No. 62-April4,
1994
Stadiums, Professional Sports, and
Economic Development:
Assessing the Reality
by Robert A. Baade
About one century ago, professional sports becameprominent in American public
life. During its early years, the businessof sports was primarily a private undertaking,
financed with private money and played in private stadiumsand arenas.
But state and local governmentsubsidiesto professional sports businesseshave
proliferated over the past few decades,and economic argumentshave been crafted to justify
the subsidies, These argumentstypically rest on the assertionthat professional sports is a
significant, evenunique, catalyst for economic growth. By this reckoning, stadiumsand
teamsare “cash cows” that expand the economy and enablefurther public investmentin
other critical areas.
Public funds are increasingly scarce.We must test the argument that professional
sports offer an important return on governmentsubsidies.The purpose of this paper is to
use economic theory and empirical techniquesto assessthe contribution of professional
sports to metropolitan area economic developmentin the United States.
The study consistsof five parts. Part 1 briefly reviews the economic literature on
professional sports and urban development. Part 2 discussesthe ways professional sports
can have an economic impact on an area and explores the challengesinherent in measuring
this effect through “expenditure” and “multiplier” analysis.
Part 3 briefly introducesthe empirical approachthe author has adoptedin this study.
In Part 4, the author presentsa new statistical analysisof the relationship between sports
and municipal economic activity. His analysis involves 48 cities and the metropolitan areas
around them over a 30-year period. Included in his analysisis every U.S. city that had
either a new professional sports stadium or arena, or a professional baseball, football,
basketball, or hockey team during this period.
Afrer a thorough examination of an unprecedentedquantity of data related to
professional sports and host area per capita personal income, the author finds no factual
basisfor the conventionalargument that professional sports stadiumsand teams have a
significant impact on a region’s economicgrowth. Part 5 is comprised of the author’s
summary and concluding remarks. A more detailed technical discussionof the methodology
and results employed in the study appearsin AppendicesA and B.
-2-
PART 1
Brief Review of the Literature
Objective assessmentsof the impact of professional sports on urban economic
developmentin the United Statesare a relatively recentphenomenon,inspired mainly by the
aggressivepromotion of professional sports as a solution to the fiscal crises that plague
American cities. What have researchersconcludedabout the efficacy of pro sports as an
economicdevelopmenttool?
Benjamin Okner elevatedthe debateabout sports stadium funclmg to a national level
in a 1974 article in Governmentand the SportsBusiness,a book published by the Brookings
Institution. < 1> Having examinedstadiumsacrossthe U.S., Olcnerconcludedthat most
producedrevenuesinsufficient to cover their costs.
In 1990, economist Dean Bairn of PepperdineUniversity lent support to Okner’s
analysis, using standardfinancial principles to analyzegovernmentsubsidiesto fourteen
sports stadiumsacrossthe country. < 2 > Bairn found that thesestadiumshad a total net
accumulatedvalue of negative $139.3 million. In thirteen of the fourteen stadiums studied,
tax dollars “invested” by financing governmentsin professional sports stadiumsproduceda
financial loss for the governmentunit when tax revenuesfrom the project were comparedto
the tax dollars invested. < 3 >
In a 1992 publication, researchersJamesQuirk and Rodney Fort reachedconclusions
similar to those of Bairn and Okner. For ten of the thirty-nine stadiumsfor which they were
able to obtain data, Fort and Quirk found that stadium revenueswere insufftcient to cover
operating expensesexclusive of debt service and depreciation. < 4 >
But if the history of sports stadiumsis indeed “written in red ink,” can it be that
stadium investmentsproduce other economicreturns that justify public subsidies?In the
standard“civic investment” argument, there is no necessaryinconsistencybetween a losing
private enterpriseand a profitable public activity. In fact, a stadium booster will argue that
it is precisely becausethe stadium producesbenefits to the public but lossesfor a private
owner that tax support of sports stadiumsis justified-indeed, essential.
