Selling the Game

8/2/2019 Selling the Game

1/18

Southern Economic Journat 2008, 74(3), 794-810

Selling the Game: Estimating the EconomicImpact of Professional Sports throughTaxable SalesR o b e r t A . B a a d e ,* R o b e r t B a u ma n n . f a n d V ic to r A . Ma th e so n J

Sports leagues, franchises, and civic boosters to ut the economic benefits of professional sportsas an incentive for host cities to construct new stadiums or arenas at considerable publicexpense. Past league-sponsored studies have estimated that new stadiums, franchises, andmega-events such as the Super Bowl increase economic activity by potentially hundreds ofmillions of dollars in host cities. A detailed regression analysis of taxable sales in Florida overthe period extending from 1980 to 2005 fails to support these claims. New stadium s, arenas, andfranchises, as well as mega-events, appear to be as likely to reduce taxable sales as increasethem. Similarly, strikes and lockouts in professional sports have not systematically lead toreductions in local taxable sales.JEL Classirication: L83

1. IntroductionSports boosters often claim that sports teams, facilities, and events inject large sums of

money into the cities lucky enough to host them. Promoters envision hoards of wealthy sportsfans descending on a city's hotels, restaurants, and businesses and showering them with fistfulsof dollars. For example, the National Football League (NFL) typically claims an economicimpact from the Super Bowl of around $400 million (National Football League 1999), MajorLeague Baseball (MLB) attaches a $75 million benefit to the All-Star Game (Selig, Harrington,and H ealey 1999) and up to $250 million for the W orld Series (Ackman 2000), and theestimated effect of the National Collegiate Athletic Association (NCAA) Men's BasketballFinal Four ranges from $30 million to $110 million (Mensheha 1998; Anderson 2001). Multi-day events such as the Olympics or soccer's World Cup produce even larger figures. The pre-Olympics estimates for the 1996 Games in Atlanta indicated that the event would generate$5.1 billion in direct and indirect economic activity as well as 77,000 new jobs in Georgia(Hum phreys and Plumm er 1995). A study of soccer's 2002 World C up (by the Dentsu Institutefor Human Studies) estimated a $24.8 billion impact for Japan and an $8.9 billion impact forSouth Ko rea. A s a percentage of national income, these figures represent 0.6% and 2.2% of the

* Depa rtment of Economics and Business, Lake Forest College, Lake Fo rest, IL 60045, USA; [email protected] Department of Economics, Box 192A, College of the Holy Cross, Worcester, MA 01610-2395, USA; E-mail


2/18

Economic Impact of Professional Sports 795total Japanese and South Korea n economies, respectively (Finer 2002). Initial economic im pactstudies of the 2010 Winter Olympics in VancouverAVhistler predict a gain to the local economyof up to $10 C billion.

Even regular season games prompt claims of huge benefits. For example, the OregonBaseball Campaign, a group dedicated to bringing MLB to Portland, reported that "a MLBteam and ballpark would g enerate between $170 and $300 million a nnually in grossexpenditures to the state of Oregon" (Oregon Baseball Campaign 2002), while a similaranalysis completed for the Virginia Baseball Authority stated that a "a major league baseballfranchise and stadium in nor thern V irginia would pu m p more th an $8.6 billion into theeconomy over 30 ye ars," or $287 million annually. The St. Louis (Missouri) Regional Cham berand Growth Association estimated that the Cardinals brought $301 million in annual economicbenefits to the region, with another potential $40 to $48 million in benefits from a post-seasonappearanc e (Saint Louis Regional Chambe r and Gro wth A ssociation 2000). Of course, baseballis not the only sport to provide rosy economic impact numbers. A study of the NFL's NewOrleans Saints estimated the impact of the team on the state at $402 million in 2002 (Ryan2003), and the Seattle Supersonics of the NBA claimed that they pump $234 million into thearea's economy annually (Feit 2006). Boosters are often vague about exactly what is beingmeasured in these claims of hundreds of millions of dollars of benefits, making directcomparisons difficult, but the overall claims are clear: Professional sports provide hugeeconomic windfalls for host cities.

Of course, leagues, team owners, and event organizers have a strong incentive to provideeconomic impact numbers that are as large as possible in order to justify heavy public subsidies.When leagues consider expansion or franchise relocations, they frequently highlight thepotential economic benefits of a new franchise in order to minimize the team's or league'srequired contribution to the funding of the stadium or arena in which the team will play.Similarly, the NFL and MLB use the Super Bowl and baseball's All-Star Game as carrots toprompt otherwise-reluctant city officials and taxpayers to provide lavish funding for newstadiums to the great financial benefit of the existing owners. For example, in baseball, of the 15new major league stadiums built between 1970 and 1997, 13 were selected by the MLB to hostan All-Star G ame within five years of their construction (Baade and M atheson 2001). Similarly,during a visit to the D alla s-F ort W orth, T exas, area just before a crucial vote on public fundingfor a new stadium, N F L Comm issioner Paul Tagliabue suggested tha t the construction of a newstadium would lead to the opportunity for the metro area to host the Super Bowl in the nextdecade. Since the NFL touts economic benefits from hosting the Super Bowl of $350 to$400 million, an amoun t th at exceeded the pro posed $325 million p ublic subsidy for thestadium, in effect. Commissioner Tagliabue was saying that combined with a Super Bowl,Arlington, Texas, would be getting a new stadium for free.

With an event like the Olympics, the huge costs of hosting the event to the standards nowrequired by the International Olympic Committee, as well as those associated with providingadequate security, almost necessitate an infusion of taxpayer money. For example, while onpaper the 2002 Winter Olympics in Salt Lake City, Utah, made a profit, the cost figures did notinclude millions of dollars of additional security provided by the U.S. Department of Defenseat no cost to the local organizing committee. For the 2004 Summer Games, the government in


3/18

796 Baade, Baumann, and M athesonstand to benefit directly from the public subsidies such reports are designed to elicit, one mustquestion whether such studies can be believed.

