Jumping in the Pool: What Determines Which Players the NBA Considers in the Draft?
The most popular position in the 2014 National Basketball Association draft was shooting guard. Of the 60 players selected, 21 players could--according to Basketball-Reference.com--play shooting guard in the NBA. After three years, twelve of these shooting guards were no longer on an NBA roster. Meanwhile, Tyler Johnson--a shooting guard from Fresno State--made Miami's roster in 2014 and played well enough to receive a $50 million contract in 2016. Johnson, though, was not one of the 21 shooting guards selected during the 2014 draft.
The fact that Johnson was overlooked may not be surprising when we consider the scope of the problem facing NBA decision-makers. There are specific rules governing the NBA amateur draft. Beginning in 2005, (1) the league's collective bargaining agreement stipulated that for non-international players to be eligible to be drafted, they must be 19 years old during the draft calendar year and at least one season has passed since graduation of high school. While there are a few exceptions, nearly all non-international players play at least one year of college prior to entering the NBA draft. (2) After one year of college, players can declare as early entrants for the NBA draft. (3) After players complete four years of their college eligibility, they are automatically eligible for the NBA draft.
The NBA draft consists of two rounds where each team gets one selection per round. The 60 players selected represent a small fraction of players eligible to be drafted (5.8%) and an even smaller fraction of all college players (1.6%) who could be eligible for the draft, had they declared for it. In essence, the NBA draft is an exercise of finding a few needles in a stack of hay. Past research has looked at how those needles are ordered. Specifically, Berri, Brook, and Fenn (2011) examined what factors impacted where a player was selected in the draft. This current study will consider a different question. How do teams determine which players are in the drafted pool and which players are not?
Understanding what leads a player to be drafted is important because of the benefits that being drafted provides. The Collective Bargaining Agreement (CBA) establishes that players drafted in the first round receive a guaranteed two-year contract with team options for a third and fourth year. The salary is also determined by the CBA and the order in which the player was drafted. (4)
Players drafted in the second round do not receive guaranteed contracts, though according to a report by Dauster (2018), "Of the 132 college players selected in the second round of the last six NBA drafts, 91 of them--or 68.9 percent--received at least a one-year guaranteed NBA contract.....Of the 72 college players selected between 31st and 45th during the last six drafts, 65 of them--or a whopping 90.3 percent--received a guaranteed contract from an NBA team."
Often, these contracts are not that much less than those selected at the end of the first round. For example, in 2017 the 31st pick (2nd round) earned just $100,000 less than the 30th pick (1st round). This goes to show, that while the first round does provide more guarantees for players, simply being selected in the draft is highly correlated with making an NBA roster in the player's first year. And as such, it is important to examine the outcome of whether or not a player is drafted.
Our analysis shows that scoring points is the largest predictor of a player being drafted, while shooting efficiency is not a strong predictor. This finding coincides with the literature, which shows that individual players are rewarded for scoring points at the expense of shooting efficiently, which is the strongest predictor of team wins. (5) Additionally, our analysis shows that players on teams who have average to below average winning percentages and do not perform well in post-season tournaments are less likely to be drafted.
Our inquiry will be organized as follows. We will begin by reviewing past literature on decision-making in sports. This will be followed by a discussion of the data we use for this study and the empirical model we estimate. Empirical results are then presented and discussed, followed by concluding observation.
Decision-Making in the NBA
One goal of the general manager of an NBA franchise is to evaluate talent in order to improve team performance. Teams can improve the quality of their team through personnel decisions in three ways: re-signing current player, signing free agents, and through the amateur draft. There is an extensive literature that points to how teams evaluate talent of players currently in the league.
In determining player salary, scoring points is the single greatest factor that leads to higher salaries. Berri, Schmidt, and Brook (2007), Berri and Schmidt (2010), Simmons and Berri (2011), and Deutscher, et al. (2017) state that points scored is the primary determination of a free agent's salary. Specifically, Berri and Schmidt (2010) report that a one-standard deviation increase in points scored per 48 minutes led to a $1.4 million increase in free agent salary. When these authors looked at other box score statistics, though, they failed to find a single factor where a one-standard deviation increase led to as much as $1 million in additional pay.
In addition to player salaries, which tend to be negotiated by a team's general manager, there is also evidence to suggest that coaches use scoring as the primary measure of a player's ability. Historically, NBA coaches voted for the First and Second NBA All-Rookie teams. (6) Berri and Schmidt (2010) reported (7) that points scored was the primary determinant of how many votes a rookie received. Specifically, these authors noted that points scored had seven times the impact on voting totals of any other box score statistics considered.
