Born in the USA: national origin effects on time allocation in US and Spanish professional basketball.
Berri, David J. ; Deutscher, Christian ; Galletti, Arturo 等
This paper examines national origin bias in professional basketball
in both the North American National Basketball Association (NBA) and
Spanish professional league (Liga ACB). Past studies into racial bias
find mixed results. In contrast, our study provides consistent evidence
that players born in the USA receive preferential treatment in both the
USA and Spain in terms of receiving additional time on the court.
Keywords: national origin discrimination; professional basketball
JEL Classifications: Z2; Z22; J7
1. Introduction
Basketball, like American football, is strongly identified with the
United States, partly because of US dominance in recent international
competitions. (1) But unlike American football, basketball is truly a
global game. For example, the New York Times reported in 2012 that about
300 million people play basketball in China. (2) That same year,
Forbes.com reported that fans from 215 different countries were expected
to watch the National Basketball Association (NBA) Finals and the games
were broadcast in 47 languages. Forbes.com noted that the NBA expected
278 million fans to follow the Finals via social media networks. (3)
Henry Abbott of ESPN, com reported in 2010 that surveys show that
basketball is now the most popular sport among young people around the
world. (4)
The identification of basketball with the United States, though,
might lead to a bias in the evaluation of talent in the sport. It is
possible that people might conclude that players from the United States
are simply better at the game invented in their country. (5) In other
words, although the game is played worldwide, basketball might suffer
from what is called 'national origin discrimination'. (6)
The US Equal Employment Opportunity Commission (USEEOC) (7) defines
'national origin discrimination' as treating employees
"unfavorably because they are from a particular country or part of
the world, because of ethnicity or accent, or because they appear to be
of a certain ethnic background (even if they are not)." In
addition, the USEEOC notes that US law "forbids discrimination when
it comes to any aspect of employment, including hiring, firing, pay, job
assignments, promotions, layoff, training, fringe benefits, and any
other term or condition of employment."
[FIGURE 1 OMITTED]
This paper estimates national origin discrimination in two of the
most prominent international basketball leagues: the National Basketball
Association (NBA) and the Spanish Liga ACB. The majority of players in
the NBA are from the United States. The number of players born outside
the United States, though, has increased from about 1 per cent in the
early 1980s to nearly 20 per cent today and, as figure 1 illustrates,
playing time by players not born in the USA increased from around 10 per
cent to more than 20 per cent in the 21st century. (8)
The Spanish Liga ACB is among the most prominent professional
leagues in the world. Spain won the silver medal in both the 2008 and
2012 Olympic Games. And the ACB is often ranked as one of the top
basketball leagues in Europe. Like the NBA, the ACB also employs players
from around the world. As table 1 notes, only about one-third of minutes
in the ACB go to players born in Spain. Meanwhile, about 25 per cent of
minutes go to players born in the USA and about 40 per cent go to
players born in places other than Spain or the USA. As the ACB is one of
the most lucrative European leagues, many American players choose to
play in Spain, something that is likely to happen if they have not been
drafted or signed to a guaranteed contract by a team in the NBA.
So in both the NBA and the Spanish ACB we see that workers are
drawn from places around the world. Do differences in national origin
impact player evaluation?
Studies into discrimination frequently examine wages or hiring.
Because we do not have salary data from the ACB and only three years of
performance data, (9) we will focus our attention on 'job
assignments'. Specifically, we analyse whether or not national
origin impacts the allocation of minutes per game in the NBA and ACB.
Our inquiry will be organised as follows. The next two sections
will discuss rules to limit discrimination in the European Union, as
well as past studies into discrimination in professional sports. This
will be followed by a discussion of a model designed to explain the
allocation of minutes in the NBA and ACB. This model will offer evidence
of national origin discrimination in both leagues. In the conclusion we
speculate about the source of the discrimination our study has
uncovered.
2. Anti-discrimination in European sports: laws and detours
Laws in the United States with respect to national origin
discrimination are designed to apply to both sports and non-sports
industries alike. And in the United States teams do not have rules in
place to limit foreign participation in sports. In Europe, on the other
hand, professional sports teams have historically limited foreign
participation. In basketball only two out of five players (40 per cent)
on the court could be from foreign countries (Primault, 2007). This all
changed with the Bosman ruling. In 1995 the European court of justice
ruled case number C-415/93 in favour of Jean-Marc Bosman and confirmed
the freedom of movement for workers. In addition it banned restrictions
on the number of foreign players from teams that were allowed to deploy.
(10)
Concerns regarding an influx of low priced foreign players led
leagues to introduce a so-called home-grown quota which set a minimum
number of local players that clubs must include on their squad.
Irrespective of nationality, players are referred to as being home-grown
if they spend a predetermined period of time (usually three years) with
the club itself or any club within the respective national association.
Precise regulations depend on the respective sport and league. From an
economic standpoint clubs are given incentives to invest in their own
talent instead of acquiring talent from foreign leagues (Frick, 2007).
So there is no minimum quota for American or Canadian players in the NBA
whereas the Spanish ACB requires that a minimum of five players (on a
roster with twelve players) must be registered with a Spanish club for
three seasons between the youth and senior categories for every team in
the Spanish ACB league (FIBA International Migration Report 2012).
Hence, national origin 'discrimination' is explicitly part of
professional sports in Spain.
The rules adopted by the Spanish ACB make it clear without the need
for any further empirical analysis that national origin plays a role in
who is named for the team. However, we focus on whether or not national
origin discrimination can be found in the allocation of playing time.