2 . Ex Ante versus Ex Post Studies

A typical ex ante economic impact study used by league and event promo ters estimates thenumber of visitors an event or team is expected to draw, the number of days each spectator isexpected to stay in the city, and the amount each visitor will spend each day. Combining thesefigures, an estimate of the "direct economic impact" is obtained. This direct impact is thensubjected to a multiplier, usually around two, to account for the initial round of spendingrecirculating through the economy. This additional spending is known as "indirect economicimpact." Thus, the total economic impact is roughly double the size of the initial spending.While such an estimation method is relatively straightforward, academic economists have beenquick to point out the failings of such ex ante studies, as they often rely on poor methodologyand also suffer from several theoretical problems.

First, many booster estimates are wildly optimistic about the number of potential guestsand their spending habits. In March 2005, Denver, Colorado, tourism officials predicted100,000 visitors for the NBA All-Star Game. Considering that the Pepsi Center, the game'svenue, only holds 20,000 fans and taking into account that Denver has only about 6000 hotelrooms, it is not clear exactly how such an influx of basketball fans would be possible.

In many cases, the variation in estimated benefits alone is enough to question the validityof the studies. A series of studies of the NBA All-Star Game produced numbers ranging froma $3 million windfall for the 1992 game in Orlando, Florida, to a $35 miUion bonanza for thegame three years earlier in H ouston , Texas (Houck 2000). Similarly, the 1997 NC AA W om en'sBasketball Final Four was estimated to have an economic impact of $7 million on the localeconom y of Cincin nati, Oh io, but the same event was predicted to produ ce a $32 millionimpact on the San Jose, California, economy just two years later (Knight R idder New s Service1999). The 10-fold disparity in the estimated impact for the same annual event illustrates the adho c nature of these studies. In some cases, economic impact figures appear to be completelyfabricated. While city or league officials may suggest a certain monetary figure for a particularevent, when pressed on the details, the "missing study" syndrome arises (Anderson 2004).

Even when ex ante studies are don e in a carefully considered m ann er, they suffer fromthree primary theoretical deficiencies: the substitution effect, crowding out, and leakages. Thesubstitution effect occurs when consumers spend money at a sporting event rather than onother goods and services in the local economy. A local resident who goes to a baseball game isspending money at the game tha t likely would have been spent at local restauran ts, theaters, orretail establishments in the absence of the game. Therefore, the local consumer's spending ona sporting event is not new economic activity; rather, it represents a reshuffiing of localspending. For this reason, most economists advocate that spending by local residents beexcluded from any economic impact estimates.


4/18

Economic Impact of Professional Sports 797a ticket to a local game, and therefore the ticket would be counted as a direct economic impactof the sports contest. The professor, however, would have come to the city and spent m oney onhotels and restaurants in the absence of the sporting match, and again, the money spent at thegame substitutes for money that would have been spent elsewhere in the local economy.

Similarly, ex ante estimates may be biased upwards if event guests engage in "time-switching," which occurs when a traveler rearranges a planned visit to a city to coincide witha mega-event. One example of time-switching is someone who has always wanted to visitHawaii who plans a trip during the NFL's Pro-Bowl. While the Pro-Bowl did influence thetourist's decision abo ut when to com e, it did no t affect the decision whether to come. Therefore,total tourism spending in Hawaii is unchanged; the Pro-Bowl simply affects the timing of suchspending.

Accounting for the su bstitution effect is likely to result in large reduction s in the estim atedeconomic impact of regular season games. In the case of mega-events, however, the sub stitutioneffect may be much smaller. Since these premier events are thought to attract large audiencesfrom outside the local economy, many of whom come specifically for the event, the amount ofspending that is new to the economy is thought to be quite a large proportion of the totalamount of spending, whereas 5-20% of fans at a typical MLB game are visitors from outsidethe local metropolitan area, the percentage of visitors at an event like an All-Star G am e or theSuper Bowl is thought to be much higher (Siegfried and Zimbalist 2000).

A second source of bias is "crowding out," which results from the congestion caused bya game that dissuades local citizens from venturing near the playing venue during the game andthereby reduces economic activity. Attractions such as Chicago's Field M useum or Cleveland'sRock and Roll Hall of Fame are located next door to NFL stadiums, and their attendancesuffers during the Chicago Bears or Cleveland Browns home games. Similarly, mega-eventsmay dissuade regular recreational and business visitors from coming to a city during that time.While a city's hotels may be full of sports fans during the Super Bowl, if the city's hotels aregenerally full of vacationers or conventioneers anyway, the Super Bowl simply displaces othereconomic activity that would have occurred. High prices charged by hotels and other businessesin the hospitality industry also tend to dissuade casual visitors during mega-events. In otherwords, the economic impact of a mega-event may be large in a gross sense, but the net impactmay be small. Scores of examples of this phenomenon exist. As a case in point, during the 2002World Cup in South Korea, the number of European visitors to the country was higher thannormal, but this increase was offset by a similar-sized decrease in the number of regular touristsand business travelers from Japan who avoided South Korea as a result of World Cup hassles.The total number of foreign visitors to South Korea during the World Cup in 2002 wasestimated at 460,000, a figure identical to the number of foreign visitors during the same periodin the previous year (Golovnina 2002).

A third source of bias comes from leakages. While money may be spent in local economiesduring sporting events, this spending may not wind up in the pockets of local residents. Thetaxes used to subsidize these events, however, are paid for by local taxpayers. The incomemultiplier for sporting events is likely to be much lower than for general expenditures as a resultof the specialized nature of the service provided. In the NBA, for example, only 29% of playerslive in the metropolitan area in which their team plays (Siegfried and Zimbalist 2002).


5/18

798 Baade, Baumann, and Mathesonpatterns. During mega-events, however, the economy within a region may be anything butnormal, and, therefore, these same inter-industry relationships may not hold. Since there is noreason to believe the usual economic multipliers apply during mega-events, any economicanalyses based on these multipliers may, therefore, be highly inaccurate.

