This paper analyzes the impact of the NHL's participation in the Winter Olympics on competitive balance outcomes within the NHL. It finds that the post-Olympic performance of NHL teams is negatively related to the number of players that the team supplied to the various Olympic rosters. This is consistent with a notion that participating in the Olympics can induce greater fatigue in players, thereby reducing their effectiveness upon their return to their domestic clubs. This effect was found to be particularly strong for players representing the host country at the Olympics.
Keywords: Olympics, NHL, competitive balance
Countries participating in premiere international sport competitions typically draw their players from club teams. For example, in football, clubs from around the world release their players to play in competitions such as the FIFA World Cup and the UEFA European Championships; in basketball, National Basketball Association (NBA) teams allow their players to participate in the summer Olympics, etc. In these particular cases, the international competitions occur during a time of year (summer) when the professional leagues are in their off-seasons. Thus, players have some time to physically and mentally recover from these international competitions before their domestic professional leagues begin play.
In ice hockey, however, the premiere international competition--the Winter Olympics--occurs at a time when the National Hockey League (NHL), the top professional league in the world, is in the midst of its season. Prior to the 1998 Olympics, the NHL did not allow its players to participate in the Olympics; thus, the rosters of Olympic hockey teams were comprised of players not in the NHL. However, for the past four Winter Olympics (1998, 2002, 2006, and 2010) the NHL has ceased operations for an almost three-week period and allowed its players to play for their respective countries in the Olympics.
Within the NHL, this decision to participate in the Olympics has been controversial, and there is now considerable uncertainty as to whether the NHL will allow its players to participate in the 2014 Winter Olympics in Sochi, Russia. The Olympic-induced NHL shutdown generally occurs in February, about two-thirds of the way through the NHL regular season. While the NHL has benefitted from the publicity it receives from allowing its players to participate in the Olympics, many within the League apparently feel that the mid-season shutdown is too disruptive to the League and its players.
One specific concern is that players who participate in the Olympics may become over-fatigued, making them more prone to injury or poor performance after they return to the NHL--this is particularly true because players have almost no time to rest either before or after the Olympics. Also, both the 1998 (Japan) and 2006 (Italy) Games were outside of North America, and hence involved large amounts of travel; this travel, combined with the grueling Olympic schedule--with as many as eight games over about 17 days--has the potential to take a heavy toll on players.
In this context, the purpose of this paper is to determine whether the Olympics have any competitive balance effects in the NHL. In particular, the paper asks whether NHL teams that supply a greater number of players to Olympic rosters are then put at a competitive disadvantage once the NHL season resumes. This is potentially an important question as the NHL considers its future participation in the Olympics--if the NHL's participation in the Olympics has the unintended effect of favoring one team over another, then the League will also have to weigh such impacts when making its decision.
This paper uses a panel data set encompassing the past four Winter Olympics to test for these possible competitive balance (CB) effects. An analysis of the data first reveals that participation in the Olympics is not uniform across NHL teams--as might be expected, the better-performing NHL teams in a given season tend to produce the greatest number of Olympic players. The question, then, is whether such differences in participation impact the NHL team's post-Olympic performance.
To test this, the formal empirical model contains a dependent variable that measures the difference between each NHL team's pre-Olympic and post-Olympic performance. This difference in performance is then regressed on a series of control variables, along with a variable measuring the number of Olympians on the team. The results indicate that this latter variable is both negative and significant, indicating that teams with a greater number of Olympians experience a decrease in team performance following the Olympics, relative to that team's performance preceding the Olympics. This lends support to the "fatigue" theory--that NHL players that participate in the Olympics experience increased levels of physical and mental fatigue, and thus exhibit declines in performance once they return to the NHL. This, in turn, harms the performance of the teams that employ these players. To the extent, then, that Olympians are not spread uniformly across all NHL teams, but are more likely to come from high-performing teams, these high-performing teams may be disadvantaged by the NHL's participation in the Olympics.
International Competitions and Competitive Balance Implications
Competitive balance has been one of the most studied issues in sport economics over the past two decades. The literature in the area is now vast, and spans several different sub-areas--numerous review articles now exist, including Humphreys and Watanabe (forthcoming), Fort (2006), and Sanderson and Siegfried (2003).