Before turning to this, we briefly review past research on
discrimination in professional sports.
3. Research on discrimination in professional basketball
The literature on wage discrimination in professional sports
suggests that the existence and magnitude of racial discrimination
depends on the observation period, estimation technique and the use of
control variables. (11) More relevant for our approach is a closer look
at the literature on nationality and its impact on remuneration. Unlike
race--which is often difficult to judge by simply looking at the player
(12)--where a player was born is a matter of fact. (13)
The study of national origin with respect to salaries in basketball
tends to provide mixed results. While Eschker et al. (2004) as well as
Yang and Lin (2012) find a negative salary effect for foreign players in
the NBA, Hoffer and Freidel (2014) estimate a wage premium of around
$900,000 per season. The observation period appears to be critical as a)
signing international players became more lucrative in recent years and
b) the outside options for players improved as international leagues
expanded. One difficulty with the study of salaries is that pay is a
function of how the decision-maker thinks the player will perform in the
future. The researcher, though, can only use past performance as a proxy
of this forecast. (14)
Playing time should be more closely linked to current performance.
Although past research has not looked at national origin and playing
time, there has been investigation of the link between race and playing
time. For example, Schroffel and Magee (2012) offer evidence of an
own-race bias by coaches, suggesting that with respect to the allocation
of playing time coaches favour players who are of the same race. (15)
This result, though, is contrasted by earlier findings from McCormick
and Tollison (2001), who find no difference between black and white
coaches with regard to playing time allocation of black and white
coaches.
It is possible that the study of race suffers from an inability to
identify it properly. Hence, we turn to our study of national origin and
the allocation of playing time in the NBA and Spanish ACB.
4. National origin and time allocation
In order to measure the impact of origin on playing time allocation
we apply regression and Oaxaca-Blinder decomposition techniques,
following two different approaches for regular season basketball in the
NBA and the ACB.
We estimate equation (1) to study the allocation of minutes in the
NBA. The dependent variable is minutes per game. A player had to play in
at least 30 games (out of 82) and average six minutes per game to be
included for a season in the data set (for the ACB a player had to be on
the court in at least 15 out of 34 games). In other words, players who
hardly ever played during a season were excluded for that season and
that season only. Naturally injuries and suspensions of players can
reduce the playing time they receive but missing games for these reasons
does not impact on minutes per game in a dramatic fashion. Players being
injured or suspended for a longer period are missing due to our data
restrictions.
The list of explanatory variables employed is given in table 2 and
includes performance statistics, talent indicators and players'
characteristics. This list begins with a collection of box score
statistics for the season being examined. Except for TOPER, ADJFG, and
FT, each of these statistics is a per 48 minute measure. As player
duties and statistics vary by their position on the court, all measures
have been adjusted for position played. (16) By calculating performance
on a 48 minute level, measures may exceed the maximum number of fouls
allowed according to the rules (which stands at six in the NBA and five
in the ACB).
In addition to a player's performance on the court, we also
consider a player's height (also adjusted for position played),
games played, (17) age and age squared, (18) a dummy variable for
players born in the USA, and draft position. Equation (1) illustrates
the specific model being estimated while table 2 notes the descriptive
statistics of the dependent and independent variables employed. As one
can see, about 82 per cent of our sample were born in the United States.
Minutes per Game = [[gamma].sub.j] + [[beta].sub.1] Productivity +
[[beta].sub.2] DraftExperience + [[gamma].sub.1] Height +
[[gamma].sub.2] Games Played + [[gamma].sub.3] Age + [[gamma].sub.4]
SquaredAge + [[gamma].sub.5] DUSA + [[epsilon].sub.i] (1)
Where Productivity = Vector of player statistics including Points,
Adjusted Field Goal Percentage, Free Throw Percentage, Rebounds,
Turnover, Steals, Assists, Blocks and Personal Fouls.
Draft Experience = Vector of variables interacting Draft Position
and Experience.
Draft position is an indicator for expected talent. Prior research
(19) has found that decision-makers in the NBA have difficulty ignoring
sunk costs. Specifically, research has shown that draft position
influences minutes per game even after one controls for on-court
performance. We address this issue by multiplying a player's draft
position by a dummy variable for each year of experience. We account for
up to ten years of experience, so DFTXP represents ten additional
independent variables. If sunk costs are ignored, draft position should
stop being significant after decision-makers have been able to see how
well a player actually plays in the NBA.
Equation (1) was estimated across data from the NBA from
2001/2-2013/14. (20) The results are reported in table 3. Before
discussing the issue of national origin we should note that these
results indicate that draft position appears to impact playing time in
the NBA until a player's sixth season in the league. In other
words, decisionmakers in the NBA do not appear to let sunk costs be
sunk. In addition, age follows the expected pattern (21) while players
who are relatively tall (relative to position played) receive more
minutes. And except for TOPER, the performance factors are statistically
significant and of the expected sign. (22)
Turning to national origin, one can see that a player born in the
USA sees an estimated 1.2 additional minutes after controlling for
performance and the other factors we believe impact minutes per game.
The approach offered in table 3 is often seen in the sports
economics literature. Non-sports studies, though, often employ the
Oaxaca-Blinder (Oaxaca, 1973; Blinder, 1973) decomposition method. This
decomposes the gap in minutes per game into explained and unexplained
portions. Here, the explained portion of the gap is characterised by
differences in observed endowments. The remaining unexplained part is
characterised by differences in returns to performance and may therefore
capture discrimination. Table 4 indicates that minutes per game are 1.3
higher for players with US nationality.
