Today we have an interview with sports economist and statistician David Berri. He kind enough to give us his thoughts on a variety of topics, including Chris Paul v. Deron Williams, what James Posey brings to the team, and whether Hilton Armstrong may pan out yet. I'd introduce him more thoroughly, but he does a pretty good job of it himself...
At the Hive: Let me start by thanking you for taking the time to talk to us. For Hornets fans who don't know who you are, who is Dave Berri?
David Berri: Most importantly, an Associate Professor of Economics at Southern Utah University. In other words, I am just a college professor. Beyond that, I am also vice-president of the North American Association of Sports Economists, lead author of The Wages of Wins (Stanford Press), author or co-author of over 20 academic articles on the economics of sports, and the primary writer at The Wages of Wins Journal
@tH: Have you always been a basketball fan? If not, how were you drawn towards analyzing it instead of, say, football?
DB: Well, I analyze football also. But I started looking at basketball in graduate school. When I started looking at the economics of sports, most articles looked at baseball. It seemed to me that basketball had not been examined as much, primarily because measuring performance was more difficult.
@tH: Can you give us a quick summary of Wages of Wins? How does the approach you use to arrive at "Wins Produced" differ from other current basketball statistics?
DB: When people think about The Wages of Wins, they tend to think about basketball. It’s our work on basketball that caught the attention of Malcolm Gladwell, as well as many other reviewers. But the book is about quite a bit more. Essentially the Wages of Wins takes work we wrote on a variety of subjects in sports and economics from academic journals to the general public. The book begins with a discussion of labor disputes and sports, then moves on to the link between payroll and wins, the measurement and determinants of competitive balance, and then the value of star power in the NBA. All of that is just the first half of the book. In the second half we introduced Wins Produced and Win Score (our measures of player performance in the NBA), discuss the ability of NBA players to "step-it-up" in the playoffs, and discuss the consistency of performance in football, baseball, and basketball. The book concludes with a look at decision-making in the NBA, presenting evidence that scoring is over-valued by coaches and general managers in basketball.
Wins Produced is detailed in Chapters Six and Seven of the book. It’s important to note that the book doesn’t include any mathematical equations. To see those you need to look at an article that was just published in The Business of Sports (a three volume collection edited by Brad Humphreys and Dennis Howard).
What makes Wins Produced different is that it is entirely based on regression analysis (the standard statistical method used by economists). This analysis begins by carefully laying forth the relationship between the statistics the NBA tracks for individual players and team wins (again, you need to see the aforementioned article for the math behind all of this). With this relationship established we can determine the value of various statistics (points, rebounds, steals, turnovers, etc…) in terms of team wins. These values are then used to determine the impact each player has on team wins.
The idea that scoring is over-valued can be seen without looking at Wins Produced. Simply modeling free agent salaries or the coaches voting for the All-Rookie team tells that story. Wins Produced, though, also tells this tale. Players who score inefficiently will simply not produce many wins. That should make intuitive sense. Launching shots that do not go in does not actually help a team win games. Unfortunately, if a player can score – even if that doesn’t happen very efficiently – he can score a major payday. Hence the incentives of players (i.e. the desire to get paid gobs of money) are not consistent with the incentives of teams (i.e. the desire to win many games).
A few last notes on the NBA...
It’s important to remember that payroll and wins are not highly correlated in the NBA. This is the same story we see in football and baseball. But in the latter two sports, performance is very inconsistent across time. So it’s hard for decision-makers in baseball and football to predict the future and therefore we should not be surprised when payroll can’t explain wins in these sports.
In basketball, the box score statistics are much more consistent across time (relative to what we see in baseball and football). And these statistics do explain wins. So we should see teams with the most money acquire the best players. But this is not what we see. Although you can see this in the regression analysis, the New York Knicks illustrate the point. Over the past few years the Knicks have been among the league leaders in payroll but nowhere near the league leaders in wins.
@tH: One of the more problematic issues today is that "all assists are not created equal." An assister gets the same credit for setting up a guarded, fade-away 26 footer as he does for breaking down two defenders and creating a layup. Due to that, are assists overrated or underrated by current media?
DB: Assists are a fairly crude statistic. It is the only stat in basketball that is entirely based on the scorer’s judgment. So we should not be entirely sure about the reliability of assists as a measure of performance. That being said, I have found that players are more productive when their teammates get more assists. Although the direction of the causality is not entirely clear, I think assists tell us something.
As for the media…members of the media do not generally have any training in statistics. In general it appears they follow this rule: If the numbers support the argument then the numbers are used, if not, the numbers are ignored.