A similar story is told with respect to minutes per game. Berri, Deutscher, and Galletti (2015) offered evidence that personal fouls have the largest impact on how minutes are allocated in the NBA. But as the authors noted, this is because the rules of the game require a player leave the game after six fouls. Of the other box score statistics, again, points scored is the dominant factor. A one-standard deviation increase in points per 48 minutes led to a 2.5 increase in minutes per game. Except for assists, a one standard deviation increase in all other box score stats failed to earn a player as much as an extra minute of playing time per game. (8)
What about the evaluation of players not in the NBA? As noted, Berri, Brook, and Fenn (2011) previously examined the order by which players were selected in the NBA draft. Consistent with all the evaluations we have noted, these authors also found that points scored dominated where a player was selected. A one-standard deviation increase in point scored per 40 minutes was linked to a player being selected approximately six slots earlier. As we have seen before, no other box score statistic yielded this sort of a return.
All of these studies tell a very clear story. Players who score more points are generally considered to be better players. Since games are won by the team that scores the most points, this might appear to be the correct way to evaluate players. But focusing on points scored can be misleading. Consider a player like Allen Iverson. Across his Hall-of-Fame career, Iverson averaged 26.7 points per game and led the NBA in this category in four different years. Iverson, though, didn't score all these points because he was an effective scorer. Across his career Iverson's effective field goal percentage was only 45.2%. An average point guard in the NBA, though, has an effective field goal percentage of 47.5%. In sum, Iverson was consistently a below average shooter.
As Berri (2015) notes, Iverson didn't create the shots he took. When he was traded in 2006 from the Philadelphia 76ers to the Denver Nuggets, the number of field goals the Sixers attempted per game didn't really change. In other words, a player like Iverson really just takes shots from his teammates. Consequently, teams need to focus on shooting efficiency to see if that was a good decision. By focusing so much on points scored (as opposed to efficiency), teams are giving players the incentive to focus more on their personal scoring and less on what actually wins games. (9)
In line with this research, our objective is to analyze systematic patterns that teams adopt in order to determine their draft pool. We limit our analysis to players who competed in collegiate basketball since that is the source of a majority of the draft pool, and because there is consistent data over time. (10) Part of our analysis is to measure the value that teams place on shooting when it comes to evaluating player talent in the draft, but also identify other factors that teams use to select which type of players to draft.
We have collected data beginning with the 2002-2003 NCAA through 2016-2017 season. Our data includes players only at Division I schools (11) and does not include international players. We want to limit our sample to who are most likely to be drafted, but we do not want to be overly restrictive. As such, we limit our sample to players with at least 300 minutes played in the season. We choose this cutoff because over the years, all players selected in the draft played at least this amount of time. As a result of this restriction the sample is reduced by about 38 percent. (12)
We couple these data with NBA draft data from 2003 to 2017, the most recent year available. Each year in our sample, the draft consisted of two rounds with one pick for each team per round, for a total of 30 picks per round. Table 1 shows the total number of players that played at least 300 minutes and the number of players eligible to be drafted in a given year. As the third and fourth columns of Table 1 show a very small fraction of college players are actually drafted.
Across our sample, about 1.6% of all players were drafted, and about 5.8% of eligible players were drafted.
The vast majority of all players drafted come from a major conference; a list that includes the Atlantic Coast Conference, the Big 10, Big 12, Big East, Southeastern Conference, and the Pacific Athletic Conference-12. (13) Table 2 shows that each year the fraction of college players drafted from a major conference ranges from a low of 65.7% (in 2003-2004) to a high of 87.5% (in 2016-2017). On average, 79% of players are selected from these top conferences.
For each season we have box score data representing player performance for each player in college basketball. Statistics representing player performance are shown in the Panel 3A of Table 3 and include: points, rebounds, assists, steals, blocks, turnovers personal fouls, and various measure of shooting efficiency. (14) While these are commonly reported statistics in box scores, this study measures player performance on a per-minute basis. As college games are 40 minutes, we adjust performance statistics using a per 40 minute measurement. Summary statistics with this adjustment are presented in Panel 3B of Table 3.
To analyze the sample by draft status, we limit the sample to players who are eligible to be drafted in a given year and separate it into two columns, players who are drafted, or not drafted.
Table 4 shows these results. Unsurprisingly, those who are drafted overall have better statistics (points scored, rebounds, assists, steals, blocks, personal fouls). In addition to player stats, biographical information on each player, such as height, position, and year in school is collected; this data indicates that players drafted tend to be taller and younger.
It is also the case that players who are drafted come from more successful teams. Teams with drafted players won, on average, 72% of their games in our sample. In addition, these teams won a higher fraction of regular season conference titles, a higher fraction of conference tournaments, made the NCAA tournament more often, and had more Final Four appearances.
Although interesting, the simple comparisons of means does not control for anything. A more sophisticated analysis is offered in the next section.