This difference cannot be explained by differences in endowments:
indeed, endowments would lead non USA players to play 0.09 minutes more
than their USA counterparts. Differences in playing time not ascribed to
differences in performance are 1.39 minutes per game in favour of
players born in the USA.
In sum, the analysis provided in both tables 3 and 4 is consistent
with the proposition that coaches in the NBA favour players born in the
USA.
Next we turn to the Spanish ACB. Again, the sample for this league
only consists of three seasons. In addition, because European leagues do
not employ a reverse order draft, this factor is not considered.
Otherwise equation (1) is being estimated for the Spanish league.
Table 5 reports descriptive statistics. Games in Europe consist of
four 10 minute quarters. So the average minutes per game--which were
close to 24 minutes per game in the NBA--are only around 20 minutes per
game in the ACB. And because the games are shorter, the box score
statistics are per 40 minutes. The remaining statistics are calculated
in the same fashion as for the NBA.
The estimation of equation (1) for the ACB is reported in table
6.23 Like the NBA, minutes per game are related to performance
statistics like points, adjusted field goal percentage, steals, assists,
and personal fouls. Although evidence of height bias doesn't
appear, games played does impact minute per game. In addition, age does
appear to matter and follows the predicted pattern. (24)
Two dummy variables were used to capture national origin bias. One
considered whether or not a player was born in the USA. The second
looked at whether or not a player was born somewhere other than Spain or
the USA.
As table 6 notes, both factors are significant at the 1 per cent
level. And the size of the effect is comparable to the NBA sample. In
the ACB a player born in the USA receives about 1.8 additional minutes
per game after we control for performance and the other factors that
impact playing time.
So players from the USA receive a boost in playing time in both the
USA and Spain. Apparently, decision-makers are biased in favour of USA
players. The success of USA players in international competition
suggests that the USA has 'better' players but we control for
performance. The fact that USA players are still given preferential
treatment controlling for performance is suggestive of coach bias in
favour of US players. It is notable, however, that non-US players born
outside Spain also spend more time on court than Spanish born players,
ceteris paribus, though the effect is not as large as the effect of
being a USA player.
Table 7 presents results from the Oaxaca-Blinder decomposition
where we compare US players with those from Spain, then Spanish versus
non-Spanish non-US players.
In a comparison of players from Spain and the USA, differences in
performance and other observables explain around 53 per cent of
difference in playing time, leaving 47 per cent of the difference
unexplained. Around 25 per cent of the difference in playing time
between Spanish and non-Spanish non-US players is explained by
observable differences.
Table 8 offers an estimate of the economic significance of the
player performance statistics employed. The rules of the game indicate
that personal fouls will tend to have the largest impact on minutes per
game. Beyond fouls we find that scoring dominates in both leagues, and
factors such as shooting efficiency, rebounds, and turnovers are of less
importance to determine playing time. Prior research from Berri (2008)
(25) indicates that wins in basketball are primarily determined by a
team's ability to gain and keep possession of the ball (i.e.
rebounds, turnovers, and steals) and shoot efficiently. Furthermore
scoring totals--which tend to be the focus of decision-makers in
basketball--are a function of a player's shot attempts. Shot
attempts, though, are often just taken from teammates.
Table 8 notes that minutes per game--in both the NBA and ACB--are
primarily dictated by a player's per-minute scoring. So the
evaluation of players in both leagues is not only similar with respect
to the evaluation of players born in the USA, we also see the same bias
in favour of scorers.
5. Concluding observations
Studies of racial bias in sport tend to provide very mixed results.
In contrast, our study of national origin bias provides a surprisingly
clear result. In both the NBA and ACB we see evidence that players born
in the USA receive preferential treatment even after we control for
performance. The magnitude of the difference ranges between 1.2 and 1.4
minutes for the NBA and the ACB. Since NBA games are 48 minutes while
ACB games are 40 minutes the impact is larger for the ACB. Given the
average playing time of players in our sample of around 20 minutes per
game, the unexplained difference in playing time accounts on average for
an extra 7 per cent playing time given identical performance.
Our empirical study into the allocation of minutes in the NBA and
the Spanish ACB indicates that national origin plays a role in the
decisions of coaches. But such a study doesn't tell us why coaches
are making such decisions.
Different theories offer reasons for why people discriminate
against others. Following Becker (1957), it is possible that
discrimination simply reflects the employer's tastes. Specifically,
employer taste discrimination can cause players of different national
origins to be treated differently if the employer prefers (or dislikes)
some players for a reason independent of on-court performance.
Phelps (1972) and Arrow (1973) focus on statistical discrimination.
Here, the employers have limited access to information about the
abilities of the workers. Hence employers use observable characteristics
which they believe indicate productivity, often race or nationality.
Hence preferred treatment of players can be credited to the
athletes' origin (Bryson, Rossi and Simmons, 2014).
The discrimination might also reflect the preferences of customers.