In fact, there is substantial reason to believe that during mega-events, these multipliers arehighly overstated, which overestimates the true impact of these events on the local economy.Hotels, for example, routinely raise their prices during mega-events to three or four times theirnormal rates. The wages paid to a hotel's workers, however, remain unchanged, and, indeed,workers may be simply expected to work harder during times of high demand without anyadditional monetary compensation. As a hotel's revenue increases without a correspondingincrease in costs, the return to capital (as a percentage of revenues) rises, while the return tolabor falls. Capital income is far less likely than labor income to stay within the area in which itis earned, and, therefore, one might expect a fall in the multiplier effect during mega-events asa result of these increased leakages (Matheson 2004).While ex ante estimates often do a credible job of determining the economic activity thatoccurs as a result of a sports team or mega-event, and although they m ay also address the issueof the substitution effect by excluding spending by local residents, they generally do a poor jobof accounting for crowding-out, and they almost never acknowledge the problems associatedwith the application of incorrect multipliers. For these reasons, numerous studies have lookedback at the actual performance of economies that have had professional franchises, built newplaying facilities, and hosted mega-events and have compared the observed economicperformance of host cities to that predicted in ex ante studies. These ex post analyses ofstadiums and franchises, including those of Rosentraub (1994), Baade (1996), Coates andHumphreys (1999, 2003), and Siegfried and Zimbalist (2000), to name just a few, generally fmdlittle or no economic benefits from professional sports teams or new playing facilities.

In the area of mega-events, Baade and Matheson (2001) examine the MLB's All-StarGame and fmd that employment growth in host cities between 1973 and 1997 was 0.38% lowerthan expected, compared to other cities. A similar examination of the 1996 Summer Olympicsin Atlanta, Georgia, found employment growth of between 3500 and 42,000 job s, a fraction ofthe 77,000 new jobs claimed in ex ante studies (Baade and Matheson 2002). An examination ofmetropolitan area-wide personal income during 30 NCAA Men's Final Four basketballtournaments found that, on average, personal incomes were lower in host cities duringtournament years (Baade and Matheson 2004a). A similar study of the 1994 World Cup in theUnited States found that personal income in host cities was $4 billion lower than predicted,a direct contradiction to ex ante estimates of a $4 billion windfall (Baade a nd M atheson2004b). Coates and Humphreys (2002) examined the effect of post-season play in all four majorU.S. sports on per capita personal incomes and found in all cases that hosting playoff gameshad a statistically insignificant impact on per capita incomes.

The remainder of this paper adds to the already-substantial body of work regarding expost analyses of franchises, stadiums, and sporting events by using taxable sales data toestimate the effect of professional sports on local economies. In addition, this paper examineslabor disputes, which serve as natural experiments for determining the economic impact of


6/18

Economic Impact of Professional Sports 799of the 1994 MLB strike on 17 metropolitan statistical areas (MSAs), and he later extends hiswork to cover the effect of this strike on spring training venues in Florida (Zipp 1997). Baadeand Matheson (2005) examine the 1981 and 1994/95 MLB baseball strikes, using personalincome data, to arrive at an average net annual economic impact of a MLB team on a host cityof between $16.2 milhon and $132.3 million, or between 5% and 50% of the figure generallysuggested by baseball's boosters. Coates and Humphreys (2001) present the mostcomprehensive analysis of the economic consequences of sports strikes and lockouts. Theiranalysis of real per capita personal income finds no statistically significant effects from thestrikes in MLB in 1972, 1981, and 1994/95 and strikes in the NFL in 1982 and 1987.

The existing studies of sports labor interruptions have two major weaknesses. First, allfour previous studies examined the 1994/95 baseball strike, but only Coates and Humphreys(2001) examine a sport other than baseball. To the best of our knowledge, no other study hasexamined the effect of labor stoppages in the NBA and the National Hockey League (NHL).

Second, a major difficulty of measuring the economic impact of sports teams, events, andstrikes is that even the imp act of large businesses may be ha rd to isolate within the large, diverseme tropolitan economies in which they reside. For exam ple, even if a M LB franchise or a SuperBowl does result in a $300 million boost to the host city, this is less than 0.1% of the annualpersonal income of a large metropolitan area like Los Angeles. Any income gains as a result ofa franchise or the "big game" would likely be obscured by normal fiuctuations in the region'seconomy. This problem is further com pounded if the labor interruption or the mega-event lastsfor only a few months or even just a few days. Even if the effects of a labor dispute or mega-event are large in the time period immediately surrounding the event, this impact is likely to beobscured in annual data. The studies of Coates and Humphreys (2001) and Baade andMatheson (2005), which examine the ex post economic impact of labor interruptions in sports,both suffer from this limitation. For example, Coates and Humphreys (2001) faced a dauntingtask when they examined the impac t of the 1972 MLB strike, which lasted 13 day s, usingannual data. Similarly, Coates and Humphreys' (2001) and Baade and Matheson's (2000, 2001,2002, 2004a, b) studies of mega-events all examine events lasting at most one month, often aslittle as a single weekend, using annual data.

3. Use of Taxable Sales

Taxable sales are ideally suited to measuring the economic impact of stadiums, teams, andlarge sporting events for several reasons. First, there is a direct connection between sales taxcollections and sporting events or facilities. Boosters often include large sums for visitorspending in their ex ante estimates of the economic impact of an event. In one of the fewexamples of a league-sponsored ex post study, the NFL reported that Super Bowl XXXIII in1999 was responsible for a $670 million increase in taxable sales in South Florida, compared tothe equivalent January-February period in 1998 (National Football League 1999). Numerouspublicly funded sports facilities have also been financed specifically from sales tax collections orthrough specific increases in the sales tax rate, making an examination of taxable sales


7/18

800 Baade, Baumann, and Mathesonon specific goods and services, such as rental cars or hotel rooms) (Baade and Matheson 2006).In addition, consumer spending, much of which is captured by taxable sales, is the single largestcomponent of gross domestic product and, therefore, is a good proxy for economic activity.