We use a logit model to analyze the factors that are correlated with being drafted. As key covariates we include player statistics measured on a per 40 minute basis and other observable measures about the player (height, position, and year in school). We also include team measures: conference affiliation, categorical dummies to measure if the team had a winning percentage below 40 percent, between 40 to 60 percent (the omitted category) and above 60 percent. Dummy variables are also included to indicate whether the player was part of a major conference, made a Final Four appearance, won the conference title, won the conference tournament, and whether the team made the NCAA tournament. The model is represented by the following equation:
[Prob(Drafted = 1).sub.it] = [beta](player [statistics.sub.it]) + [theta](player [observables.sub.t]) + [gamma] (team [characteristics.sub.it] + [u.sub.it] (1)
This model is similar to one employed by other previous studies utilizing college basketball data (Berri, Brook and Fenn, 2011). However, one of the limitations of this data is that it is subject to selection bias. In order to be in the analyzed sample of this model, players must be eligible to be drafted. Thus, they either have to complete their years of eligibility or declare early. In the 2017 draft, 48 underclassmen declared for the draft, of which 35 were drafted, which accounted for 58.3 percent of all draft picks. Players entered the draft early based on expectations of being drafted, while those who were less certain about their draft status remained in college.
As players can self-select into the draft pool before they exhaust their eligibility there is a selection issue that we need to address. Our first approach is to use the panel nature of the data to analyze the survival of each player towards being drafted. To model this selection, we implement a Cox Proportional Hazard Model represented by the following function:
[lambda][(Drafted).sub.it] = [lambda](t) exp[[beta](player [statistics.sub.it]) + [theta](player [observables.sub.t]) + [gamma](team [characteristics.sub.it]) + [u.sub.it]] (2)
The coefficients in this model provide hazard-ratios of the covariates and the likelihood of being drafted.
This hazard model may not account for the unobservable factors that lead players to enter early into the draft. Thus our second approach limits the sample to observing players only during their freshmen year. We then look at the outcome if they were ever drafted, as shown in the model below:
[Prob(Ever Drafted = 1).sub.it] = [beta](player [statistics.sub.it]) + [theta](player [observables.sub.i]) + [gamma](team [characteristics.sub.it]) + [u.sub.it] (3)
This model removes the censoring issue that is caused by players entering the draft early and not allowing the previous models to observe their collegiate performance.
The logit model examines the probability of being drafted conditional on being in the draft pool. We report the average marginal effects and calculate the change in probability of being drafted for a standard deviation change for each of the player statistics. These results are reported in Table 5.
Controlling for effective field goal percentage and other statistics, a standard deviation change in points scored leads to a 2.9 percent increase in probability of being drafted. Comparing this coefficient to that of effective field goal percentage (0.007) shows that scoring points has an impact over four times greater on the probability of being drafted than effective field goal percentage. Other statistics also have a positive impact on the probability of being drafted (defensive rebounds, assists, steals, blocks, and fewer personal fouls), but points scored has the largest impact of any box score statistic. Similar to all the previously cited research on player evaluation in basketball, our results continue to support the argument that teams value scoring as the most important player attribute, even though this does not necessarily contribute to team wins.
In addition to scoring, playing on successful teams also has a large impact on the probability of being drafted. Players on a winning team--defined as having a 60 percent team winning percentage or better--are 2.5 percent more likely to be drafted than players on an average team. A similar result is seen when we look at most measures of post-season success. Winning a conference regular season title has no impact on draft status, but being a part of a team that wins the conference tournament increases the probability of being drafted by 4.8 percent. Making the NCAA tournament adds a one percent increase, and making a Final Four appearance increases the probability of being drafted by an additional 1.8 percent. As such, controlling for player and team performance during the regular season, a team that wins a series of consecutive games in the last weeks of the season can increase the likelihood of being drafted by 7.6 percent. So as we saw with the simple analysis of means, a player with better teammates is more likely to be drafted than an equally productive player on a worse team.
Focusing our analysis on the factors that lead a player to select into the draft, Table 6 reports the hazard rate ratios. (15) Here the "hazard" being modeled is that of being drafted. While we cannot compare these coefficients directly to the logit estimates of Table 5, we convert the coefficients in order to represent the change in the hazard rate ratios for a standard deviation change in the variable of interest. As such, we can compare the relative impact of each coefficient within the hazard model.
Similar to the logit model, a standard deviation change in scoring points has a much larger impact on being drafted than effective field goal percentage and other player statistics. Playing on a winning team or in a major conference also have large impacts on the hazard of being drafted.