The literature notes that one can treat professional sports as a
customer-based service sector (Kahn, 2012). Consequently, consumers may
prefer certain types of players, resulting in differentiated willingness
to pay for trading cards (Stone and Warren, 1999), varying sizes of TV
audience (Kanazawa and Funk, 2001) and arena attendance (Brown, Spiro
and Keenan, 1991). Customer discrimination can hence directly impact the
economic performance of teams. General Managers may consider
consumers' tastes not only during hiring and contract negotiation
(Hamilton, 1997). Coaches, also hired by the general managers, could
follow the general manager's lead to give extended minutes to
certain players in response to fan preferences (Kahn, 2012). (26)
Although we cannot conclusively isolate the source of the
discrimination, we can note that statistical discrimination is unable to
capture the results presented in this paper. After all, decision-makers
in sports have an abundance of data on player performance. We cannot,
though, differentiate between the preferences of employers and
customers. One can simply argue that the allocation of playing time
reflects the preferences of coaches. But one could also argue that
players from the USA draw more fans to the arena or are increasingly
attractive to team sponsors.
We should note that coaches consider factors beyond the extensive
list of box score statistics we control for. For example, taking
charges, deflecting passes and intimidating shooters are factors a coach
could consider that are not in the standard box score. If US-born
players perform systematically better than non-US born players with
respect to the non-box score factors, then coaches might be justified in
giving additional minutes to US born players. In addition, it is
possible that externalities between US-born and non US-born players
would justify difference in how players are treated. And again, our data
would not capture such an effect.
Perhaps further research could look into these non-box score
factors. In addition, future research could examine whether the source
of any discrimination uncovered is driven by employers' or
customers' tastes. One could also expand this study to more leagues
outside Spain and the USA. In sum, future research might examine whether
or not leagues around the world exhibit a preference for players born in
the nation that invented basketball.
REFERENCES
Arrow, K.J. (1973), 'The theory of discrimination', in
Ashenfelter, O. and Rees, A. (eds), Discrimination in Labor Markets,
Princeton, NJ, Princeton University Press.
Becker, G.S. (1957), The Economics of Discrimination, Chicago,
University of Chicago Press.
Berri, D.J. (2006), 'Economics and the National Basketball
Association: surveying the literature at the tip-off, in Fizel, J.
(ed.), The Handbook of Sports Economics Research, M.E. Sharpe, Inc., pp.
21-48.
--(2008), 'A simple measure of worker productivity in the
National Basketball Association', in Humphreys, B. and Howard, D.
(eds), The Business of Sport, 3 volumes, Westport, Conn., Praeger, pp.
1-40.
Berri, D.J., Brook, S.L. and Fenn, A. (2011), 'From college to
the pros: predicting the NBA amateur player draft 'Journal of
Productivity Analysis, 35, I, February, pp. 25-35
Berri, D.J., Brook, S.L., Fenn, A., Frick, B., and Vicente-Mayoral,
R. (2005), 'The short supply of tall people: explaining competitive
imbalance in the National Basketball Association', Journal of
Economics Issues, 39, 4 (December), pp. 1029-41.
Berri, D.J., Brook, S.L. and Schmidt, M.B. (2007), 'Does one
simply need to score to score?', International Journal of Sport
Finance, 2(4), pp. 190-205.
Berri, D.J. and Schmidt, M.B. (2010), Stumbling on Wins: Two
Economists Explore the Pitfalls on the Road to Victory in Professional
Sports, Princeton, N.J., Financial Times Press.
Berri, D.J., Schmidt, M.B. and Brook, S.L. (2006), The Wages of
Wins: Taking Measure of the Many Myths in Modern Sport, Stanford
University Press (released in paperback in September, 2007).
Berri, D.J., Van Gilder, J. and Fenn, A. (2014), 'Is the
sports media color-blind?', International Journal of Sport Finance,
9(2), pp. 130-48.
Blinder, A.S. (1973), 'Wage discrimination: reduced form and
structural estimates', The Journal of Human Resources, 8(4), pp.
436-55.
Brown, E., Spiro, R. and Keenan, D. (1991), 'Wage and nonwage
discrimination in professional basketball: do fans affect it?'
American Journal of Economics and Sociology, 50(3), pp. 333-45.
Bryson, A., Rossi, G. and Simmons, R. (2014), 'The migrant
wage premium in professional football: a superstar effect?',
Kyklos, 67(1), pp. 12-28.
Camerer, C.F. and Weber, R.A. (1999), 'The econometrics and
behavioral economics of escalation of commitment: a reexamination of
Staw and Hoang's NBA data', Journal of Economic Behavior and
Organization, 39(1), pp. 59-82.
Eschker, E., Perez, S.J. and Siegler, M.V. (2004), 'The NBA
and the influx of international basketball players', Applied
Economics, 36(10), pp. 1009-20.
Fair, R.C. (1994), 'How fast do old men slow down?',
Review of Economics and Statistics, 76(1), pp. 103-18.
Fort, R. and Gill, A. (2000), 'Race and ethnicity assessment
in baseball card markets', Journal of Sports Economics, 1(1)
(February), pp. 21-38.
Frick, B. (2007), 'The football players' labour market:
empirical evidence from the major European leagues', Scottish
Journal of Political Economy, 54(3), pp. 442-6.
--(2009), 'Globalization and factor mobility. The impact of
the "Bosman-Ruling" on player migration in professional
soccer', Journal of Sports Economics, 10(1), pp. 88-106.
Groot, L. (2008), Economics, Uncertainty and European Football:
Trends in Competitive Balance, Cheltenham, Edward Elgar.
Groothuis, P.A. and Hill, J.R. (2013), 'Pay discrimination,
exit discrimination or both? Another look at an old issue using NBA
data', Journal of Sports Economics, 14(2), pp. 171-85.
Hamilton, B.H. (1997), 'Racial discrimination and professional
basketball salaries in the 1990s', Applied Economics, 29(3), pp.
287-96.