A second major reason that taxable sales are a useful tool in measuring the economicimpact of professional sports is that, as noted previously, even significant economic events maybe hard to isolate within the large, diverse metropolitan economies in which they take place.Any income gains as a result of the game or team would likely be obscured by normalfluctuations in the region's economy. If the event or franchise can be isolated within space andtime, however, any potential impact is more likely to be identified. For example, while thepresence of a World Series might have a large effect on neighborhood businesses, the overalleffect on a state or country's economy will be minuscule and hard to identify. Furthermore,these same economic effects may be large for the time period immediately surrounding theevent, but over the course of an entire year, the impact during a week-long period is not likelyto show up as an important change.Most previous studies of professional sports have used personal income (Baade andMatheson 2004a, b), per capita income (Coates and Humphreys 1999, 2002, 2003), oremployment data (Baade 1996; Baade and Matheson 2000, 2001, 2002; Coates and Humphreys2003) to estimate the ex post economic impact of sports. Generally, these data are availableonly annually and at the county or metropolitan area level, and, therefore, these studies sufferfrom the limitations mentioned previously. Taxable sales data, on the other hand, are oftenpublished either monthly or quarterly and can cover areas down to the city level or smaller.Therefore, these data can be analyzed to identify activities that are much smaller in scale andduration.Several previous attempts to measure the effect of teams and mega-events through taxablesales data have been made. Baade and Matheson (2000) challenge the NFL's claim ofa $670 million boost in South Florida's taxable sales from the Super Bowl and arrive at a figureof a mere $37 million increase. Th eir ana lysis is quite simplistic, however; their estimatesaccount for only GDP growth, infiation, and population growth. Baade and Matheson (2001)examined taxable sales in California to determine the effect of MLB's All-Star Game on localeconomies. They found that the three California cities that hosted All-Star Games between1985 and 1997 suffered an averag e dro p in taxable sales of roughly $30 million in the qu arter inwhich the game took place. Their study, however, is limited only to baseball's All-Star Game.Porter (1999) provides detailed analysis of taxable sales with respect to mega-events, usingregression analysis to determine that the economic impact of the Super Bowl was statisticallyinsignificant, that is, not measurably different from zero. After reviewing short-term data onsales receipts for several Super Bowls, Porter concluded:

Investigator bias, data measurement error, changing production relationships, diminishingreturns to both scale and variable inputs, and capacity constraints anywhere along the chain ofsales relations lead to lower multipliers. Crowding out and price increases by input suppliers inresponse to higher levels of demand and the tendency of suppliers to lower prices to stimulatesales when demand is weak lead to overestimates of net new sales due to the event. Thesecharacteristics alone would suggest that the estimated impact of the mega-sporting event will belower than the impact analysis predicts, (p. 65)


8/18

Economic Impact of Professional Sports 801event having a beneficial impact on any type of sales, total, retail, or services, or of the parts ofthose sales actually subject to the sales tax" (p. 240). Coates and Depken (2006) extend thisstudy by exam ining 126 jurisdic tions in Texas from 1990 thro ugh early 2006. They find tha tregular season games in the NBA, NFL, NHL, and MLB have divergent effects, with NHL andMLB games increasing tax revenues, while NBA and NFL regular season games decreaserevenue. Collegiate regular season football games are revenue generators for small cities andtowns home to Division I and Division IAA football. The Super Bowl had the largest effect ofany event in the sam ple, with H ouston garnering an additional $2 million in tax revenuesduring the big game.

This paper expands the scope of previous work in the area of taxable sales by includinga much broader array of stadiums, franchises, and mega-events; a larger number of host cities;a longer time series; and a more detailed regression analysis in its examination than wereincluded in previous studies.

4. The DataWe use taxable sales data, which are available monthly, to estimate the economic impact

of professional sports on local economies. These data include just over 25 years' w orth ofmon thly sales tax data (Ja nuary 1980 through June 2005) for every county in Florida. Florida isan ideal candidate for this analysis since the state is home to at least two teams from each of the"Big Four" American professional sportsfootball, baseball, basketball, and hockey. Thestate has welcomed multiple new sports franchises to the area over the past 25 years, includingthe Orlando Magic and Miami Heat in the NBA, the Florida Marlins and Tampa Bay DevilRays in the MLB, the Tampa Bay Lightning and the Florida Panthers in the NHL, and theJacksonville Jaguars in the NFL. In addition, at least nine major new stadiums or arenas havebeen constructed in the state since 1980. Furthermore, each of the major labor interruptionssince 1982 has impacted at least one franchise in Florida: the 1982 and 1987 NFL strikes, theNHL's 1994/95 and 2004/05 lockouts, the 1998/99 NBA lockout, and the 1994/95 MLB strike.Only two other labor interruptions in the Big Four have resulted in the loss of games, the 1972and 1981 MLB strikes, which we do not analyze, since no city in Florida hosted a MLBfranchise until the arrival of the Florida Marlins in 1993. Finally, Florida cities have hostedchampionship events for each of the "Big Four" American professional sports as well assoccer's World Cup and the NCAA Men's Final Four basketball tournament. In addition,Florida cities have also hosted all-star games in professional basketball, hockey, and soccer.

In order to maximize the chance that the economic effects of the events can be isolated(i.e., to minimize statistical "noise"), it is crucial to find data as specific as possible to the areain which the franchise is located or in which the mega-event occurred, and with the highestfrequency possible. Florida provides monthly data on taxable sales for individual counties, andthese data meet our criteria. In the analysis, taxable sales from several counties are addedtogether, corresponding to the four specific Florida MSAs that will be examined: Miami-Fort


9/18

802 Baade, Baum ann, and MathesonMSA taxable sales are used in lieu of county taxable sales for two reasons. First, in several

cases, stadiums are very near to county lines. For example. Dolphins Stadium, home of theMiam i Dolphins and the Florida M arlins, resides in M iami-Dad e C ounty b ut is less than 1milefrom the Broward Cou nty border. In a case like this it seems unreasonable to exclude Browa rdCounty from the economic analysis because of the stadium's close proximity. Furthermore, asnoted previously, it is im por tant to differentiate between the gross and net impact of a franchise.If a lockout simply causes residents to spend money elsewhere in the local economy rather thanat the ballpark, analysis of only a single county may not capture this substitution effect.Exam ining an en tire MSA will acco unt for the substitution effect if money is redirected b etweencounties within a single MSA as a result of a strike, expansion, or a new stadium.