One additional result stands out from the estimation of the hazard model; players on a winning team are, statistically, at a much higher risk of being drafted. Worded differently, players on average and losing teams are less likely to be drafted. Over the span of this analysis, only 9 out of the 11,189 players on teams with a win percentage under .400 have been drafted. Of the 17,086 players on average teams--defined by a winning percentage between .400 and .600--96 have been drafted, or about 0.56 of a percent. On winning teams, 4.1% of the 13,475 players are drafted.
Model (3) further accounts for potential selection bias by looking at the relationship between performance during the freshmen year and the probability of ever being drafted. These results are presented in Table 7, and similar patterns emerge as when looking at the full sample. The player statistic that increases the probability of being drafted the most is still points scored. Other statistics, like defensive rebounds, assists, steals, and blocks, are also positively correlated with being drafted, but the magnitudes are smaller than those in the full model. These results further support the general findings of this research.
One might think that having one top talent could, by itself, lead to significant team success. The data, though, does not support that argument. Specifically, Berri (2018) notes that how many wins a basketball player produces can be empirically measured. For men's college basketball, NCAA.org presents data back to the 2002-2003 season. Since 2002-2003, Anthony Davis, of the University of Kentucky produced the most wins in 2011-2012. His mark of 12.96, though, would not have been enough to ensure Kentucky had a winning record. This indicates that to have a winning record, a productive player must also have productive teammates.
In sum, NBA decision-makers essentially ignore a very large pool of talent. Again, as noted above, one player cannot produce a winning record by themselves. But if a player's teammates are not sufficient to produce a winning percentage above 0.400, there is very little chance the NBA will consider that player. Hence, the NBA appears to be missing out on many players who could be productive in the NBA (where the teammates are clearly going to be better).
The 2005 Collective Bargaining Agreement established a rule that all non-international players needed to be one year removed from high school to be eligible for the draft. This rule took effect beginning with the 2006 draft. To test whether there were systematic changes as a result of this policy, we separated the sample into pre- and post-policy drafts. Estimating the logit model on the probability of being drafted and the hazard model show no significant changes in the coefficients between these two time periods. (16)
The results of this study show the statistic with the largest impact on getting a player drafted is scoring points. This result is consistent with other research that draft order, earnings, and career length are greatly influenced by a player's ability to score. The statistic that does not have a consistent effect through the different specifications and models is shooting efficiency. This shows a misalignment between team and player incentives. A player's incentive is to score as much as possible, even if that means taking more shots at the expense of higher quality shots by teammates. A team's incentive is to win games, but this is done by players who shoot efficiently, that is players make shots that they take.
In addition to over-emphasizing scoring points, decision-makers also fail to separate a player from his teammates. Evidence of this is shown by NBA teams essentially ignoring all players on teams that fail to win 40% of their games. In addition, we see that post-season team performance is one of the biggest factors explaining why a player is drafted. Making the Final Four is the result of winning just 4 games, given they are the consecutive games in a tournament, but it is a small sample in which to evaluate talent. But teams consistently draft players at higher rates because they were part of a team that played well in this tournament.
These misaligned incentives are interesting, in that teams are the decision-makers and could simply draft players who will contribute to team wins. But they don't do that. The question then is why? Part of it might be that teams are not using statistics to properly evaluate talent. The scouting process itself involves watching the players. It seems reasonable to suspect that this process itself would be biased towards seeing scorers on winning teams as the best players. Certainly scorers on winning teams would likely appear to be better players.
To combat such a bias, decision-makers would have to be trained in data analysis and learn to rely on this objective information. Although teams reportedly hire data analysts, the actual decision-makers may not have this training. Consequently, it is possible that the analysis provided is discounted in favor of what the decision-maker actually sees when looking at each player.
Of course, this is just speculation. We are only reporting that the draft process in the NBA is biased towards factors that are either not good indicators of a player's impact on wins (i.e. points scored) or fail to separate the player from teammates (i.e. team success). Why teams behave in this fashion is certainly a good subject for future research.
For now, we can offer some clear thoughts on why Tyler Johnson was ultimately left home on draft night in 2014. As a 6'4" senior shooting guard, Johnson was not particularly tall and relatively old. He also was below average (relative to other drafted players) points scored per game. Finally, Fresno State only won 54% of its games in 2013-2014, a mark that is also below average for drafted players. Consequently, given what we have learned about how the pool of drafted talent is constructed, it is not surprising Johnson didn't hear his name called.
(1) Prior to 2005, the draft rules stipulated that players only needed to be out of high school in order to be eligible to be drafted.
(2) Some players have decided to forgo college to play a year of professional basketball in a foreign country. For example, Brandon Jennings played one year in Italy prior to being selected in the 2009 NBA draft. Most recently, Emmanuel Mudiay played one year in China prior to being selected in the 2015 NBA draft.