A major focus of the literature has been to develop mechanisms to define and measure CB. No single, universally accepted measurement has emerged; for example, Fort (2006) categorizes three broad approaches to measuring CB--game uncertainty, playoff uncertainty, and consecutive-season uncertainty. Game uncertainty pertains to balance within a given season, and has typically been measured by the standard deviation of win-percent of teams in a league, relative to the idealized (i.e., equal balance) standard deviation. As Fort notes, however, if fans are more sensitive to the tails/extremes of the distribution rather than to win-percents around 0.50, then instruments such as Herfindahl indices (Humphreys, 2002) and Gini coefficients (Schmidt & Berri, 2001) may be more appropriate measures of balance. With playoff uncertainty, one research track has focused on the closeness of playoff races (see, for example, Krautmann, Lee, & Quinn, 2011), while another broad approach has been to ex-post analyze the concentration of championships within a league (see Quirk & Fort, 1997). Lastly, consecutive-season uncertainty is concerned with the turnover in league outcomes from season to season--in this regard, Humphreys (2002) develops a metric that simultaneously captures both the within-season variation of teams within a league, and the team-specific variation in league standings across seasons.
Given these various measures of CB, a parallel stream in the literature has attempted to determine the degree to which CB has changed across time within a given league, and/or the extent to which it has varied across leagues. Focusing here on only hockey, Fort (2006) finds, using a within-season measure of CB, that balance in the NHL reached a low point during the 1970s, but steadily increased after that era, to the point where, by the 2000s, it was only marginally lower than the NFL's. Somewhat similarly, Berri et al. (2005) examine average CB (using the standard deviation of win-percents, relative to the ideal) across the four major North American leagues since their respective inceptions, and find that the NHL has had the second-highest level of balance, trailing only the NFL. They also find that all of the major professional sports leagues, with the exception of the NBA, have better balance after 1990, compared to pre-1990.
While most of the literature focuses on factors internal to leagues (disparities in market sizes across franchises, the presence of salary caps and revenue sharing mechanisms, etc.) when analyzing CB outcomes, this paper takes a somewhat different approach by examining how CB within a league might be impacted by factors external to the league itself. In particular, it investigates the role that elite international competitions might have on CB outcomes in domestic leagues. These competitions potentially impact league-play because national squads draw their players from domestic clubs; thus, if a player's future performance is somehow impacted (either positively or negatively) by his participation in these international events, then the player's domestic team will also be impacted. If some domestic teams supply more players to international competitions than do other teams, then the effects of these international competitions are not neutral with respect to domestic play; some teams will be impacted more than others.
Taking this another step further, one might expect that high-talented (and often large-market) domestic clubs will supply more players to international rosters than will less-talented clubs--for example, Arsenal will generally supply more players to World Cup squads than would, say, Sunderland. To the extent this is true, the domestic CB impacts of these international competitions depend upon whether subsequent player-performance is either enhanced by or hampered by their international participation. If it is the former, then international events would decrease domestic CB--the best domestic teams supply the most international players, these players future performance is then further enhanced by their international participation relative to those who didn't participate, thus giving the player's domestic club a further advantage over those clubs who supplied fewer international players. Conversely, if participating in international events reduces the future performance of players, then international competitions serve to increase CB--the best domestic clubs are more negatively impacted than are their other league competitors.
The ultimate question, then, is whether international events enhance or hamper the future performance of those players who participate. In general, there are two possible effects. Since players who participate in international competitions must play in more games, often in high-pressure situations, they may experience additional fatigue--both physical and mental--thereby reducing their effectiveness upon their return to their domestic clubs. Conversely, the excitement and aura that surround many international events can be exhilarating and inspiring, and these psychic benefits may have residual positive effects on the player after his return to domestic competition. For economists, it largely becomes an empirical question as to which of these effects are greater.