Since the current annual GDPs of large MSAs in Florida, such as Miami or Tampa,exceed $50 billion in nominal terms, even the effects ofa potential major economic event, suchas a new franchise or stadium, a prolonged strike, or a mega-event, can be obscured by thenormal economic fluctuations of these economies. Many factors, including the local, regional,and national business cycle; state and federal government policies; monetary policy andinfiation; international factors; consumer and business confidence; wealth effects; and a host ofother ingredients, tend to infiuence taxable sales. In order to prevent these other factors fromclouding the true effects of the labor interruption, it is essential to find a method to account forthem.

5. The ModelIn order to examine the impact of the individual sporting events on taxable sales in the

relevant MSAs of Florida, we estimate the following reduced-form model for each of the fourMSAs in our analysis:

p=\ m=\

where y* is the natural log of taxable sales; x, is a vector of macroeconom ic con trol variables; z,is a vector of variables that capture the impact of strikes, lockouts, expansions, stadiumconstruction, and mega-events; and S^ is a vector of monthly dummies that account forseasonal variation in taxable sales. We include lagged values of the dependent variable tocorrect for autocorrelation in taxable sales and to purge out carryover effects of taxable salesfrom one mon th to the next. We use the Breusch-Godfrey Lagrange m ultiplier test to determinethe appropriate number of lagged dependent variables for each MSA model. As Coates andHumphreys (2001) point out, lagged dependent variables bias their own coefficients, but not thecoefficients of other variables in a linear model. Thus, we attach no interpretation of thesecoefficients.

The macroeconomic control variables in x, are the state population in Florida and the


10/18

Economic Impact of Professional Sports 803monthly data. We employed a trial-and-error approach to find the best specification for eachMSA, and we omitted the controls that did not have an impact distinguishable from zero ontaxable sales and that failed to produce an increase in explanatory power, as measured by theadjusted r^ .

We identify two non-sports controls that affect the taxable sales ratio for the Miami MSAto improve the fit of the model. Hurricane Andrew, which devastated the South Floridaeconomy in 1992, had a dramatic effect on taxable sales in the Miami MSA. Taxable salesinitially fell in the area in the wake of the storm, then surged as residents rebuilt homes andreplaced damaged property. This pattern is modeled using two intervention variables: an initialpenalty during the month of the storm (August 1992) and a convex "ramp" (above pre-stormlevels) that lasted for three months after the storm. See Baade, Baumann, and Matheson (2007)for details and a sensitivity analysis of this specification. Hurricane Andrew controls for theTampa, Orlando, and Jacksonville models and did not produce significant results. In addition,controls for other costly hurricanes during our sample frame (e.g., the many hurricanes ofAugust-October 2004) did not produce significant results for any of the MSAs in our analysis.We also include a dumm y variable for the mo nth of September 2001 to account for the terroristattacks of 9/11, which had a negative and statistically significant impact on Jacksonville andMiami, but not on Tampa or Orlando.

The labor interruption variables in z, include the NBA lockout from November 1998through January 1999, the MLB strikes in August and September 1994 and April 1995, theNFL strikes in October and November 1982 and October 1987, and the NHL lockouts fromOctober 1994 through Janua ry 1995 and O ctober 2004 through April 2005. When strikescarried over into the off-seasons of their sports, only the months during which regular seasongames were lost were counted as strike periods. Of course, not every labor interruptionoccurred at the beginning or end of a m onth. T he results in Table 1 designate labor interruptionmonths as those in which at least half of the games were lost. Alternative specifications wereattempted and made little difference in terms of the results.

We also include controls for franchise expansions, which represent the entry of sportsfranchises in a manner similar to that by which labor interruptions represent a temporary exit.The franchise variables in z, include the expansion of the NBA into Miami (Miami Heat,November 1988) and Orlando (Orlando Magic, November 1989), the expansion of the NHLinto Miami (Florida Panthers, October 1993) and Tampa (Tampa Lightning, October 1993),the expansion of the MLB into Miami (Florida Marlins, April 1993) and Tampa (Devil Rays,April 1998), and the expansion of the NFL into Jacksonville (Jaguars, September 1995).

In addition, we also include controls for stadium and arena construction. The NFLstadiums constructed in Florida during our sample frame are the Raymond James Stadium inTampa (September 1998) and the Dolphins Stadium (originally Joe Robbie Stadium) nearMiami (September 1987). In the NHL, we include controls for the BankAtlantic Center in theMiami metropolitan area (October 1998), the St. Petersburg Times Arena in Tampa (October1996), and the Tam pa T hund erdom e (April 1990), which was used by the Tam pa Bay Lightningof the NHL until the 1996/97 season and was then subsequently renamed the Tropicana Fieldand was used for the expansion Tampa Bay Devil Rays of the MLB. In the NBA, we includea control for American Airlines Arena in Miami (January 2000). There are not separate


11/18

804 Baade, Bauman n, and MathesonTable 1. Sample 1980:1-2005:6: Miami. Dependent Variable: y* = In (Taxable Sales)VariableConstantPopulationTwo-month change in unemployment9/11 impactServices taxedHurricane Andrewinitial penaltyHurricane Andrewconvex rampNFL Strike, 1982NFL Strike, 1987NBA Strike, 1998/99NHL Strike, 1994/95.NHL Strike, 2004/05MLB Strike, 1994/95Joe Robbie Stadium, NFL, 9/1987BankAtlantic Center, NHL, 10/1998American Airlines Arena, NBA, 1/2000MLB Expansion, 4/1993NHL Expansion, 10/1993NBA Expansion, 11/1988Super Bowl, 1989Super Bowl, 1995Super Bowl, 1999NBA All-Star Game, 1990World Series, 1997World Series, 2003Stanley Cup Finals, 1997NHL All-Star Game, 2003In (taxable sales,-1)In (taxable sales,_2)In (taxable sales,-3)In (taxable sales,-4)Adjusted r^

Coefficient4.32050**5.48E-08**

-0.01137**-0.07116*0.05992**-0.07228*0.12611**-0.032080.0186-0.013210.010370.01965-0.01139-0.02568*-0.00249-0.008250.00871-0.017360.03186*0.03859-0.00320.0093-0.002680.012610.10201**-0 .01296-0.07173*0.33638**0.23928**0.15032*0.06150.9977