(3) Players with remaining college eligibility have until April 29th to enter the NBA draft. In 2011, a new rule was implemented which allowed players to declare for the draft and withdraw their names by April 11th without losing their college eligibility.
(4) For the players drafted in 2018, the first pick will receive about $9 million the first year, the 30th pick will receive about $1.8 million the first year, with the average first-year contract being about $3.6 million.
(5) Berri, Schmidt, Brook (2006), Berri (2008), and Berri and Schmidt (2010) all detail a model of player performance called Wins Produced. This model builds on the observation that wins in basketball are primarily a function of offensive efficiency (points per possession employed) and defensive efficiency (points per possession acquired). This link between efficiency and wins indicates that for teams to win they must acquire the ball without the opponent scoring, maintain possession, and ultimately turn that possession into points. And that means wins in the NBA are primarily about rebounds, creating and avoiding turnovers, and shooting efficiency.
(6) The 2013 team was the last time NBA coaches voted for this award. Starting in 2014, this award was voted on by members of the media.
(7) This study looked at voting for this award from 1995 to 2009.
(8) Berri, Deutscher, and Galletti (2015) noted that a one standard deviation increase in assists led to 1.3 additional minutes per game.
(9) As Berri (2015) notes: "Teams win in the NBA because they are able to gain possession of the ball without the other team scoring (i.e., grab defensive rebounds, force turnovers); keep possession of the ball (i.e., avoid turnovers and grab offensive rebounds), and ultimately turn possessions into points (shoot efficiently from the field, get to the line and hit free throws)." Given this list, player evaluation shouldn't focus on scoring totals but rather on rebounds, steals, turnover, and shooting efficiency.
(10) International players, though eligible for the draft, are not included in the analysis because of reliable sources of data which contain performance statistics prior to being drafted.
(11) Over this time period, there has been only one Division II player drafted, Ronald Murray, thus we only analyze those players at Division I schools.
(12) The results of this study do not change significantly by including all players in a given season.
(13) We measure conference affiliation based on the conference that each school belonged to each year, even though the conferences and names have changed over this time period.
(14) We also include measures of effective field goal percentage, free throw percentage, and three-point field goal percentage.
(15) One advantage of using the Cox Proportional Hazard Model is that the underlying hazard function does not need to be estimated. The trade-off is that without the underlying hazard function, we cannot calculate the marginal effects.
(16) These results are available from the authors by request.
Basketball-Reference.com (2018). https://stats.ncaa.org/ [website]. Retrieved from https://stats.ncaa.org/rankings/change_sport_year_div
Berri, D. (2018) Sports Economics. New York, New York: Macmillan Education.
Berri, D. (2015). Think you know Basketball? You need to know the numbers to know the game. Sport & Entertainment Review: 1, 6-13.
Berri, D. (2006). Economics and the National Basketball Association: Surveying the literature at the tip-off. in The Handbook of Sports Economics Research. John Fizel (Ed). (pp. 21-48). New York: Routledge.
Berri, David, J. (2008). A simple measure of worker productivity in the national basketball association. In Humphreys and D. Howard, (Eds.), The Business of Sport (pp. 1-40). Westport, Connecticut: Praeger.
Berri, D., Brook, S., and Fenn, A. (2011). From college to the pros: Predicting the NBA amateur player draft. Journal of Productivity Analysis, 35, 25-35.
Berri, D., Brook, S., & Schmidt, M. (2007). Does one simply need to score to score? International Journal of Sport Finance, 2(4), 190-205.
Berri, D., Deutscher, C, and Galletti, A. (2015) Born in the USA: National origin effects on time allocation in US and Spanish professional basketball. National Institute Economic Review, R41-R50.
Berri, D., and Schmidt, M. (2010). Stumbling on wins: Two economists explore the pitfalls on the road to victory in professional sports. Princeton, NJ: Financial Times Press.
Berri, D. J., Schmidt, M. B., and Brook, S. L. (2007). The Wages of wins: Taking measure of the many myths in modern sport. Stanford, California: Stanford University Press.
Bodvarsson, O., and Brastow, R. (1998). Do employers pay for consistent performance? Evidence from the NBA. Economic Inquiry, 36(1), 145-160.
Bodvarsson, O., and Brastow, R. (1999). A test of employer discrimination in the NBA. Contemporary Economic Policy, 17, 243-255.
Bodvarsson, O., and Partridge, M. (2001). A supply and demand model of co-worker, employer and customer discrimination. Labour Economics, 8, 389-416.
Brown, E., Spiro, R., and Keenan, D. (1991). Wage and non-wage discrimination in professional basketball: Do fans affect it? American Journal of Economics and Sociology, 50(3), 333-345.
Deutscher, C, Gurtler, O., Prinz, J., & Weimar, D. (2017). The payoff to consistency in performance. Economic Inquiry, 55(2), 1091-1103.
