Entry discrimination in the NHL:Evolution and the KHL effect. w

advertisement
Entry
discrimination in
the NHL:
Evolution and the
KHL effect
Tom Christie
and Marc Lavoie
Preliminaries
• There have been lots of studies on salary
determination in the NHL, and by extension on salary
discrimination
• There have been fewer studies on the issue of entry
discrimination in the NHL.
• This discrimination could affect Europeans,
Americans and French Canadians, to the benefit of
English Canadians
• One way of ascertaining the presence of entry
discrimination is to compare the future performance
of players relative to the rank at which they were
drafted.
Click View then Header and Footer to change this text
DISCRIMINATION IN THE NHL
Quebec Hockey Players Sidelined
Bob Sirois
2009, 2010
Click View then Header and Footer to change this text
Entry and hiring discrimination
• “When I talk about discrimination, I’m talking about equal or
comparable talent. Most NHL teams, at equal talent or
comparable talent, will not take a Quebecer, will not take a
European, they will not even take an American player, they will
take a good ol’ Canadian boy.”
• …. Former Devils Star Bobby Holik replied, “I agree with him
100%. Because I’ve experienced it myself where I believe I had
to fight for my spot, fight for my ice time, more and longer than
the Western Canadians or Ontario-born and raised.”
• Bob Sirois, author of Discrimination in the NHL
Montreal, November 16, 2010
Click View then Header and Footer to change this text
Prejudices and poor information
• “Talent scouting is much more a question of
intuition, feeling or even gut instinct. Myths,
prejudices, stereotypes and favouritism are
an integral part of each National Hockey
League draft”
• (Sirois 2010, p. 30).
Click View then Header and Footer to change this text
Previous studies, with simplest model
Dependent variable: career points per game
Lavoie et al.
1987
Lavoie and
Grenier 1992
McLean and
Veall 1992
Lavoie 2003
1983-84
R² = .27
N= 362
1989-90
AdjR² = .22
1990-91
R² = .20
N = 280
1993-94
AdjR² = .16
N = 281
Constant
.752
.745
.737
.668
Draft Number
− .00467
(5.89)***
− .00372
(3.79)***
− .00411
(4.28)***
− .00228
(2.90)***
Draft Number²
+.0000205
(4.54)***
+.0000132
(2.51)**
+.0000167
(3.16)***
+.0000110
(2.17)**
Defense
− .266
(9.59)***
− .271
(7.49)***
− .270
(7.78)***
− .230
(6.18)***
French
Canadian
+ .110
(2.61)***
+ .089
(1.54)
+ .064
(1.43)
+ .122
(2.44)**
Click View then Header and Footer to change this text
Have things changed in nearly 20 years?
• 20 years have gone by
• People may hold less prejudices
• Financial constraints on teams limited by the salary
caps may induce teams to be more efficient in their
scouting and hiring decisions
• New technology allows scouts to get access to more
information and more visuals of prospective draftees,
thus reducing uncertainty in appraisal
• Hence, with less uncertainty, less biased decisions
should be taken
Click View then Header and Footer to change this text
Dependent variable: career points per game
2009-2010
R2 = 0.236, N = 295
Constant
0.550***
(0.0233)
Draft Number
-0.00281***
(0.000531)
Draft Number2
0.00000785***
(2.23e-06)
Defense
French Canadian
-0.151***
(0.0255)
0.0271
(0.0362)
Click View then Header and Footer to change this text
It would seem that scouting is more
efficient ….
1993 – 1994
y = -0.00228x + 0.0000110x2
∂y/∂x = -0.00228 +
2(0.0000110)x
Set ∂y/∂x = 0
Therefore draft rank where y is
minimized:
x = 0.00228/0.000022
= 103.64
2009 – 2010
y = -0.00281x + 0.00000785x2
∂y/∂x = -0.00281 +
2(0.00000785)x
Set ∂y/∂x = 0
Therefore draft rank where y is
minimized:
x = 0.00281/0.00000785
= 178.98
Click View then Header and Footer to change this text
Two more more complex model (constant and draft
rank left out of results) ….
Defence
Defensive Play
(1)
2009-2010
R2 = 0.237
N = 295
(2)
2009-2010
R2 = 0.301
N = 295
-0.154***
(0.0272)
-0.134***
(0.0270)
0.00947
(0.0266)
0.0253
(0.0261)
Height
-0.0261***
(0.00784)
Weight
0.000941
(0.00110)
Penalties
-0.0578**
(0.0223)
French Canadian
0.0281
(0.0363)
Click View then Header and Footer to change this text
0.0212
(0.0350)
What about Europeans and Americans?
• More information available than in the past, thus less
uncertainty, and less possible bias.
• However, European leagues may be more developed
than in the past, in particular the Russian KHL, so
that NHL teams run the risk of seeing their European
draft picks declining to play in North America.
• Repeat regression (2), adding Europeans and
Americans to the sample of 2009-2010 players
Click View then Header and Footer to change this text
How about Europeans and Americans?
1993-1994
R2 = 0.18
N = 436
2009-2010
R2 = 0.285
N = 561
French Canadian
0.116***
(0.044)
0.015
(0.0358)
European
(Russians included)
0.135***
0.071***
(0.038)
(0.0222)
0.025
(0.038)
0.010
(0.0224)
American
Click View then Header and Footer to change this text
How about isolating the Russians?
2009-2010
R2 = 0.311
N = 561
French Canadian
Non- Russian European
0.0154
(0.0356)
0.0884***
(0.0221)
American
0.00995
(0.0223)
Russian
0.194***
(0.0423)
Click View then Header and Footer to change this text
Conclusions
• There is still no apparent entry discrimination against
American hockey players.
• Entry discrimination against French Canadian hockey
players seems to have disappeared.
• There is still apparent entry discrimination against
non-Russian European hockey players, but less so
than before.
• There is a strong and obvious KHL effect against
Russian hockey players, which discourages NHL
teams from drafting young Russian players as early
as they should.
Click View then Header and Footer to change this text
Download