Hoffer, A. and Freidel, R. (2014), 'Does salary discrimination
persist for foreign athletes in the NBA?', Applied Economics
Letters, 21(1), pp. 1-5.
Kahn, L.M. (2000), 'The sports business as a labor market
laboratory' Journal of Economic Perspectives, 14(3), pp. 75-94.
--(2012), 'The economics of discrimination: evidence from
basketball', in Schmanske, S. and Kahane, L. (eds), Oxford Handbook
on Sports Economics, Economics Through Sports, Volume 2, Oxford, Oxford
University Press.
Kanazawa, M.T. and Funk, J.P. (2001), 'Racial discrimination
in professional basketball: evidence from Nielsen ratings',
Economic Inquiry, 39(4), pp. 599-608.
Lee, Y.H. and Fort, R. (2012), 'Competitive balance: time
series lessons from the English Premier League', Scottish Journal
of Political Economy, 59(3), pp. 266-82.
McCormick, R.E. and Tollison, R.D. (2001), 'Why do black
basketball players work more for less money?', Journal of Economic
Behavior and Organization, 44(2), pp. 201-19.
Milanovic, B. (2005), 'Globalization and goals: does soccer
show the way?', Review of International Political Economy, 12(5),
pp. 829-50.
Oaxaca, R. (1973), 'Male-female wage differentials in urban
labor markets', International Economic Review, 14(3), pp. 693-709.
Phelps, E.S. (1972), 'The statistical theory of racism and
sexism', The American Economic Review, 64(4), pp. 659-61.
Price, J., Lefgren, L. and Tappen, H. (2013), 'Interracial
workplace cooperation: evidence from the NBA', Economic Inquiry,
51(1), pp. 1026-34.
Price, J. and Wolfers, J. (2010), 'Racial discrimination among
NBA referees '.The Quarterly Journal of Economics, 125(4), pp.
1859-87.
Primault, D. (2007), 'European professional basketball in
crisis: 1992-2002', in Andreff, W. and Szymanski, S. (eds),
Handbook of the Economics of Sport, Edward Elgar Publishing.
Robst, J., Van Gilder, J., Coates, C. and Berri, D.J. (2011),
'Skin tone and wages: evidence from NBA free agents', Journal
of Sports Economics, 12(2), pp. 143-56.
Schroffel, J.L. and Magee, C.S. (2012), 'Own-race bias among
NBA coaches', Journal of Sports Economics, 13(2), pp. 130-51.
Staw, B.M. and Hoang, H. (1995), 'Sunk costs in the NBA: why
draft order affects playing time and survival in professional
basketball', Administrative Science Quarterly, 40(3), pp. 474-94.
Stone, E.W. and Warren, R.S. (1999), 'Customer discrimination
in professional basketball: evidence from the trading-card market',
Applied Economics, 31 (6), pp. 679-85.
Yang, C.H. and Lin, H.Y. (2012), 'Is there salary
discrimination by nationality in the NBA? Foreign talent or foreign
market '.Journal of Sports Economics, 13(1), pp. 53-75.
NOTES
(1) The United States dominated the first Olympic competitions,
winning the gold medal in each of the summer games played from 1936 to
1968. In 1972 the Soviet Union managed to win a gold, an outcome
repeated in the Soviet Union in 1988. And Yugoslavia won in 1980 (when
the US boycotted the game). Professionals began playing in 1992, and
with that change the USA again returned to dominance. Except for the
2004 Olympics (where the gold went to Argentina), the gold medal has
gone to Team USA in each of the games played from 1992 to 2012. For more
see http://www.sports-reference.com/olympics/sports/
BAS/mens-basketball.html
(2) http://www.nytimes.com/2012/02/05/magazine/NBA-in China.
html?n=Top/Reference/Times per cent20Topics/People/Y/ Yardley, per
cent20Jim?ref=jimyardley&_r=1&pagewanted=all.
(3) http://www.forbes.com/sites/aliciajessop/2012/06/14/the-surge-of-the-nbas-international-viewership-and- popularity/.
(4) http://espn.go.eom/blog/truehoop/post/_/id/19814/is-basketball-taking-over-the-world.
(5) During the winter of 1891-92, James Naismith invented the sport
of basketball at Springfield College in Massachusetts. For the story of
the origins of basketball one is referred to the website of Springfield
College: http://www.springfieldcollege.
edu/welcome/birthplace-of-basketball/index#.VFUn5_nF_h4.
(6) Discrimination against individuals due to their origin or
religious beliefs remains a phenomenon in modern societies. Becker
(1957) outlines potential sources of discrimination in labour markets
where workers are treated differently despite being equally qualified.
The availability of workers' performance, individual
characteristics and salaries in professional sports data means it is a
fruitful setting in which to test for labour market discrimination
(Kahn, 2000).
(7) http://www.eeoc.gov/laws/types/nationalorigin.cfm.
(8) The international players employed by the NBA examined in our
dataset are drawn from 54 different nations. European nations are the
most common nationality observed (outside the USA). But players are also
drawn from Africa, South America, Asia, Australia, and other North
American nations. The international search for talent is likely related
to the issue of the "short supply of tall people" (see Berri
et al., 2005).
(9) For the ACB we are relying on data from RealGM.com. This site
reports for the 2011/12-2013/14 seasons data on performance and personal
characteristics of the players employed (including age and national
origin). For the NBA we will be utilising data from
basketball-reference.com. This site also reports performance data and
personal characteristics.