Dey, M. 1997. Racial differences in National Basketball Association players' salaries. American Economist, 41, 84-90.
Gius, M, & Johnson, D. (1998). An empirical investigation of wage discrimination in professional basketball. Applied Economics Letters, 5(11), 703-705.
Hamilton, B. (1997). Racial discrimination and professional basketball salaries in the 1990s. Applied Economics, 29, 287-296.
Hill, J. (2004). Pay discrimination in the NBA revisited. Quarterly Journal of Business and Economics, 43(1-2), 81-92.
Hoang, H., and D. Rascher. (1999). The NBA, exit discrimination, and career earnings. Industrial Relations, 38(1), 69-91.
Kahn, L. M. (1991). Discrimination in professional sports: A survey of the literature. Industrial Labor Relations Review, 44, 395-418.
Kahn, L. M. (2000). The sports business as a labor market laboratory. The Journal of Economic Perspectives, 14(3), 75-94.
Kahn, L. M., and Sherer, P. D. (1988). Racial differences in professional basketball players' compensation. Journal of Labor Economics, 6(1), 40-61.
Krautmann, A., Von Allmen, P., and Berri, D. (2009) The underpayment of restricted players in North American sports leagues. International Journal of Sport Finance, 4(3), 155-169.
Prinz, J., Weimar, D., & Deutscher, C. (2012). Popularity kills the talentstar? Einflussfaktoren auf Superstargehalter in der NBA. Zeitschrift fur Betriebswirtschaft, 82(7-8), 789-806.
Stiroh, K. J. (2007). Playing for keeps: Pay and performance in the NBA. Economic Inquiry, 45(1), 145-161.
Wallace, M. (1988). Labor market structure and salary determination among professional basketball players. Work and Occupations, 15(3), 294-312.
Yang, C, & Lin, H. (2012). Is there salary discrimination by nationality in the NBA? Foreign talent or foreign market. Journal of Sports Economics, 13(1), 53-75.
Tiffany Greer, (1) Joshua A. Price, (1) and David J. Berri (1)
(1) Southern Utah University
Tiffany Greer graduated from Southern Utah University in Economics and Mathematics and then from University of North Carolina-Charlotte in May 2018 with a dual masters degree in Mathematical Finance and Economics. She is currently employed in the Financial Services industry.
David Berri, PhD, is a professor of economics in the Department of Economics and Finance. His current research focuses on the economics of sports, specifically the topics of the efficiency of decision-making, competitive balance, worker productivity, and women in sport.
Joshua A. Price is an associate professor of Economics at Southern Utah University. His research analyzes the application of behavioral economic tools in educational, health, and sports settings.
Table 1. Players Drafted by Season Total Number of Players in NCAA who play 300 Number of Players Season minutes Declared for Draft 2002-03 2,565 558 2003-04 2,610 688 2004-05 2,614 698 2005-06 2,633 713 2006-07 2,699 732 2007-08 2,790 779 2008-09 2,845 764 2009-10 2,781 792 2010-11 2,841 853 2011-12 2,834 787 2012-13 2,876 785 2013-14 2,909 838 2014-15 2,909 801 2015-16 2,923 823 2016-17 2,921 802 Totals/Averages 41,750 11,413 Number of Percent of Percent of players NCAA players Eligible NCAA Season drafted drafted Players drafted 2002-03 32 0.01 0.06 2003-04 35 0.01 0.05 2004-05 36 0.01 0.05 2005-06 44 0.02 0.06 2006-07 47 0.02 0.06 2007-08 47 0.02 0.06 2008-09 46 0.02 0.06 2009-10 53 0.02 0.07 2010-11 46 0.02 0.05 2011-12 51 0.02 0.06 2012-13 46 0.02 0.06 2013-14 44 0.02 0.05 2014-15 44 0.02 0.05 2015-16 42 0.01 0.05 2016-17 48 0.02 0.06 Totals/Averages 661 0.016 0.058 Table 2. Players Drafted by Season from Major Conferences Players from Total Number of Number of Major Major