Thus, an assessmentof the worthinessof stadium subsidiesrequires that one
consider the public, opposedto merely the private, benefits and costs. Recognizing this,
scholarshave sought ways to measurethe “externalities” or “spin-off benefits” accruing
from stadium activities. Researchon this issuehas producedwidely divergent results.
-3-
For example, a 1985 study by University of PennsylvaniaresearcherEdward Shils
estimatedthat Philadelphia’s professional sports teamscontributed more than $500 million
to the city’s economyin 1983. <5> More recently, a report by Kenneth Clark concluded
that the New York Yankeescontribute $200 million annuatly to New York City’s
economy. < 6 > In contrast, a 1986 analysisby Baltimore researcherHal Lancaster
estimatedthat the Colts (now the Indianapolis Colts) generateda mere $200,000 in
additional economic activity when they played in Baltimore. < 7 >
These divergent results were compelledby different assumptionsabout spending
patterns. < 8 > It is also worth noting that me Philadelphia study was funded by a
consortium of the City’s professional teams, while the Baltimore study was written shortly
after the Colts had bolted for “greener pastures” in Indianapolis.
In 1987, Robert Baadeperformed an independentempirical study of the effect of
sports stadiumsand pro sports teamson a metropolitan economy. < 9 > Baade’s
examinationof economic growth data for nine cities before and after the adoption of new
stadiumsor professional sports teamssuggestedthat pro sports impact was not significant.
In five of the nine cities analyzed, stadium renovation or construction, or a city’s adoption
of a professional football or baseballteam had a significant negativeimpact on metropolitan
income as a fraction of regional income. In no instancedid a positive, significant
correlation surfaceamong stadiums,professional sports, and city income as a fraction of
regional income. < 10>
Similar researchby Baadeand Richard Dye in 1988 < 11> and again in
1990 < 12> supportedBaade’s earlier fmdings. Two recent popular books by
economists-Puydirt and Baseball and Billions < 13>-have demonstrateda similar
skepticismregarding the unique economic impact often attributed to professional sports.
Summary. The literature concerninginvestmentin pro sports stadiums suggeststhey
often fail to provide a positive direct return on the money investedin them. Studies of the
value of a pro sports team or stadium to an area’s economyhave produced widely divergent
results, dependingon the methods employed and the assumptionsmade in crafting the
study. Thesemethods and assumptionsare discussedin Parts 2 and 3.
-4-
PART 2
Basic Principles,
Three Types of Economic
Expenditures,
and Multipliers
Impact
Stadiums, arenas,and professional sports can bring three sourcesof economic
benefits to a metropolitan area: direct expenditures,indirect expenditures,and
“psychological” benefits.
Direct expendituresrepresentthe money spentby the professional sports franchise,
its employees,and its patrons. As JamesQuirk and Rodney Fort explain,
The procedurethat is used to estimatethe economicbenefits provided by a
team or a facility is first to estimatethe direct expendituresby the team for
goods and servicesin the city, and then to add to this expendituresby fans on
goods and services(other than game tickets) purchasedin the city, together
with expendituresby players on purchasesof goods and servicesin the city.
The resulting sum is the amount of direct expenditurebenefits to the city
provided by the team. < 14>
Thesedirect expenditures-on restaurants,hotels, transportation, souvenirs, food,
and the like-are received as income by metropolitan-areabusinesses.Thesebusinessesand
their employeesthen spenda fraction of this income on other goods and serviceswithin the
metropolitan area, and thosebusinessesand their employeesspenda fraction of their
income at other metropolitan-areabusinesses.The processcontinues, and the second,third,
and subsequentrounds of spendingconstitutethe indirect expendituresthat benefit a region.
Since direct expendituresresult in indirect expenditures,the direct expendituresare
said to “multiply” through me economy. Economistsattempt to quantify this effect by
calculating a “multiplier. ” For a given level of direct expenditures,higher multipliers
indicate higher levels of economic impact.