However, there is also an additional factor at work here--the specific nature of the international competition itself may matter. Both the timing and frequency of the competition are potentially important. In general, for example, European football players are more frequently involved with international competitions than are their counterparts in the four major North American professional leagues. For European football players, there is a major international competition every two years--alternating between the World Cup and the European Championships. While these competitions occur during the off-season of domestic leagues, the qualifying rounds for these tournaments generally occur during the domestic season. These qualifying games are, however, spread-out across a long time period, usually many months, and do not have the same compressed intensity that accompany the major competitions themselves. Nevertheless, given the frequency of international competitions in European football, one might expect such competitions to have an externality effect on domestic play. Curiously, there is almost a complete absence of any literature that examines this issue.
In contrast to Europe, for players in North American sports, playing in major international events is a relatively recent occurrence. In fact, for NFL players, there is no international competition--American-style football is relatively unique to North America, and not played at a sufficiently high level in most countries in the world to make an international competition viable. In baseball, MLB players have never participated in the Olympics, largely because the Summer Olympics occur at a time when the MLB season is still in progress. In order to help build a stronger international component to the game, MLB introduced the World Baseball Classic in 2006. Teams from 16 countries participated in the tournament, with MLB players comprising many of those participating. The event was held again in 2009, and is scheduled to occur every four years. The event occurs in March, just before the start of the MLB season. Despite MLB's best efforts to promote the event, some MLB players have opted to by-pass the tournament, and the event has not yet gained the credibility and publicity of international tournaments in other sports. In the NBA, players have been participating in the Olympic basketball tournament since the 1992 Barcelona Games. That followed the 1989 ruling by the International Basketball Federation (FIBA) to allow professionals to participate in the Olympics.
In hockey, NHL players first began participating in the Olympics in the 1998 Nagano Games, and have participated in the three Olympics since then--2002 in Salt Lake City, 2006 in Torino, Italy, and 2010 in Vancouver. However, unlike the aforementioned situations with soccer, basketball, and baseball, the Olympic hockey tournament occurs while the NHL's season is in progress. For the past four Olympics, the NHL has taken an approximately 17-day break from its schedule to allow its players to participate for their respective countries. This break occurs in mid-to-late February, about two-thirds of the way through the NHL regular season.
The NHL's continued participation in the Olympics is now being re-examined by the League, particularly with the 2014 Games being in Sochi, Russia, and thus necessitating long travel times for NHL players who participate. The NHL's original goal in participating in the Olympics was to increase the profile of the League--particularly in Europe, but also in the US, where the game of hockey is not nearly as popular on a national level as it is in Canada--before a world audience. However, some League executives have now questioned the lasting impact of this publicity, and combined with the disruption and loss of momentum caused by shutting down NHL play for an extended time, the prospects for the NHL participating in the Sochi Games remains very uncertain. (1) An additional and ongoing concern is the risk of injury to players at the Olympics, and its potential subsequent impact on the player's club-team once the Olympics are over.
It is within this context that this paper conducts its empirical analyses. Specifically, it examines whether the NHL's participation in the Olympics has potential CB implications for the League. In particular, it asks whether teams that supply the most players to Olympic rosters--that generally being the NHL's best-performing teams--are subsequently advantaged or disadvantaged once the NHL season resumes. (2) The fact that the Olympic tournament occurs during the NHL season, rather than in the offseason, would seem to increase the likelihood that some type of effect--either positive or negative--would be found.
Data and Model
Table 1 provides information on the Olympic participation of NHL players over the four Olympic years under study. On average, about 4.47 players from each NHL team participated in the Olympics, representing about 22% percent of an NHL team's 20 player roster. (3) However, as hypothesized earlier, this participation was not even across teams. For example, the top five NHL teams--as measured by goal differential at the time the Olympics began--produced an average of 6.40 players, whereas the bottom five NHL teams produced only 2.90 players. This disparity is particularly evident in the 2002 and 2006 seasons, where the top five teams produced almost five more Olympians than did the bottom five.
The empirical analyses then attempts to assess the extent to which these disparities in participation impact--either positively or negatively--the post-Olympic performance of NHL teams. The general form of the model is given in Equation 1.