(10) The liberalisation of the labour market in sports was followed
in detail by the economic literature to analyse the impact on player
migration and remuneration as well as its consequences for clubs and
leagues (e.g. Milanovic, 2005; Frick, 2007, 2009; Groot, 2008; Lee &
Fort, 2012).
(11) Berri (2006) offered an extensive literature review of past
studies. This review illustrated that results varied in the literature.
More recently, Robst et al. (2011) failed to find evidence of
discrimination while Groothuis and Hill (2013)--in a comprehensive study
of wage and employment discrimination in the NBA--noted that evidence of
racial discrimination depended on how the model employed was
constructed.
(12) This point was made by Fort and Gill (2000). It was also noted
in a study of bias in the sports media by Berri, Van Gilder, and Fenn
(2014).
(13) However, nationality can be changed by the player to
circumvent restrictions for usage of foreign players. If a player has
citizenship for the country he is playing in, he is treated as a
domestic in our data.
(14) In addition--as noted in Berri and Schmidt (2010) and Berri,
Schmidt and Brook (2006)--the evaluation of this past performance is
inconsistent with how that performance actually impacted outcomes in the
NBA.
(15) Price et al. (2013) do not find any own-race bias for players,
showing that cooperative behaviour on the basketball court is
independent of teammates' race.
(16) Performance relative to position is calculated following the
approach taken originally in Berri, Schmidt, and Brook (2006).
Specifically, the average per 48 minute performance in a player's
primary position was subtracted from his per 48 minute performance. Then
the overall average per 48 minute performance (across all positions) was
added back to the outcome of the first step. This gives us a measure of
performance that can allow comparisons across all positions. Without
this step the inclusion of players at different positions is
problematic, since performance varies systematically across position
played. Specifically, big men tend to get rebounds and not many assists
or steals. Smaller players tend to get fewer rebounds and more assists
and steals.
(17) Games played are included as a proxy for injury. But it also
reflects the evaluation of the coach.
(18) The squared term is necessary because we suspect a younger
player might earn more minutes with experience while older players might
see playing time decline with age (Fair, 1994).
(19) Prior research into this topic was offered by Staw and Hoang
(1995) and Camerer and Weber (1999). Each of these authors examined the
escalation of commitment in the NBA, defined by Camerer and Weber (pp.
59-60) as follows: "when people or organizations who have committed
resources to a project are inclined to 'throw good money after
bad' and maintain or increase their commitment to a project, even
when its marginal costs exceed marginal benefits." Both of the
aforementioned studies offer evidence that, after controlling for the
performance of the player, where a player was chosen in the draft still
impacts the amount of playing time the player receives after the first
two years of the player's career and the completion of the rookie
contract. Such a result indicates that NBA decision-makers cannot let
sunk costs go.
(20) The estimation of this model included team specific fixed
effects and robust standard errors.
(21) The model indicates minutes per game reach a maximum at 26.6
years of age.
(22) TOPER is found to be positive and significant. This means that
players are rewarded for being more careless with the ball. Many
'stars' tend to score in large numbers but are also somewhat
careless with the ball, a result that is consistent with a failure of
decision-makers to evaluate inefficient stars (as detailed in Berri and
Schmidt, 2010).
(23) The estimation of this model includes team specific fixed
effects and robust standard errors.
(24) The model indicates that minutes per game peak around 32.1
years of age in the ACB.
(25) This was also noted in Berri, Schmidt, and Brook (2006),
Berri, Brook, and Schmidt (2007), and Berri and Schmidt (2010). The
focus on scoring has also been seen in past studies of the NBA draft
(see Berri, Brook, and Fenn, 2010) and the media's voting for the
NBA's MVP award (see Berri, Fenn, and Van Gilder, 2012).
(26) Even if no fan preference exists towards a certain type of
player it can still be favourable to award additional playing time in
response to referee bias. Such behaviour would be consistent with Price
and Wolfers (2010), who offer evidence that calls made by referees are
impacted by the race of the players in the game.
David J. Berri, Department of Economics and Finance, Southern Utah
University. E-mail: berri@suu.edu.
Christian Deutscher, Department of Sport Science, University of
Bielefeld. E-mail: christian.deutscher@uni-bielefeld.de.
Arturo Galletti, Senior Calibration Manager, Coopervision, Juana
Diaz, and contributor to www.boxscoregeeks.com. E-mail:
Arturo.galletti@gmail.com.
Table 1. Percentage of minutes played by players of
different national origins in the ACB: 2011 /12-2013/14
Season Spain USA Non Spain/Non USA
(%) (%) (%)
2013/14 34.5 22.8 42.6
2012/13 32.7 27.7 39.6
2011/12 32.0 26.9 41.0
Average 33.1 25.8 41.1
Source: RealGM.com.