Players in Major Conference Players Conference Season Conference declared for Draft drafted 2002-03 535 142 23 2003-04 530 136 23 2004-05 533 156 28 2005-06 581 173 37 2006-07 584 144 39 2007-08 596 163 38 2008-09 592 162 35 2009-10 609 176 40 2010-11 607 195 36 2011-12 607 175 44 2012-13 618 170 33 2013-14 599 169 36 2014-15 629 177 37 2015-16 629 192 34 2016-17 639 193 42 Totals/Averages 8,888 2,523 525 Percent of Percent of Major Major Conference Percent of Drafted Conference Eligible players Players from Major Season drafted drafted Conference 2002-03 0.043 0.162 0.719 2003-04 0.043 0.169 0.657 2004-05 0.053 0.179 0.778 2005-06 0.064 0.214 0.841 2006-07 0.067 0.271 0.830 2007-08 0.064 0.233 0.809 2008-09 0.059 0.216 0.761 2009-10 0.066 0.227 0.755 2010-11 0.059 0.185 0.783 2011-12 0.072 0.251 0.863 2012-13 0.053 0.194 0.717 2013-14 0.060 0.213 0.818 2014-15 0.059 0.209 0.841 2015-16 0.054 0.177 0.810 2016-17 0.066 0.218 0.875 Totals/Averages 0.059 0.208 0.790 Note: Major Conferences include Atlantic Coast Conference, Big 10, Big 12, Big East, Southeastern Conference, Pacific Athletic Conference - 12 Table 3. Summary Statistics 3A Per Game Statistics Variable Obs Mean Std. Dev. Min Max Points 35,906 8.29 4.39 0.36 28.86 Effective FG Pct 35,906 0.39 0.09 0.10 0.80 FG pct 35,906 0.44 0.08 0.11 0.80 FT pct 35,899 0.68 0.12 0.00 1.00 3pt pet 31,529 0.31 0.13 0.00 1.00 Offensive Rebounds 35,906 1.13 0.78 0.00 5.94 Defensive Rebounds 35,906 2.59 1.28 0.20 11.24 Assists 35,906 1.58 1.22 0.00 9.97 Steals 35,906 0.80 0.47 0.00 3.96 Blocks 35,906 0.39 0.49 0.00 6.33 Turnovers 35,906 1.60 0.74 0.14 5.73 Personal Fouls 35,906 2.11 0.63 0.13 4.19 Note: statistics are measured on a per game basis 3B Per 40 Minute Statistics Variable Obs Mean Std. Dev. Min Max Points 35,906 13.39 4.36 1.19 34.72 Offensive Rebounds 35,906 1.99 1.28 0.00 8.19 Defensive Rebounds 35,906 4.37 1.65 0.69 15.38 Assists 35,906 2.53 1.55 0.00 11.81 Steals 35,906 1.32 0.60 0.00 6.19 Blocks 35,906 0.70 0.86 0.00 9.13 Turnovers 35,906 2.69 0.87 0.36 7.22 Personal Fouls 35,906 3.78 1.32 0.41 10.92 Note: statistics are measured on a per 40 minute basis Table 4. Table Summary Statistics by Draft Status Not Drafted Drafted Diff in mean Variable Mean Std. Mean Std. p-value Points 14.17 4.55 19.92 4.02 0.00 Offensive Rebounds 2.01 1.28 2.45 1.40 0.00 Defensive Rebounds 4.52 1.69 5.88 2.17 0.00 Assists 2.58 1.55 3.13 1.91 0.00 Steals 1.32 0.59 1.51 0.63 0.00 Blocks 0.69 0.83 1.29 1.28 0.00 Turnovers 2.61 0.83 2.88 0.74 0.00 Personal Fouls 3.60 1.26 2.96 0.81 0.00 Height 76.90 5.56 78.78 3.47 0.00 Freshmen 0.00 0.02 0.15 0.36 0.00 Sophomore 0.00 0.05 0.18 0.38 0.00 Junioe 0.01 0.09 0.24 0.43 0.00 Senior 0.99 0.11 0.43 0.50 0.00 Team Win pct 0.52 0.17 0.72 0.13 0.00 Final Four Appearance 0.01 0.09 0.15 0.36 0.00 Conference Title 0.11 0.31 0.25 0.44 0.00 Conference Champion 0.00 0.06 0.10 0.30 0.00 Make NCAA Tournament 0.20 0.40 0.72 0.45 0.00 Observations 9,217 571 Table 5. Logit Regression on the Outcome Y=1 if Drafted (Marginal Effects of a Logit Model) Average p-value from Marginal logit Variables Effects Std. Err. estimation Points 0.007 0.000 0.000 Offensive Rebounds -0.004 0.002 0.070 Defensive Rebounds 0.007 0.001 0.000 Assists 0.008 0.001 0.000 Steals 0.011 0.003 0.000 Blocks 0.015 0.002 0.000 Turnovers -0.004 0.003 0.162 Personal Fouls -0.015 0.002 0.000 Effective FG Pct 0.077 0.027 0.004 Height 0.002 0.001 0.000 Position Shooting Guard -0.008 0.006 0.184 Small Forward -0.002 0.007 0.779 Power Forward 0.000 0.008 0.960 Center -0.013 0.008 0.108 Class Freshmen 0.300 0.057 0.000 Sophomore 0.135 0.020 0.000 Junior 0.079 0.010 