[GoalDifChng.sub.i] = [B.sub.0] + [B.sub.1] [Games.sub.i] + [B.sub.2][Home.sub.i] + [B.sub.3] [Conf.sub.i] + [B.sub.4] [PrevYr.sub.i] + [B.sub.5][Playoffs.sub.i] + [B.sub.6] [GoalDif.sub.i] + [B.sub.7] [Olympians.sub.i] (1)
The dependent variable, GoalDifChng, measures team i's change in performance from the pre-Olympic period to the post-Olympic period. Following Kahane, Longley, and Simmons (forthcoming), team performance is measured by goal differential--i.e., the difference between goals-for and goals-against. They note that, with the NHL's introduction of the shootout in 2005, win-percent may no longer be the best measure of team performance, since teams often gain a "cheap" point in the standings simply by winning the shootout contest. The variable GoalDifChng then measures the difference between the team's goal differential per game after the Olympics, relative to its goal differential per game prior to the Olympics. Positive values indicate the club's performance increased following the Olympics, while negative values indicate a postOlympic drop in performance.
With the independent variables, Olympians is the focus variable, with all other variables acting as controls. Games measure the number of games that a team has remaining following the Olympics. It attempts to account for any pre- and post-Olympics imbalance in the team's schedule. In the approximately seven weeks of the regular season that remain following the Olympics, some teams had as many as 27 games remaining in their 82-game schedule, while other teams had as few as 19 games remaining. The hypothesis is that teams that have to compress more games into the post-Olympic time period are more susceptible to fatigue, thus having possible detrimental effects on their performance. In a similar vein, Home measures the proportion of a team's postOlympics games that are played at home. Teams that played a disproportionate number of games at home in the pre-Olympic period must then play a disproportionate number of games away from home after the Olympics. Some teams played as many as 64% of their post-Olympic games at home, while others played as few as 36%. Like Games and Home, Confalso attempts to capture schedule imbalances--it measures the proportion of post-Olympic games that are played against teams from the same conference. The idea is that, since teams compete for playoff positions only against teams from the same conference, teams may have additional motivation to compete harder in within-conference games. Values for Confin the data set range from 0.50 to 1.00.
The variable PrevYr is a dummy variable set equal to one if the team made the playoffs in the previous year, and zero otherwise. The notion is that teams that made the playoffs in the previous year may be more adept and experienced at navigating the late-season pressures that go along with competing for a playoff spot, and may thus be more likely to increase their performance in the later stages of the season. Playoffs is also a dummy variable--it equals one if a team is within six points (plus or minus) of the last playoff spot (i.e., the 8th position in the conference standings) at the time the Olympic break occurred. The idea here is that these teams are "on the bubble" in terms of securing a playoff spot, and as such, may have more at stake in the latter part of the season than teams that are already reasonably assured to be in the playoffs or teams that have already dropped far out of contention. The final control variable, GoalDif, measure the team's total goal differential over the entire season in question. It attempts to capture overall team quality--better teams will have higher values of GoalDif. With overall goal differential as a control variable, the dependent variable then measures how this goal differential is distributed between the pre-Olympic and post-Olympic periods of the season.
Olympians is the final independent variable and is the focus of the analysis. It measures the number of players from an NHL team's roster that participated in the Olympics during the season in question. Its sign is a priori indeterminate since it captures two opposing effects. First, there is the possibility that the Olympics creates greater fatigue in those players that participate, thus lowering their post-Olympic performance. While the eight (for some) games that these players play in the Olympics is about the same number of NHL games they would play over that same 17-day time period in a non-Olympic year, these NHL games are not eliminated and still must be played, and thus get compressed into the rest of the season. Therefore, while these players play 82 (regular season) games during non-Olympic years, some will play as many as 90 games during Olympic years. Working in parallel with this effect is that NHL teams with more Olympians have, by necessity, fewer non-Olympians, and hence fewer players that get the valuable rest that the Olympic break provides for those not participating. Countering these negative effects on NHL teams of having players participate is the possibility that Olympic participants are inspired and energized by the Olympic experience, thus enhancing their performance upon their return to the NHL.
