said...

Good call, dquinn. Did those teams have particularly low payrolls for their performance? Because the study did control for payroll.

In any case, I'm surprised that two coaches, one with a short tenure, could make that big a difference. The authors did say they took into account the clustering of coaches when calculating statistical significance, didn't they? I should check again.

At Friday, July 11, 2008 1:48:00 PM, Blogger DA said...

I'm all for statistical analyses, but can't it be possible to observe a phenomenon that can't be totally explained statistically?

Are there no intangibles that can't themselves be measured, but can be measured in the measure of their results?

7 more wins IS the statistic. Why?

- maybe because All-stars have been coached by all-star coaches (during all-star games), and thus have a better feel for the game.
- maybe because all-stars command more respect from their players who do what they're told and get the results
- maybe because the all-stars themselves know better how to play a respective position to be successful, and get better results from their players, especially those who play in the position in which that coach excelled...


Post a Comment

<< Home