Table 2. Descriptive statistics of variables in equation (1) for the
NBA 2001 /2-2013/14
Dependent variable Label Mean
Minutes per game MGM 23.837
Independent variables
Points (a) PTS 18.486
Adjusted Field Goal Percentage (b) ADJFG 0.486
Free Throw Percentage (a) FT 0.738
Rebounds (a) REB 9.145
Turnover Percentage (d) TOPER 13.764
Steals (a) STL 1.361
Assists (a) AST 2.894
Blocked Shots!3) BLK 1.149
Personal Fouls!3) PF 4.729
Draft Position, year 1 DFT1 3.337
Draft Position, year 2 DFT2 3.239
Draft Position, year 3 DFT3 2.839
Draft Position, year 4 DFT4 2.732
Draft Position, year 5 DFT5 2.332
Draft Position, year 6 DFT6 2.052
Draft Position, year 7 DFT7 1.844
Draft Position, year 8 DFT8 1.556
Draft Position, year 9 DFT9 1.426
Draft Position, year 10 DFT10 l.117
Height (a) Height 79.012
Age Age 26.578
Age, Squared SQAge 724.409
Games Played GM 63.363
Dummy Variable, Bom in USA DUSA 0.812
SD Minimum
Minutes per game 8.993 6.091
Independent variables
Points (a) 5.462 3.31
Adjusted Field Goal Percentage (b) 0.053 0.224
Free Throw Percentage (a) 0.112 0
Rebounds (a) 2.005 1.32
Turnover Percentage (d) 3.997 1.755
Steals (a) 0.520 0.023
Assists (a) 1.667 -3.792
Blocked Shots!3) 0.707 -0.825
Personal Fouls!3) 1.355 1.309
Draft Position, year 1 1 1.415 0
Draft Position, year 2 11.092 0
Draft Position, year 3 10.322 0
Draft Position, year 4 9.386 0
Draft Position, year 5 9.386 0
Draft Position, year 6 8.739 0
Draft Position, year 7 8.350 0
Draft Position, year 8 7.608 0
Draft Position, year 9 7.287 0
Draft Position, year 10 6.254 0
Height (a) 1.617 70.182
Age 4.248 18
Age, Squared 236.133 324
Games Played 15.520 30
Dummy Variable, Bom in USA 0.391 0
Maximum
Minutes per game 43.7
Independent variables
Points (a) 40.073
Adjusted Field Goal Percentage (b) 0.729
Free Throw Percentage (a) 1
Rebounds (a) 19.537
Turnover Percentage (d) 36.188
Steals (a) 3.581
Assists (a) 10.912
Blocked Shots!3) 6.35
Personal Fouls!3) 13.104
Draft Position, year 1 61
Draft Position, year 2 61
Draft Position, year 3 61
Draft Position, year 4 61
Draft Position, year 5 61
Draft Position, year 6 61
Draft Position, year 7 61
Draft Position, year 8 61
Draft Position, year 9 61
Draft Position, year 10 61
Height (a) 85.754
Age 41
Age, Squared 1681
Games Played 82
Dummy Variable, Bom in USA /
Notes: (a) Each of these variables is calculated relative to position
played, (b) Adjusted field goal percentage--or effective field goal
percentage--is calculated as follows: [(Points--Free Throws Made)-
Field Goal Attempts], This measure takes into account that players can
shoot from two point and three point range. It is detailed at
basketball-reference.com. (c) Free throw percentage = free throws
made-free throws attempted, (d) Turnover percentage is calculated as
follows: [(Turnovers)-(Turnovers + Field Goal Attempts + 0.44 * Free
Throw Attempts)] * 100. This measure--detailed at basketball-
reference.com--allows one to compare turnovers across a sample of
players who handle the ball more or less.
Table 3. Explaining minutes per game for the NBA: 2001 /2-2013/14
Independent variable Label Coefficient
Points PTS 0.413 ***
Adjusted Field Goal Percentage ADJFG 8.802 ***
Free Throw Percentage FT 8.109 ***
Rebounds REB 0.347 ***
Turnover Percentage TOPER 0.049 *
Steals STL 0.841 ***
Assists AST 0.823 ***
Blocked Shots BLK 0.437 **
Personal Fouls PF -2.478 ***
Draft Position, year 1 DFT1 -0.114 ***
Draft Position, year 2 DFT2 -0.088 ***
Draft Position, year 3 DFT3 -0.062 ***
Draft Position, year 4 DFT4 -0.057 ***
Draft Position, year 5 DFT5 -0.049 ***
Draft Position, year 6 DFT6 -0.039 ***
Draft Position, year 7 DFT7 -0.021
Draft Position, year 8 DFT8 -0.026 *
Draft Position, year 9 DFT9 0.009
Draft Position, year 10 DFT10 -0.009
Height Height 0.329 ***
Age Age 2.012 ***
Age, Squared SQAge -0.038 ***
Games Played GM 0.134 ***
Dummy Variable, Bom in USA DUSA 1.159 ***
Constant term -50.598 ***
Season Dummies included
Observations 4521
R-squared 0.67
Independent variable Standard t-statistic
error
Points 0.022 19.05
Adjusted Field Goal Percentage 1.862 4.73
Free Throw Percentage 1.098 7.38
Rebounds 0.063 5.51
Turnover Percentage 0.027 1.78
Steals 0.254 3.31
Assists 0.072 11.39
Blocked Shots 0.167 2.61
Personal Fouls 0.083 -29.72
Draft Position, year 1 0.009 -12.82
Draft Position, year 2 0.008 -10.93
Draft Position, year 3 0.009 -6.73
Draft Position, year 4 0.008 -6.78
Draft Position, year 5 0.010 -5.01
Draft Position, year 6 0.011 -3.44
Draft Position, year 7 0.016 -1.31
Draft Position, year 8 0.014 -1.79
Draft Position, year 9 0.015 0.60
Draft Position, year 10 0.014 -0.61
Height 0.067 4.90
Age 0.312 6.45
Age, Squared 0.006 -6.82
Games Played 0.008 16.84
Dummy Variable, Bom in USA 0.226 5.12
Constant term 5.484 -9.23
Season Dummies
Observations
R-squared
Notes: * significant at the 10 per cent level, ** significant at the 5
per cent level. ***significant at the 1 per cent level.