The model is estimated as a panel data set, with 116 observations, one for each NHL team in each of the four Olympic seasons. (4) The results are reported in Column A of Table 2. The only control variable that is significant is Playoffs. Its positive sign indicates that teams that are close to the playoff cutoff point are more likely to have stronger post-Olympic performances, presumably due to the fact that these teams have more at stake than do teams either well above or well below the cutoff point. With respect to the focus variable, the coefficient on Olympians is negative and significant at the 1% level. Thus, the more players an NHL team supplies to the Olympics, the greater is that team's post-Olympic decline in performance. The results lend support to the "fatigue" argument--the extra mental and physical stress of playing in the Olympics reduces a player's effectiveness upon his return to the NHL, thus decreasing the post-Olympic performance of the team for which he plays.
The magnitude of the coefficient on Olympians indicates that each additional player that an NHL team sends to the Olympics reduces that team's post-Olympic goal differential by 0.088 goals per game. Thus, if we consider two teams at opposite ends of the NHL standings, and the best team sends, say, five more players to the Olympics than the other, this translates into a 0.44 relative reduction (0.088 x 5) in goal differential per game for the former. With teams having, on average, 23 post-Olympic games remaining, the team with the higher number of Olympians will see a reduction in overall goal-differential of about 10.12 (i.e., 0.44 x 23). To put this in context, over the sample period, the top five NHL teams had an average season-long goal differential of +57, while the bottom five had a differential of -65. Thus, the estimated Olympiceffect of 10.12 goals is certainly not negligible.
In essence, then, the Olympics do seem to change competitive outcomes in the NHL. The best-performing NHL teams--because they send the most players to the Olympics--are disadvantaged relative to weaker NHL teams. Thus, the Olympics have somewhat of a compressing effect on NHL standings--in some sense, the best NHL teams become victims of their own success. Whether this compressing effect is ultimately beneficial for the NHL is unclear. To the extent that it makes playoff races potentially more exciting over the post-Olympic portion of the NHL's regular season, it may increase fan interest for some teams. However, to the extent that it reduces the performance of the League's high-profile teams, it may work against League interests.
The issue is explored in further depth by examining whether certain types of players are particularly impacted by the Olympics. One such question is whether a player's nationality impacts his post-Olympic performance. While a priori expectations are not entirely clear here, some informal hypotheses are possible. For example, in Canada, hockey is the national sport, and is deeply engrained within the social fabric of the country--no other sport comes even close to hockey in terms of popularity. The country's hockey team is usually favored to win any premiere international event, and the media attention and fan expectations can be enormous. Compare this with, say, the US, where the sport of hockey has a much lower profile, and where many Americans do not regularly follow the game. A key question, then, is whether such differences ultimately impact players themselves, and whether the presumed increased pressure faced by Canadian Olympians ultimately helps or hurts their post-Olympic performance.
Nationality issues may also have other potential impacts. For example, there may be differences between European and North American players. One could argue that, in Europe, international events, including the Olympics, are a stronger and much more significant part of the popular sport culture than in North America; as discussed earlier, North American sport has traditionally been focused more internally on domestic leagues, with international events tending to be of secondary importance. To the extent this may be true, European players, particularly those from countries like the Czech Republic, where hockey is the most popular sport, may experience increased pressure resulting from the high scrutiny that their Olympic performances receive back in the homeland.
To test for this possible nationality effect, seven additional dummy variables were added to the base model of Equation 1. The variables--Canada, US, Russia, Czech, Sweden, Finland, and Slovakia--represent the proportion of an NHL team's Olympians that come from each of the seven leading hockey-producing nations. The results are reported in Column B of Table 2. Both Playoffs and Olympians remain significant as in the base model of Column A, but only one of the nationality variables--Russia--is significant. Thus, other than for Russians, nationality had no independent effect on the post-Olympic performance of NHL players--thus, while participating in the Olympics reduces player performance after their return to the NHL, players from one country are generally no more affected than players from any other country. With Russia, the positive coefficient indicates that NHL teams with more Russian Olympians actually improve their team performance following the Olympics, all else equal. Perhaps the long and distinguished history of that country's hockey performance at the Olympics--the Soviet Union won seven gold medals in hockey over the nine Olympics between 1956 and 1988 (and winning the silver and bronze in the other two Olympics)--is so inspiring to players that their selection to the Russian team motivates them to perform at sustained high levels even after their return to the NHL. (5)
A second extension to the base model is to investigate whether the success of a player's Olympic team has any impact on the post-Olympic performance of that player's NHL team. For example, if a player is a member of the team that wins the gold medal, the effects on that player might be quite different than if his Olympic team was eliminated in, say, the quarterfinals. There may be several factors involved. First, players on teams that advance to the gold medal game will play more games, and stay at the Olympic site longer, than players on teams that are eliminated earlier in the competition, further increasing possible fatigue effects. Second, these additional games are, by their very nature, particularly high-pressure situations, and are undoubtedly psychologically stressful. In addition, players advancing to the gold medal game will experience an extreme emotion, either the exhilaration from winning or the disappointment from defeat. Players on the winning side may be so inspired by their victory that they go back to their NHL teams and perform at an even higher level than before the Olympics; alternatively, the physical and psychological stresses of attaining victory may precipitate a "letdown" following the Olympics, and may thus reduce a player's effectiveness upon his return to the NHL. In essence, there are many possible impacts, none of which are a priori predictable from any type of economic theory, and thus empirical testing is necessary if one is to gain any type of clarity on the issue.