In addition to direct and indirect expendituresare “psychological” benefits, which
are less easily quantified. Some have suggested,for instance,that television coverageof a
city’s professional sports teamsencouragesbusinessesto locate there.
It is conceivablethat relocating businessesmay view professional sports as adding to
their employees’ “quality of life” at a potential new location. Speculationaside, no evidence
exists to suggestthat professional sports is an important factor in businesslocation
decisions. For example, Baadeand Dye < 15> found no connectionbetweenprofessional
snorts and manufacturing activity in approximately 90 percent of the tests conductedfor a
number of cities in the United States.It is likely that other factors, such as the tax
environmentand the existenceof a skilled labor force, determinebusinesslocation to a far
greater extent than the presenceof professional sports.
It is, of course, possible that hosting
professional sports may have negative
exists to suggest hat professional
psychological effects that ultimately burden
the economy. In any event, the empirical
test employedlater in this study will detect
economicgrowth derived from any source
related to the presenceof professional
sports, including psychological benefits that hosting a team may induce. < 16>
How Direct and Indirect
Expenditures
Can Become Net Benefits
In standardeconomicmodels, metropolitan economicgrowth comesfrom new
“export sales” and “import substitution.” Increasedexport salesresult from attracting net
new inflows of spendingfrom outside the area. This regional increasein exports might
occur if, for example, people from another region decide to attend a baseballgame in the
area, rather than go to their local movie theater. If, on the other hand, people from another
region spendmoney at an area stadium rather than at a movie theater or restaurantnear the
stadium, the stadium is not increasingexport sales-it is simply shifting them.
Import substitution occurs when a community keepsmoney that residentsmight have
spent elsewhere.If residentschooseto go to a local football stadium instead of an
entertainmentevent outside the area, it can be said that the stadium has becomean import
substitute.If, however, residentsspendmoney at the stadium rather than at other local
businesses,the stadium causesonly a shift in spending, and no import substitution occurs.
The net new local spendingthat results from increasedexport sales and import
substitutionmay multiply and drive an expansionof locally produced secondarygoods and
services.The size of the multiplier dependson the locus of the subsequentspending.If the
new income is likely to be spenton locally producedgoods, and if this income is likewise
likely to be spenton locally producedgoods, the multiplier will be larger. If, however, the
athletesand team executivesspendtheir income outside the area, or if the concessionaires
import their semiftished goods from outside the area, the multiplier will be smaller.
How high are multipliers for professional sports?The answersto this questionhave
varied greatly. In studiessupporting stadium subsidies,multipliers greater than
three-suggesting a considerableeconomic impact, given the size of the direct
expenditures-have been estimated. < 17> The recent norm for stadium multipliers,
however, is two or less.
Even these lower multipliers are suspect,however, since they are basedon
assumptionsabout where eachdollar is spentand would be spent in the absenceof sports.
Determinii this information for a particular sports activity is extremely difficult. The
amount of money, the number and residenceof consumers,and the number of businesses
involved meansthat tracing each individual dollar is extremely difficult. < 18>
Not surprisingly, those who estimatethe greatesteconomic impact for professional
sports assumethat all the direct expendiruresrepresentsa net increasein local spending.
These studiesimply that all sports-relatedspendingis either new export salesor import
substitution, and that much or even all of the re-spendingstays inside the
community. < 19> Less optimistic studies, including some of the author’s previous work,
have suggestedthat sports income representsmainly a shift of metropolitan-areaspending,
and that there are few new export salesand little import substitution. < 20 >
Opportunity
Cost
Any analysis of the economicimpact of sports would be incomplete without an
acknowledgementthat a city, county, or stategovernmentforgoes other opportunities when
it builds a stadium or otherwise subsidizesa sports franchise. The value of the best of these
other options is defined as the “opportunity cost” of the subsidy.