Table 4. Decomposition results for the NBA: 2001/2-
2013/14
Independent variable Coeff. Standard z-stat.
error
USA 24.08 *** 0.15 160.40
Non USA 22.77 *** 0.29 78.83
Difference 1.31 *** 0.33 4.03
Explained -0.08 0.29 0.28
Unexplained 1.39 *** 0.201 6.91
Observations total 4521
Observations USA 3671
Observations non USA 850
Notes: * significant at the 10 per cent level, ** significant at the 5
per cent level. *** significant at the 1 per cent level.
Table 5. Descriptive statistics of variables in equation (1) for the
ACB 2011/12-2013/14
Dependent variable Label Mean SD
Minutes per game MGM 19.830 5.887
Independent variables
Points PTS 14.326 4.041
Adjusted Field Goal ADJFG 0.506 0.07
Percentage
Free Throw Percentage FT 0.75 0.12
Rebounds REB 5.456 1.716
Turnover Percentage TOPER 16.341 5.348
Steals STL 1.429 0.485
Assists AST 2.986 1.254
Blocked Shots BLK 0.375 0.535
Personal Fouls PF 4.096 1.359
Height Height 78.349 1.52
Age Age 27.76 4.277
Age, Squared SQAge 788.881 237.931
Games Played GM 31.348 6.994
Dummy Variable, Bom in USA DUSA 0.231 0.422
Dummy Variable,
Born outside Spain (not USA) DnonSpain 0.417 0.493
Dependent variable Minimum Maximum
Minutes per game 5.135 33.815
Independent variables
Points 2.548 26.133
Adjusted Field Goal 0.211 0.728
Percentage
Free Throw Percentage 0 1.000
Rebounds -0.641 11.813
Turnover Percentage 0 41.91279
Steals 0.059 3.078
Assists -1.746 8.28
Blocked Shots -0.843 3.234
Personal Fouls 1.108 13.099
Height 69.0 87.00
Age 18.0 39.0
Age, Squared 324.0 1521.0
Games Played 15.0 46.0
Dummy Variable, Bom in USA 0.0 1.0
Dummy Variable,
Born outside Spain (not USA) 0.0 1.0
Table 6. Explaining minutes per game for the ACB: 2011/12-2013/14
Independent variable Label Coefficient
Points PTS 0.397 ***
Adjusted Field Goal Percentage ADJFG 5.592
Free Throw Percentage FT 3.076 *
Rebounds REB 0.1556
Turnover Percentage TOPER -0.075
Steals STL 1.047 **
Assists AST 0.908 ***
Blocked Shots BLK 0.247
Personal Fouls PF -1.429 ***
Height Height -0.021 ***
Games Played GM 1.554 ***
Age Age 1.470 *
Age, Squared SQAge -0.023 *
Dummy Variable, Born in USA DUSA 1.791 ***
Dummy Variable, Born outside
Spain (not USA) DnonSpain 0.963***
Constant Term -0.464
Season dummies included
Observations 596
R-squared 0.487
Independent variable Standard t-statistic
error
Points 0.053 7.51
Adjusted Field Goal Percentage 3.23398 1.69
Free Throw Percentage 1.549 2.04
Rebounds 0.099 1.58
Turnover Percentage 0.036 -2.11
Steals 0.391 2.68
Assists 0.190 4.79
Blocked Shots 0.311 0.79
Personal Fouls 0.136 -10.70
Height 0.061 -3.42
Games Played 0.034 4.56
Age 0.705 2.08
Age, Squared 0.013 -1.79
Dummy Variable, Born in USA 0.445 4.02
Dummy Variable, Born outside
Spain (not USA) 0.289 3.33
Constant Term 12,225 0.97
Season dummies
Observations
R-squared
Notes: * significant at the 10 per cent level, ** significant at the 5
per cent level. *** significant at the I per cent level.
Table 7. Decomposition results for the ACB: 2011/12-2013/14
Independent variable
Coeff. Standard z-stat.
error
Spain vs non Spain
Spain 18.77 *** 0.42 43.9
non Spain 20.43 *** 0.28 71.83
Difference -1.66 *** 0.51 -3.23
Explained -0.42 0.41 -1.01
Unexplained -1.24 *** 0.4 -3.14
Obs. total 596
Obs. Spain 209
Obs. non Spain 387
Independent variable
Coeff. Standard z-stat.
error
Spain vs USA
Spain 18.77 *** 0.42 43.9
USA 22.05 *** 0.47 47.28
Difference -3.28 *** 0.63 -5.18
Explained -1.75 *** 0.5 -3.5
Unexplained -1.53 *** 0.51 -3.01
Obs. total 347
Obs. Spain 209
Obs. USA 138
Notes: * significant at the 10 per cent level. ** significant at the 5
per cent level. *** significant at the 1 per cent level.
Table 8. Impact of one standard deviation on minutes per
game in NBA and ACB
Independent variable Impact of a one Impact of a one
standard standard
deviation change deviation change
in NBA sample in ACB sample
Personal fouls -3.22 -1.99
Points 2.48 1.65
Assists 1.3 1.08
Rebounds 0.63 not significant
Free throw percentage 0.47 not significant
Adjusted field goal percentage 0.43 0.46
Steals 0.34 0.5
Turnover percentage 0.28 not significant
Blocked shots 0.22 not significant
Note: Estimated coefficient from above regressions must be significant
at the 10 per cent level or better to be included in table.