Breusch-Godfrey Lagrange Multiplier TestsLags Chi-Squared Statistic12345

0.550.6881.3554.3135.841

SE0.873431.32E-080.002970.03260.015950.032620.023490.023920.035010.021450.017470.013760.018980.014670.011240.011060.014140.015280.014190.023880.024280.025450.032360.03240.032850.022910.032570.060550.062040.062890.05418

/-Statistic4.954.16

-3 .8 3-2 .1 83.76-2 .2 25.37-1 .3 40.53-0 .6 20.591.43- 0 . 6-1 .7 5-0 .2 2-0 .7 50.62-1 .1 42.251.62- 0 . 1 30.37-0 .0 80.393.11-0 .5 7- 2 . 25.563.862.391.14

/-Value0.4580.7090.7160.3650.322

All dollar impact values are in 1982-1984 dollars using the CPI. The coefficients are reported with their associated/-statistic for the null hypothesis that the estimated value is equal to zero. SE = standard error.* Significant at the 10% level.** Significant at the 1% level.prevents us from separately estimating the effect of the franchise and the arena. Similarly, therenovation of Alltel Stadium in Jacksonville cannot be disentangled from the arrival of the new


12/18

Economic Impact of Professional Sports 805Orlando in 1992 and in Miami in 1990; the NHL Stanley Cup in Miami in 1996 and in Tampain 2004 and the NHL All-Star Game in Tampa in 1999 and in Fort Lauderdale (Miami MSA)in 2003; the M LB W orld Series in M iami in 1997 and 2003; the NCAA Men's Basketball FinalFour in Tampa in 1999; and FIFA's World Cup in Orlando in 1994 and the Major LeagueSoccer All-Star Game in 1998.As the Super Bowl generally occurs in either the last weekend of January or the firstweekend of February, the dummy variables for all Super Bowl years include both January andFebruary. This captures spending in preparation for the event, economic activity during SuperBowl week, and spending occurring several weeks after the big game, which should capturesome portion of the multiplier effect as local businesses and residents spend part of their SuperBowl windfall. Similarly, dummy variables for both the NBA and NHL finals cover both Mayand June, since the playoffs and finals can cover portions of both months. All other sportsvariables cover only the specific month in which the game(s) is played.

The results of ordinary least squares analysis on the model for the four cities are shown inTables 1 ^ . The tables show the coefficients, standard e rrors, and ^-statistics for each variable in themodel. The seasonal dummy variables are omitted from the tables for brevity. It should be notedthat several of these coefficients imply a reasonably large economic impact from professional sports(although they are much smaller than the average residuals reported in studies using annual data,such as those of Baade and Matheson 2004a, b). Again, given the size of these economies, the effectof even a large event with hundreds of millions of dollars of potential im pact is likely to be obscuredby natura l, unexplained variations in the economy that even a sophisticated econometric model m ayfail to detect. It is, therefore, unlikely that the models for an individual city will accurately capturethe effects of any one specific event, such as a new stadium or a Super Bowl.If professional sports do have a positive impact on a region's economy, however, acrossa large number of cities and events one should expect a recognizable pattern of increasingtaxable sales during periods with new stadiums, franchises, or large events and a pattern ofdecreasing taxable sales during labor interruptions. In fact, the coefficients of the sportsvariables appe ar to be almost entirely ran dom . Despite booste rs' claims that professional sportsresult in huge economic windfalls for host cities, over half of the sports variables examined inthis study resulted in reductions in taxable sales in host cities. Of the 43 sports variablesincluded in the four models, only 21 (49%) have the "c orre ct" sign indicating tha t the presenceof teams and events contributes positively to taxable sales in a metropolitan area. Just six of the11 labor interruptions have negative coefficients, only four of the 13 new stadium/franchisecoefficients are positive, and only 11 of the 19 mega-event coefficients are positive. This ishardly the economic stimulus that local taxpayers are routinely promised when teams andleagues claim, "If you build it, they will come."

Five of the sports variable coefficients are statistically significant at the 10% level, but again,roughly half of the signs are in the "wrong" direction. While the NBA expansion in Miami, the2003 World Series in Miami, and the 2001 Super Bowl in Tampa display a statistically significantpositive impact on their local economies, the 2003 NHL All-Star Game and the opening of JoeRobbie Stadium in 1987, both in Miami, produce statistically significant negative results. Ofcourse, with 43 sports variables in the model, one would normally expect four or five coefficientsto be statistically significant at the 10% level even if no true correlation between spo rts and taxab le


13/18

806 Baade, Baumann , and MathesonTable 2. Sample 1980:1-2005:6: Tampa. Dependent Variable: y* = ln (Taxable Sales)VariableConstantPopulationTwo-month change in unemployment9/11 impactServices taxedNFL Strike, 1982NFL Strike, 1987NHL Strike, 1994/95NHL Strike, 2004/05Raymond James Stadium, NFL, 9/1998Tropicana Field, MLB, 3/1990St. Pete Times Forum, NHL, 10/1996MLB Expansion, 4/1998NHL Expansion, 10/1993Super Bowl, 1984Super Bowl, 1991Super Bowl, 2001Super Bowl, 2004NHL All-star Game, 1999NCAA Men's Final Four, 1999ln (taxable sales,-1)ln (taxable sales,_2)ln (taxable sales,_3)Adjusted r^Breusch-Godfrey Lagrange Multiplier Tests

Coefficient2.06014**2.93E-08**0.03951**0.011070.02206-0.028430.0277-0.007830.01078-0.01724-0.004070.008130.00413-0.00203-0.021580.011660.07534**0.023440.033610.025040.39080**0.23144**0.27388**0.9974

Lags Chi-Squared Statistic1 0.382 0.8913 3.5624 3.7345 3.863

SE0.456969.87E-090.053930.035220.014170.02530.037380.018360.01480.016590.009940.010320.017520.008980.025060.025380.02510.025760.035250.035170.057540.058830.05393

/-Statistic4.512.975.080.311.56-1 .1 20.74-0 .4 30.73-1 .0 4-0 .4 10.790.24- 0 . 2 3-0 .8 60.4630.910.950.716.796.795.08

p-Yalue0.5380.6410.3130.4430.569All dollar impact values are in 1982-1984 dollars using the CPI. The coefficients are reported with their associated(-statistic for the null hypothesis that the estimated value is equal to zero. SE = standard error.* Significant at the 10% level.