In this regard, the base model regression in Equation 1 was re-run with three additional variables added--Gold, Silver and Bronze--indicating the proportion of an NHL team's Olympians that were on one of these medal-winning squads. Also added to the model was a fourth additional variable, one that captured what could be termed the "host country" effect. Players on those Olympic teams that represent the host country--i.e., American players in 2002 and Canadian players in 2010 (6) --face additional pressure of being under intense media scrutiny, and of having the burden of high fan expectations.
Column C of Table 2 shows the results. Again, both Playoffs and Olympians remain significant, as in the base model. However, none of the three medal-winning variables were significant--the performance of a player's team at the Olympics had no apparent effect on the player's NHL team. (7) Host, on the other hand, does seem to matter. Its negative and significant coefficient suggests that NHL teams that supply more players to the host-country roster experience a greater post-Olympic decline in team performance. This result is consistent with the notion that players from host countries face particularly high scrutiny and expectations, and the resulting pressures and stresses ultimately cause reductions in their post-Olympic performance.
A Robustness Check
One possibility with all of the specifications in Table 2 is that the negative coefficient on Olympians is simply reflecting some type of "regression to the mean" over the course of a season, and is not actually measuring the impacts of the Olympics. For example, some teams that performed well during the pre-Olympic period (and are correspondingly also likely to have more Olympians) may have played above their true talent level, and will inevitably "cool off" at some point, and vice-versa.
To test for this possibility, we examined the four seasons immediately subsequent (8) to the Olympics--i.e., for 1998-99, 2002-03, 2006-07, and 2010-11--and ran the same regression as given in Column A of Table 2. For each of these seasons, we broke the season into two halves at the point where the Olympics would have started (which was with 23 games remaining in the NHL season). We measured the variables in the same manner as described above, and adjusted the Olympians variable for each NHL team to reflect how many players on that team were on an Olympic roster in the previous season.
The results are reported in Table 3, and show that the Olympians variable in these non-Olympic years is no longer significant. This indicates that NHL teams that have a greater number of Olympic-caliber players do not seem to experience a late-season decline in performance in non-Olympic years. This result provides increased assurance that the significance of Olympians in Table 2 is, in fact, reflecting the impact of the Olympics, and is not simply an artifact of some type of regression to the mean within a season. (9)
It has become commonplace for elite professional athletes in both Europe and North America to represent their respective countries in major international competitions. However, what is unique about hockey is that the premier competition, the Winter Olympics, occurs during the NHL season itself, thus causing the NHL to suspend League play for about two-and-a-half weeks at a point in the season that is approximately two-thirds of the way through the schedule.
This disruption has always caused some internal debate within the NHL as to whether the League's participation in the Olympics was worth the potential costs--costs both in terms of player fatigue and injury risk, and in terms of loss of continuity in the NHL season. This debate takes on a more urgent nature as the 2014 Sochi Games approach. The NHL has not yet committed to participating in these Games, and the long travel distances from North America to Russia appear to be serious cause for concern.