A thorough assessmentof the value
Offkials must evaluate whether
of sports as a public investmentrequires not
the subsidized sports business
merely an examination of whether a sports
would have the greatest net impact
subsidy would have any net impact on a
metro area’s economic development.
of available alternative uses of the
Officials must evaluatewhether the
money.
subsidizedsports businesswould have the
grearesrnet impact of available alternative
usesof the money. The alternative of not taxing the money from citizens in the first place
should always be included in this evaluation. < 21>
l-
All resourcesare scarce.Local developmentauthorities have limited budgets;
tax-exemptindustrial developmentbonds once used for stadium finance are now
restricted; < 22 > and receipts from gambling and lotteries are not available for schoolsor
public transportationwhen used for stadiums.
The “political capital” to sell projects to those who pay the taxesor lose from the
economicredistribution is also limited. Coaxing taxpayersto subsidizea stadium may mean
the defeat of higher tax levies for other projects. Public officials must inevitably ask, Which
is more valuable-more money for schools, roads, airports, police, tax reduction, or
stadiums?
Short-term
Versus Long-term
Thinking
Even if expenditureand multiplier estimatessuggestthat subsidizedsports
developmentis better than other options, the long-term impact of a sports strategy must still
be considered.Since the presenceof a stadium or sports team may changethe structure of
the local economy, the long-run impact may be less desirable than that suggestedby the
direct expenditure-multiplier approach.If money spenton sports shifts resourcesfrom the
manufacturing sector, and if manufacturing sectorjobs are higher-wage and more stable,
then pursuit of a sports developmentstrategymay ultimately slow a region’s economic
growth.
Summary
The discussionabovesuggeststhe following points:
(9
Direct expendituresbenefit a region’s economyonly when they representnet new
spendingthrough import substitutionor new export sales, or when they generate
re-spending(indirect expenditures)that is new to the community.
(ii)
Multiplier estimatesare necessarilybasedon assumptionsabout indirect
expenditures.Bias in theseassumptionscan substantially influence the estimates
either in favor of or againstthe thesis that professional sports is a valuable
economicdevelopmenttool.
(iii)
Economic impact estimates,even when rigorously prepared, must be
comparedto economic impact estimatesgeneratedfor other usesof scarce
public funds.
(9
Expenditure and multiplier estimatescan overlook the importance of long-term
economic trends, encouraginga short-term developmentstrategy that ultimately
leavesan area lagging in long-term economicgrowth.
PART 3
Approaches
The Standard
to Testing the Economic Development
Potential of Professional Sports
Model
The orthodox “impact model” of economicdevelopmentuses expenditureand
multiplier estimatesto determine the value of sports-baseddevelopmentstrategies.But
without knowledge of the origins of direct expenditures,this type of study, which often
treats all direct spendingon sports as net new spending,can vastly overstatea stadium’s or
team’s contribution to the urban economy. <23 > Indirect expendituresare derived as a
multiple of direct expenditures,meaning that errors in estimatingnet direct expenditure
increasesare compoundedin calculating overall spendingincreases.
In defending the treatment of all sports expendituresas net new expenditures,some
have opined that there is no alternative to doing so, since it is not feasible to identify the
spendingfor which pro sports substitutes.While it is true that identifying the origins of
sports spendingand its net contribution to spendingis extremely difficult, such an
identification is precisely what a rigorous expenditure-multiplier appraisal of the economic
impact requires. Given thesedifficulties with the standardimpact model, this study proceeds
by analyzing empirical evidenceabout the economicimpact of sports. This evidenceis
basedon a comparisonof metropolitan area income before and after hosting a team or a
new stadium or arena.
Empirical
Comparison
of “Before”
and “After”
Profiles
“Before-and-after” comparisonsrecognizethe complex nature of the relationship
betweenpro sports and economic activity. Rather than trying to calculateeconomic impact,
the analyst observesthe changesthat actually occur in the urban economiclandscape.If
indeed sports teamsor stadiumsresult in economicgrowth, there should be a measurable
changefrom “before” to “after.”