** Significant at the 1% level.claims are demonstrably wrong. For example, in 1999-the NFL reported, "Thanks to SuperBowl XX XI II, there was a $670 million increase in taxable sales in South F lorida comp ared tothe equivalent January-February period in 1998" (National Football League 1999, p. 1).

Forgetting for the moment the questionable statistical practice of drawing a conclusionbased on a comparison of two years' worth of data, the data do indeed show that the FloridaDe par tment of Revenue rep orted tha t taxable sales increased by $640 million in the three-county region, including Broward, Dade, and Palm Beach Counties, in January-February1999, compared to the same period in 1998. (The $30 million discrepancy between the officialfigures and the numbers reported by the NFL is of little significance, except to indicate possiblesloppiness on the part of the League.)


14/18

Economic Impact of Professional Sports 807Table 3. Sample 1980:1-2005:6: Orlando. Dependent Variable: y* = In (Taxable Sales)VariableConstantPopulation

Coefficient1.57543**3.20E-08**

Tw o-mon th change in unemploym ent 0.02240*9/11 impactServices taxedNBA Strike, 1998/99NBA Expansion, 11/1989NBA All-Star Game, 1992World Cup, 1994NBA Finals, 1995MLS All-Star Game, 1998In (taxable sales,-1)In (taxable sales,_2)In (taxable sales,_3)In (taxable sales,_4)In (taxable sales,_5)Adjusted r^Breusch-Godfrey LagrangeLags12345

0.041470.03387*-0.00961-0.00239-0.04135-0.009670.00008-0.0250.27905**0.21591**0.17002**0.17089**0.07609**0.9974

Multiplier TestsChi-Squared Statistic0.6751.0141.5916.96

9.788

SE0.503221.14E-080.012770.052030.01950.029550.012120.051220.051270.036930.051140.061480.062110.063150.061560.05992

f-Statistic3.132.81.750.81.74-0 .3 3- 0 . 2-0 .8 1-0 .1 90-0 .4 94.543.482.692.781.27

/7-Value0.4110.6020.6610.1380.082All dollar impact values are in 1982-1984 dollars using the CPI. The coefficients are reported with their associated/-statistic for the null hypothesis that the estimated value is equal to zero. SE = standard error.* Significant at the 10% level.** Significant at the 1% level.

the presence of the Super Bowl. The model presented in this paper attempts to control for thesefactors (among others) and derives an increase in taxable sales from the 1999 Super Bowl inMiami of only $99.6 million, with a 95% confidence interval of $-434.8 to $634 million. Thus,a $670 m illion increase in taxable sales due to the Super Bowl can be rejected solely on the basisof this single observation. In concert with the host of other coefficients in this paper indicatinglittle or no positive economic impact from professional sports, the typical boosters' claims canbe even more soundly rejected.

One final note: Taxable sales in the area in January-February 2000, the year after thegame, were $1.26 billion higher than in the same months during the preceding year, yet theNFL never publicized a story proclaiming, "Thanks to the lack of a Super Bowl, there wasa $1.26 billion increase in taxable sales in South Florida compared to the equivalent January-February period in 1999."

6. Conclusions


15/18

808 Baade, Baumann, and MathesonTable 4. Sample 1980:1-2005:6: Jacksonville. Dependent Variable: y* = In (Taxable Sales)VariableConstantPopulationTwo-month change in unemployment9/11 initial impact9/11 permanent impactServices taxedNFL expansion, 9/1995Super Bowl, 2005In (taxable sales,-1)In (taxable sales,-2)In (taxable sales,-3)In (taxable sales,_4)In (taxable sales,-5)Adjusted r^

CoefTicient3.1061**5.06E-08**0.02576**-0.47586**-0.06291**0.04224**-0.044230.033580.38395**0.15316**0.16287**0.068690.060630.9955

Breusch-Godfrey Lagrange Multiplier TestsLags Chi-Squared Statistic12345

0.2334.0664.4125.1355.246

SE0.623561.15E-080,010840.044080.016320.016870.043330.031210.053550,056180.054070.054640.04864

/-Statistic4.984.422,38-1 0 .8-3 .8 62.5-1 .0 21.087.172.733.011.261.25

/7-Value0.6290.1310.220.2740.387All dollar impact values are in 1982-1984 dollars using the CPI. The coefficients are reported with their associated(-statistic for the null hypothesis that the estimated value is equal to zero. SE = standard error.

* Significant at the 10% level.* Significant at the \% level.incentive for host cities to construct new stadiums or arenas at considerable public expense. Inthe past, league- and industry-sponsored studies have estimated that mega-events such as theSuper Bowl and all-star games increase economic activity by hundreds of millions of dollars inhost cities. Similar studies claim that new stadiums or franchises also can have hundreds ofmillions of dollars of annual local economic impact. Our detailed regression analysis of taxablesales in Florida over the period from 1980 to mid-2005 fails to support these claims. Newstadiums, arenas, and franchises, as well as mega-events, appear to be as likely to reducetaxable sales as to increase them. Similarly, strikes and lockouts in professional sports have notsystematically reduced local taxable sales. While these results, like any econometric estimates,are subject to some degree of uncertainty, they clearly place doubt on boosters' claims of hugeeconomic windfalls. Cities would be wise to view with caution economic impact estimatesprovided by sports boosters, who have a clear incentive to inflate these estimates. It wouldappear that "padding" is an essential element of many games both on and off the field.

ReferencesAckman, Dan . 2000. "In M oney Terms, the Subway Series Strikes Out." Forbes Magazine, Oct 21 , 2000. Available http://

www.forbes.com/2000/10/21/1021series/html. Accessed 15 Augu st 2007.