It is within this context that the paper's empirical analysis is conducted. The results indicate that NHL teams are differentially impacted by the Olympics. In general, teams that send more players to Olympic rosters are disadvantaged once the NHL season resumes after the Olympics. This reduced team performance is presumably due to a fatigue factor, where the travel and additional games that the Olympics entail simply cause players to wear out. These differential impacts across teams have potential competitive balance implications (albeit unintended), since the best-performing NHL teams tend to supply the most players to Olympic rosters. Thus, Olympic participation may tend to compress NHL standings, with the best teams having their post-Olympic performance hampered relative to those teams that supply fewer players to the Olympics. Thus, one might expect, all else equal, that the standard deviation of winpercent would be lower in seasons in which there is an Olympics.
Whether this outcome is beneficial or harmful to the NHL is unclear and is beyond the scope of the paper, but is a direction for future research. To the extent that it makes playoff races potentially more exciting over the post-Olympic portion of the NHL's regular season, it may increase fan interest for some teams, and hence have positive impacts on revenues. However, to the extent that it reduces the performance of the League's high-profile teams, it may work against League interests in that it lowers the chances of these teams making the playoffs. As well, to the extent that the Olympics do increase fatigue for those that participated, these players may be more prone to injury upon their return to the NHL--if a "star" player is injured, this could have significant negative revenue consequences on the NHL for the remainder of the season. Of course, balanced against these more short-term revenue considerations is the broader long-term benefit that the NHL gains from exposing its game to a larger world audience.
Berri D., Brook, S., Frick, B., Fenn, A., & Vicente-Mayoral, R. (2005). The short supply of tall people: Competitive imbalance and the National Basketball Association. Journal of Economic Issues, 39(4), 1029-1041.
Fort, R. (2006). Competitive balance in North American professional sports leagues. In J. Fizel (Ed.), Handbook of sports economics research (pp. 190-208. Armonk, NY: M.E. Sharp.
Humphreys, B. (2002). Alternative measures of competitive balance in sports leagues. Journal of Sports Economics, 3(2), 133-148.
Humphreys, B., & Watanabe , N. (forthcoming). Competitive balance. In L. Kahane & S. Shmanske (Eds.), Oxford handbook on sports economics. Oxford, UK: Oxford University Press.
Kahane, L, Longley, N., & Simmons, R. (forthcoming). The effects of coworker heterogeneity on firm level output: The case of Europeans in the National Hockey League. Review of Economics and Statistics.
Krautmann, A. C., Lee, Y. H., & Quinn, K. (2011). Playoff uncertainty and pennant races. Journal of Sports Economics, 12(5), 495-514.
Quirk, J., & Fort, R. (1997). Pay dirt: The business of professional team sports. Princeton, NJ: Princeton University Press.
Sanderson, A., & Siegfried, J. (2003). Thinking about competitive balance. Journal of Sports Economics, 4(4), 255-279.
Schmidt M. B., & Berri, D. J. (2001). Competition and attendance: The case of Major League Baseball. Journal of Sports Economics, 2(2), 147-167.
(1) See "Can NHL Afford to Skip Sochi Games?" Fox Sports, February 20, 2010. Retrieved from http://msn.foxsports.com/nhl/story/can-nhl-af^ord-to-skip-sochi-winter-olympic-games-022210
(2) Since this paper examines CB issues, its focus is at the team level. However, as an alternate line of research, one could take the analysis back a step and instead examine the impact of the Olympics on the performance of individual players. However, in practice, such an analysis is difficult to conduct because there is no single, comprehensive, measure of player performance, particularly because hockey players simultaneously play both offense and defense. While offensive performance is relatively easy to objectively measure, defensive performance is not, and the latter may impacted just as much as the former by the Olympics. Conducting team-level analysis largely avoids these measurement issues.
(3) Teams normally carry about 23 players on their NHL roster. However, they can only dress 20 players for any given game.
(4) Team-level fixed effects are not employed in the model. The primary reason for their exclusion is that each observation on a team is made only once every four years, and over that time the great majority of players on that team change, as do the team's coaches and general manager. Thus, there is little reason to believe that anything is actually "fixed" at the team level. However, as a robustness check, we did run a regression with fixed effects, and found that our key results remained.