Economiststypically use a statistical method called “regressionanalysis” to identify
the degreeto which a changein one characteristicinfluencesa changein another
characteristicover time. Regressionanalysiscan thus help determine the extent to which an
area’s economicperformance varies in conjunction with changesin the number of the area’s
sports teamsor sports stadiums. Such an investigation can capturedifferences in economic
-9-
performanceboth over time (time-seriesanalysis) and acrosssimilar economicunits, such
as cities (cross-sectionalanalysis).
In applying regressionanalysis to the questionof pro sports’ economic impact,,the
fundamentalconsiderationis whether professional sports and sports stadiumshave an
economicimpact that is statistically significant-i.e., an effect greater than what could
reasonablybe expectedto occur by chance.This task must be performed with care, since
changesin the professional sports industry are often accompaniedby other important
changes,such as modification of the state’s tax code or the decline of key industries. If
important factors are omitted from the model, the researchermay incorrectly attribute
accelerationor decline in economicgrowth to the presenceof a team or a stadium.
In Part 4, the question of whether pro sports have a statistically significant economic
impact is addressedwith a new analysisof the relationship betweensports and municipal
economicactivity. A “fxed effects” model is used to help separatethe effects of sports
investmentfrom those of other economicchanges.The results provide data sufficient to
evaluatethe performance of professional sports as an economic developmenttool.
- 10
PART 4
Analysis of the Economic Impact
of Professional Sports
Methodology
A test for the statistical significanceof pro sports’ economic impact should adjust for
the fact that a city’s pattern of economic activity mimics national, regional, and statetrends
even as it follows a general trend line peculiar to itself. The extent to which a city follows
national, regional, state, and city seculartrends may be termed the “fxed” economic effects
on the city. Once thesefmed effects have been factored out of changesin a city’s economic
activity, what remains are changesinspired by factors unique to a city. This “unique
behavior” can be investigatedfor evidencethat the presenceof professional sports teamsor
sports stadiumshas contributed significantly to an area’s economy.
The empirical study conductedfor this paper employs the approachdescribedabove
by applying the equation given in Appendix A (page28) to 36 metropolitan areasthat
hostedprofessional sports teamsduring the thirty-year period from 1958 through 1987. The
equationrelates the presenceof pro sports or new pro sports stadiumsor arenasto a metro
area’s real, per capita personal income growth relative to the other cities in the sample, and
relative to its own growth pattern. (Becausethesereal, per capita personal income growth
figures are adjustedrelative to income trends, they are referred to as “trend-adjusted”in the
discussionbelow.)
These 36 cities were chosenbecausethey comprise all of the U.S. metropolitan
areasthat hosted either a professional team in one of the four major team sport-baseball,
football, basketball, or hockey-or a new professionalsports stadium or arena-i.e., ten
years old or less-during this period. < 24 > Each “metropolitan area” is representedby
the city’s “Metropolitan Statistical Area” (MSA) as defined by the U.S. Department of
Commerce. < 25 >
To ensurea broader, more representativesample, 12 MSAs that have nor hosted a
professionalteam, ostensibly becauseof their size, also were included in the sample. Thus,
data for a total of 48 MSAs were incorporatedinto the model.
This empirical investigation allows us to estimatethe extent to which yearly changes
in eachmetro area’s per capita real income, adjustedfor general trends in the 48 cities’
economicgrowth, are explained by changesin either the number of new stadiumsor
professionalsports franchisesthe area acquires. < 26 > The study includes the entire MSA,
- 11 -
rather than just the central city, since the entire region is more likely to capture all of the
indirect expendituresthat result from hosting professionalsports than is the host city alone.
A technical discussionof the methodology appearsin Appendix A. Key points about
the model are discussedbelow. Throughout this discussion,the word “stadiums” will refer
to both stadiumsand arenas.