16/18

Economic Impact of Professional Sports 809Baade, Robert, and Victor Matheson. 2001. Home run or wild pitch? Assessing the economic impact of major leaguebaseball's all-star game. Journal of Sports Economics 2:307-27.Baade, Robert A. 1996. Professional sports as a catalyst for metropolitan economic development. Journal of Urban

Affairs 18:1-17.Baade, Robert A., Robert Baumann, and Victor A. Matheson. 2007. Estimating the economic impact of natural andsocial disasters with an application to Hurricane Katrina. Urban Studies 44 .

Baade, Robert A., and Victor A. Matheson. 2000. An assessment of the economic impact of the American footballchampionship, the Super Bowl, on host communities. Reflets et Perspectives 34:35-46.

Baade, Robert A., and Victor A. Matheson. 2002. Bidding for the Olympics: Fool's gold? In Transatlantic sport: Thecomparative economics of N orth American and European sports, edited by C. P. Barros, M. Ibrahimo, and S.Szymanski. London: Edward Elgar Publishing, pp. 127-51.Baade, Rob ert A., and Victor A. Matheso n. 2004a. The quest for the cup: Assessing the economic impact of the Wo rldCup. Regional Studies 38:341-52.Baade, Rob ert A., and Victor A. Ma theson. 2004b. An economic slam d unk or march madness? Assessing the economicimpact of the NCAA basketball tournament. In Economics of college sports, edited by J. Fizel and R. Fort.Westport, CT: Praeger Publishers, pp. 111-33.

Baade, Robert A., and Victor A. Matheson. 2005. Striking out? The economic impact of major league baseball workstoppages on host communities. College of the Holy Cross, Department of Economics Working Paper No. 05-07.Baade, Robert A., and Victor A. Matheson. 2006. Have public finance principles been shut out in financing newstadiums for the NFL? Public Finance and Management 6:284-320.

Coates, Dennis. 2006. The tax benefits of hosting the Super Bowl and the MLB all-star game: The Houston experience.International Journal of Sports Finance 1:239-52.Coates, Dennis, and Craig A. Depk en, II. 2006. Mega-events: Is the Texas-Baylor game to W aco what the Super Bowl isto Houston? International Association of Sports Economists Working Paper No. 06-06.

Coates, Dennis, and Brad R. Humphreys. 1999. The growth effects of sports franchises, stadia and arenas. Journal ofPolicy Analysis and Management 14:601-24.Coates, Dennis, and Brad R. Humphreys. 2001. The economic consequences of professional sp orts strikes and lockouts.Southern Economic Journal 67:737-47.Coates, Denn is, and Brad R . Hum phreys. 2002. The economic impact of post-season p lay in professional sports. Journatof Sports Economics 3:291-9.Coates, Dennis, and Brad R. Humphreys. 2003. The effect of professional sports on earnings and employment in theservices and retail sectors in US cities. Regional Science and Urban Economics 33:175-98.Feit, Josh. 2006. Key amendmen ts: Council member Nick Licata challenges Sonics subsidy. Th e Stranger. Seattle's O nlyNews Section. 23 February-1 March 2006.

Finer, Jonathan. 2002. The grand illusion. Far Eastern Economic Review 7 March:32-6.Golovnina, Maria. 2002. "S. Korean Tourism Sector in Blues Despite World Cup." Forbes.com. Available http://www .forbes.com/newswire/2002/06/19/rtr636036.html. Posted 19 Jun e 2002. Accessed 20 June 2002.Houck, J. 2000. "High-Stake Cou rtship." FoxSportsBiz. Available http://www.foxsports.com/business/trends/z000120aUstarl.sml. Accessed 21 Janua ry 2000.Humphreys, Jeffrey M., and Michael K. Plummer. 1995. The economic impact on the state of Georgia of hosting the 1996Summer Olympic games. Athens, GA: Selig Center for Economic Growth, The University of Georgia WorkingPaper.Knight Ridder News Service. 1999. "Final Four's Financial Impact Hard to Gauge." Enquirer Sports Coverage, March25, 1999. Available http://enquirer.com /editions/1999/0 2/25/spt_final_fou rs.html. Accessed 30 Aug ust 200 1.

Matheson, Victor A. 2004. Economic multipliers and mega-event analysis. College of the Holy Cross Working PaperSeries No. 04-02.Mensheha, M. 1998. "Home-Court Edge: Final Four Promises to be Economic Slam Dunk." San Antonio Business

Journal, March 27, 1998. Available http://sanantonio.bizjournals.com/sanantonio/stories/1998/03/30/story3.html.Accessed 15 August 2007.National Football League. 1999. Super Bowl XXXIII generates $396 million for South Florida. NFL Report 58:7.Oregon Baseball Campaign. 2002. "Why MLB, Why Oregon, Why Now?" Available http://www.oregonbaseballcompaign.

com/fact.htm. Accessed 5 September 2002.Porter, Philip. 1999. Mega-sports events as municipal investments: A critique of impact analysis. In Sports economics:Current research, edited by J. Fizel, E. Gustafson, and L. Hadley. Westport, CT: Praeger Press, pp. 61-74.

Rosentraub, Mark. 1994. Sport and downtown development strategy. Journal of Urban Affairs 16:228-39.


17/18

810 Baade, Baumann, and M athesonSelig, Bud, John H arring ton, and Jim Healey. "Ne w Ballpark Press Briefing: July 12, 1999." Available http://

www .asapsports.com/baseball/l999allstar/071299BS.html. Accessed 29 August 2000.Siegfried, John, and Andrew Zimbalist. 2000. The economics of sports facilities and their communities. Journal of

Economic Perspectives 14:95-114.Siegfried, John, and Andrew Zimbalist. 2002. A note on the local economic impact of sports expenditures. Journal of

Sports Economics 3:361-6.Zipp, John F. 1996. The economic impact of the baseball strike of 1994. Urban Affairs Review 32:157-85.Zipp, John F. 1997. Spring training. In Sports, jobs, and taxes: The economic impact of sports teams and stadiums, edited

by Roger G. Noll and Andrew Zimbalist. Washington, DC: The Brookings Institution Press, pp. 427-51.


18/18

Documents

Selling the Game