@tH: If I recall correctly, you looked at Chris Paul's and Deron William's college careers a while back, and came to the conclusion that CP should have been the higher pick. By numerous statistics, Paul has bested Williams for three years now. How wide is the gap currently, according to WoW? And is there any truth to the idea that Deron Williams is a better "fit" for Utah than Chris Paul would be?
DB: Chris Paul led the NBA last year with 25.4 Wins Produced. So he was the most productive player in the game. Deron Williams was very good, producing 15.4 wins. This mark ranked 16th in the league. Among point guards, only Paul, Jason Kidd, Steve Nash, Chauncey Billups, and Jose Calderon produced more than Williams. In sum, Williams is a very good point guard. But he is not nearly as productive as Paul.
As for the better fit issue… I think Utah would be better off with Paul. Player performance can be negatively impacted by changing teams. That being said, the effect is not that great. So I think given the very large difference in productivity numbers between Paul and Williams, I think Paul would likely be more productive than Williams in Utah.
@tH: Staying on the subject of Paul, various statistics suggest that he had one of the greatest offensive seasons by a point guard in history (#1 all time PER, #1 all time Win Shares). In your estimation, where did his last campaign rank historically and what could he have improved?
DB: Since 1991-92, no guard in the NBA has produced more than 25 wins in a single season. So what Chris Paul did in 2007-08 was very impressive. Basically you have to go back to the 1980s – when Magic Johnson and Michael Jordan surpassed the 25 win mark a few times – to find a guard who played better than Paul.
@tH: I don't mean to open Pandora's Box here, but there's a question I intend to ask of every sports statistician I ever talk to, heh. What's your take on so-called clutch ability? Does it exist?
DB: As noted earlier, we have looked at performance in the playoffs and the regular season in the NBA. This analysis failed to find any evidence that a player could systematically play better in the post seasons. Although I think people can look at this in different ways, I think the underlying story is suspect. The "clutch ability" story is that a player can simply turn "it" on when his team needs it. So, according to this story, when a team needs a three pointer to win, certain players can simply step up and hit that shot. But if that were true, why can’t that same player turn it on earlier in the game? If he did the team probably wouldn’t need a last second three-point shot.
My sense is that certain players generally get to take last second shots on teams. Some of these are bound to go in, and when that happens, the player hitting the shot (and other people around the player) tend to think the player has some special ability. In reality, a certain percentage of these shots -- given the general shooting ability of the player – are bound to go in the basket. In other words, hitting a last second shot (or even several such shots) doesn’t mean a player has any special skills.
@tH: The James Posey signing brought up an intriguing question- what exactly is the "value" of winning a championship? (ie, how many consequent mediocre seasons are worth one title?) Is there an answer to this seemingly subjective conundrum from an economics standpoint?
DB: First of all, James Posey is an above average player (at least that is the Wins Produced story). I talked about this in the following post:
So Posey helps the Hornets with his production on the court. Now does the fact Posey has played on championship teams help? I don’t buy this story. The Celtics were led in the 2008 NBA Finals by Garnett, Ray Allen, Paul Pierce, and Rajon Rondo. None of these players had won an NBA title before 2008. That lack of experience didn’t seem to matter in the NBA Finals.
@tH: Last, is it true that most big men come into their own during their 3rd seasons? For example, Hilton Armstrong of the Hornets has not shown much improvement over his first 2 campaigns. How do you determine if it's too early to give up on a player?
DB: I am not sure about big men, but players in general do get better in their third season. When we look at Armstrong, we see that he posted a 0.050 WP48 [Wins Produced per 48 minutes] his rookie season. Average WP48 is 0.100 for an NBA player in general, although what Armstrong did is close to average for a rookie.
In his second season, though, Armstrong’s WP48 fell to -0.094 (yes that is a negative sign). So Armstrong was way below average. Going back to your question, I think it is incorrect to say Armstrong "has not shown much improvement". What the numbers show is that Armstrong got much worse.
So should the Hornets give up on Armstrong? I am not sure, but it is important for the coaches to figure out why he played so badly in 2007-08. If those things can be fixed, then the team should keep him. If not, it is time to look elsewhere.
@tH: Thanks for taking the time to talk to us, really appreciate it.
DB: Glad to talk to you. Fans of the Hornets should be very excited right now. New Orleans is clearly one of the top teams in the NBA and the 2008-09 season should be fun.
Thanks once again to Dr. Berri, hope you enjoyed the interview as much as I did.