(5) The importance of the Olympics to the Russians is illustrated by the views of Alex Ovechkin, star forward of the Washington Capitals. Ovechkin has been a strong proponent of the NHL participating in the Sochi Games, and has even gone as far as stating that he would (temporarily) leave the Capitals to play for Russia in 2014. See "Can NHL Afford to Skip Sochi Games?" Fox Sports, February 20, 2010. Retrieved from http://msn.foxsports.com/nhl/story/can-nhlafford-to-skip-sochi-winter-olympic-games-022210. Added to this is that Russia reportedly pays its gold medalists a bonus of approximately 100,000 Euros, much more than any other country, further showing the importance to Russia and Russians of Olympic success.
(6) The host countries in 1998 and 2006, Japan and Italy, respectively, had no NHL players on their Olympic rosters.
(7) As a robustness check, alternate variable definitions were also tried. For example, rather than Gold, Silver, and Bronze being separate dummies, a single dummy was created that included all medal winners. Other variations included a single dummy variable that captured just the two finalists, and one that captured the top four teams (the three medal winners, as well as the fourth-place finisher, i.e., the loser of the bronze medal game). As when Gold, Silver, and Bronze were included separately, none of these alternate dummies were significant.
(9) We chose the season subsequent, rather than prior, to the Olympics because there was no season prior to the 2006 Olympics due to the season-long work stoppage in the NHL.
(9) Relatedly, another possibility is that the top clubs prior to the Olympics will obviously be closer to clinching a playoff spot, and will thus be more likely to give some rest to their better players late in the season. While such a scenario is certainly possible, there are several factors that assure us that this is not materially affecting our results. First, while imperfect, we do attempt to control for this with the variable Playoffs (and which is significant in all of our Table 2 regressions). Second, the results from Table 3 suggest that in non-Olympic years there is no late-season drop-off in performance for those teams with high numbers of Olympic-caliber players. Lastly, while teams may no doubt provide some rest to stars after clinching a playoff spot, coaches are still concerned with winning, both to maintain momentum and to secure the best possible seeding (and the commensurate home-ice advantages that go with it).
The author would like to thank both an anonymous referee and an associate editor for helpful comments on an earlier draft of the paper. Any errors are mine.
University of Massachusetts, Amherst
Neil Longley, PhD, is a professor in the Isenberg School of Management. His research interests include labor economics and public choice economics.
Table 1: Olympic Participation Across NHL Teams, 1998-2010 2010 2006 2002 1998 Average All NHL Teams: Average 4.63 4.87 4.07 4.31 4.47 Number of Olympians Top-5 NHL Teams: Average 5.80 7.40 6.40 6.00 6.40 Number of Olympians Bottom-5 NHL Teams: 3.20 2.60 1.80 4.40 2.90 Average Number of Olympians Table 2: Olympic Participation and NHL Team Performance Column A Column B Column C Constant .513 (.640) .274 (.341) 1.032 (1.250) Games -.005 (-.218) -.011 (-.436) -.027 (-1.016) Home -.281 (-.299) -.256 (-.270) -.129 (-.139) Conf .006 (.014) .228 (.485) -.040 (-.095) PrevYr .057 (.464) .092 (.746) .097 (.788) Playoffs .251 (2.364) .264 (2.512) .233 (2.209) GoalDif .000 (-.081) .000 (-.222) -4.788E-5 (-.033) Olympians -.088 (-3.084) -.078 (-2.659) -.087 (-3.027) Canada .225 (.676) US -.274 (-.737) Russia .816 (2.340) Czech -.061 (-.170) Sweden -.140 (-.377) Finland .019 (.056) Slovakia .189 (.356) Gold -.158 (-.514) Silver .360 (1.095) Bronze -.275 (-.991) Host -.882 (-2.104) [R.sup.2] = .15 [R.sup.2] = .24 [R.sup.2] = .20 N= 116 N=116 N=116 t-stats in parentheses Table 3: Seasons Immediately Subsequent to Olympics Constant 1.486 (.915) Games -.057 (-1.088) Home -.822 (-.677) Conf .279 (.602) PrevYr -.198 (-1.323) Playoffs .001 (.078) GoalDif .004 (2.234) Olympians .025 (.839) [R.sup.2] = .07 N= 116