0.000 Major Conference 0.057 0.005 0.000 Win pct <.40 0.002 0.009 0.798 Win pct <.60 0.025 0.004 0.000 Final Four Appearance 0.018 0.011 0.099 Conference Title 0.002 0.004 0.578 Conference Championship 0.048 0.021 0.021 Made NCAA Tournament 0.010 0.004 0.014 Probability of Being Drafted for a Std. Dev Variables change in X Points 0.029 Offensive Rebounds -0.005 Defensive Rebounds 0.011 Assists 0.013 Steals 0.007 Blocks 0.013 Turnovers -0.003 Personal Fouls -0.020 Effective FG Pct 0.007 Height 0.011 Position Shooting Guard Small Forward Power Forward Center Class Freshmen Sophomore Junior Major Conference Win pct <.40 Win pct <.60 Final Four Appearance Conference Title Conference Championship Made NCAA Tournament Table 6. Hazard Model (Hazard defined as being drafted) Variables Hazard Ratio Std. Err. p-value Points 1.181 0.013 0.000 Offensive Rebounds 0.931 0.060 0.265 Defensive Rebounds 1.057 0.034 0.086 Assists 1.131 0.045 0.002 Steals 1.416 0.114 0.000 Blocks 1.305 0.065 0.000 Turnovers 1.370 0.099 0.000 Personal Fouls 0.632 0.040 0.000 Effective FG pct 1.139 0.937 0.875 Height 1.021 0.015 0.149 Position Shooting Guard 0.357 0.067 0.000 Small Forward 0.678 0.154 0.088 Power Forward 1.924 0.504 0.013 Center 1.097 0.307 0.742 Class Freshmen 2.488 0.379 0.000 Sophomore 2.484 0.336 0.000 Junior 3.320 0.394 0.000 Major Conference 4.776 0.593 0.000 Win pct<.40 0.343 0.146 0.012 Win pct <.60 2.300 0.344 0.000 Final Four Appearance 1.358 0.266 0.119 Conference Title 1.233 0.138 0.061 Conference Championship 0.841 0.194 0.454 Made NCAA Tournament 1.111 0.149 0.432 Change in Hazard Rate Variables Ratio for a Std. Dev change Points 5.15 Offensive Rebounds -0.09 Defensive Rebounds 1.75 Assists 1.75 Steals 0.84 Blocks 1.12 Turnovers 1.19 Personal Fouls -0.49 Effective FG pct 0.11 Height 5.26 Position Shooting Guard Small Forward Power Forward Center Class Freshmen Sophomore Junior Major Conference Win pct<.40 Win pct <.60 Final Four Appearance Conference Title Conference Championship Made NCAA Tournament Table 7. Logit Regression on the Outcome Y=1 if Player Was Ever Drafted (based on statistics during freshman season) Average Marginal Variables Effect Standard Error p-value Points 0.003 0.000 0.000 Oifensive Rebounds 0.000 0.001 0.947 Defensive Rebounds 0.002 0.001 0.012 Assists 0.005 0.001 0.000 Steals 0.005 0.002 0.002 Blocks 0.007 0.001 0.000 Turnovers 0.001 0.001 0.689 Personal Fouls -0.006 0.001 0.000 Effective FG Pct 0.025 0.015 0.088 Height 0.001 0.000 0.001 Position Shooting Guard 0.007 0.004 0.099 Small Forward 0.013 0.006 0.036 Power Forward 0.020 0.009 0.027 Center 0.014 0.009 0.114 Major Conference 0.034 0.003 0.000 Win pct<.40 -0.013 0.002 0.000 Win pct<.60 0.003 0.003 0.253 Final Four Appearance 0.020 0.010 0.058 Conference Title 0.004 0.003 0.182 Conference Championship 0.007 -0.370 -0.015 Made NCAA Tournament 0.009 0.003 0.003 Probability of Being Drafted for a Std. Dev. Variables change in X Points 0.012 Oifensive Rebounds 0.000 Defensive Rebounds 0.003 Assists 0.007 Steals 0.003 Blocks 0.006 Turnovers 0.001 Personal Fouls -0.008 Effective FG Pct 0.002 Height 0.007 Position Shooting Guard Small Forward Power Forward Center Major Conference Win pct<.40 Win pct<.60 Final Four Appearance Conference Title Conference Championship Made NCAA Tournament
|Printer friendly Cite/link Email Feedback|
|Author:||Greer, Tiffany; Price, Joshua A.; Berri, David J.|
|Publication:||International Journal of Sport Finance|
|Date:||Feb 1, 2019|
|Previous Article:||A Market Test for Ethnic Discrimination in Major League Soccer.|
|Next Article:||Insolvencies in Professional Football: A German Sonderweg?|