Possible Objections to the Model. Someanalystsmight argue that this model
demandstoo much from professional sports, Since the professional sports industry is small
relative to a large city’s economy, <27 > it is unlikely that professional sports plays a
statistically significant role in determining real per capita income.
In fact, however, the model’s test for statistical significancedoes not demandsuch
impressiveeconomicperformance from professional sports. To establishthat professional
sports contributespositively to a city’s economy, the model requires only that sports be
significant in determining changesin real, per capita personal income that are not explained
by trends.
It might also be argued that the model missesgrowth that personal income numbers
do not capture. White it is true that other measuresmay prove to be of interest-and indeed
the author himself investigatessuch measureselsewhere < 28 > -it is also true that
economistsgenerally seereal income growth as the best single measurementof economic
growth. Other measuresof growth-such as increasesin the numbers of jobs or businesses
in an area-will tend to be reflected in real personalincome growth.
This study focuseson a stadium’s economicimpact only during the years in which it
is being used, not during its construction. The economicimpact of the actual constructionof
a stadium is likely to be relatively small and of limited duration. Even if the constructionof
a stadium does have a significant, lasting impact on an area’s economy, this effect would be
measuredas it persistedinto the early years of the stadium’s use.
A fmal objection to the model might be that our fmdings will be incomplete even if
the regressionanalysis doesnot indicate a significant economicbenefit for Metropolitan
Statistical Areas. By this reasoning, a downtown stadium investmentmay not benefit an
MSA as a whole, but it may changethe distribution of income within the MSA by reversing
flight from downtown to the suburbs. A stronger “core” area in the central city would then
be the basis for long-term growth for the entire region.
This is an interesting proposition, and one deservingresearch.But severalthings
may be said about it immediately. Fiist, testing this theory empirically will be difficult. The
trend over the past few decadeshas not been the movementof teams from the suburbsto
downtown, but rather from downtown to the suburbs.Instancesof movement from the
- 12 -
suburbsto downtown are few and recent, meaning that little valid statistical analysis can be
performed at present regarding the actual outcome of such inner city “reinvestment.”
Second,this theory dependsin part on enoughof the indirect expendituresgenerated
by the stadium staying in the downtown region. Otherwise, net downtown spending
increasesmay not be sufficient to offset the cost to the inner city of building the stadium.
As noted in Part 2, it is quite possible that indirect expendituresoccur outside the area
around the stadium. This “leakage” is the reasonthe present study considersentire
Metropolitan Statistical Areas, rather than core urban areas.
Third, if pro sports developmentsubsidiesmake inner cities stronger, thus providing
a fm basis for regional growth that could not otherwise occur, the model employedhere
should detect this fact. If thoseMSAs with downtown stadiumsexhibit significant growth
related to pro sports investment,while those with stadiumsin suburbanareasdo not exhibit
such growth, the model will provide some support for this theory of downtown sports
investment.If such a trend does not emerge, the model will provide no support for the
theory.
Strengths of the Model. The model employedhere addressestwo concernsraised
by the analysisin Part 2. Fist, the model provides evidenceof whether sports increases
economicactivity or simply realigns it within the metropolitan area. It is unlikely that a
statistically significant relationship betweensports and economicgrowth will emergeunless
there is a net increasein economic activity.
Moreover, the model provides evidenceregarding the power of a sports-development
strategy relative to alternative strategiesfor creating economicgrowth. The empirical study
includes data for cities that did not host professional sports, and the model likewise
comparessports cities undertaking new sports investmentswith sports cities that are not. If
sports developmentdoesnot demonstratea statistically significant impact, the results will
suggestthe possibility that after controlling for other trends, economicgrowth in areasthat
are making no new sports investmentscomparesfavorably with economicgrowth in areas
that are. This suggestionwould provide additional reasonto question the assertionthat
sports developmentsubsidiesprovide a unique economic stimulus. Alternatively, a
consistentresult of statistical significancewould suggestthat new sports developmentbrings
unique economicbenefits not realized by sports and non-sportscities that do not undertake
new sports development.
Note three other features. First, only stadiumsten years old or less are consideredin
the model. This restriction was adoptedbecauseresearchhas suggestedthat stadiumswill
have less effect on sports spending,especiallyby fans, as the stadiumsbecomeolder. In
Government and the Sports Business, < 29 > Stanford University economistRoger No11
determinedthat a “novelty effect” influencesbaseballattendanceafter the constructionof a
new stadium, This effect lasts somewherebetweensevenand elevenyears. Including only
- 13 -
stadiumsless than elevenyears old increasesthe chancethat pro sports will attract large
direct expendituresand demonstratea significant, positive impact on metro area economic
growth.
Second,no attempt has beenmade to focus on publicly subsidizedsports teamsand
stadiums,or to treat suchprojects differently in the model. This model considersonly the
presenceof the facilities or teamsthemselves. < 30 > This stipulation was also designedto
favor the thesis that sports brings significant economicdevelopment.
Third, the model usesper capita personal income growth, rather than total personal
income growth, to measureeconomicgrowth rates. Unlike total personal income growth,
per capita figures do not prejudice the caseagainstbig city sports investment, where large
total personal income is often the norm. The model should be able to detect pro sports’
economic impact-if it exists-in cities of any size.
In summary, then, the model employed in this paper should capture the economic
benefits, if any, of professional sports teamsand stadiums. The model will also help us
answer questionsraised by the economicanalysis earlier in the paper.
Before proceedingto the results of the regressionanalysis, it is important to note
that this analysis is concernedonly with the “special impact” hypothesisused to justify
governmentsports subsidies.Even if the findings below do not support this hypothesis, the
rationale for private sports investmentis in no way affected. Private investmentin sports
teamsor sports stadiums makessenseeconomically if the owner can turn a profit from
them. A demonstrationof a significant, positive economiceffect on a host area should not
be seenas a prerequisite for allowing private sports investment.Most investmentin private
businesses-evenvery successfulones-would fail to demonstratethis kind of impact.
Results
Table A of Appendix B (page 34) shows detailed results from the regression
analysis. For the interestedreader unfamiliar with econometrics,Appendix B also includes a
“key” to understandingthe regressionresults (page 33). A summary of the results of the
analysisis presentedbelow.
Impact on metropolitan areas. As noted earlier, the model includes the real,
per capita income growth of 48 U.S. cities from 1958 through 1987. Duriig this period, 36
cities hosted a professional team in one of the four major team sports (baseball, football,
basketball, and hockey). The results of the tests for these36 MSAs are recorded in Table I.
In those caseswhere a statistically significant impact emerged,the table also
indicates whether the impact was “positive” or “negative.” A positive outcome indicatesthat
14
teamsor stadiumshad a statistically significant andpositive effect on the area’s real,
per capita income growth. A negativefmding indicatesthat teamsor stadiumshad a
statistically significant and negative effect on the area’s real, per capita income growth.
Although we have discussedrepeatedlyhow professionalsports might serve as an economic
stimulus, it is also possible that professionalsports might serve as a net economicburden.
The results presentedin Table I overwhelmingly indicate that professional sports is
not statistically significant in determining real, trend-adjusted,per capita personal income
growth:
W Of the 32 MSAs where there was a change in the number of sports teams, 30
MSAs showed no significant relationship between the presenceof the teams and
real, trend-adjusted, per capita personal income growth. In the remaining two
cases,the presenceof sports teamswas significantly positive once (in Indianapolis)
and significantly negativeonce (in Baltimore).
R Of the 30 MSAs where there was a change in the number of stadiums or arenas
ten years old or less, 27 MSAs showed no significant relationship between the
presenceof a stadium and real, trend-adjusted, per capita personal income
growth. In all three of the remaining cases(St. Louis, San Francisco/Oakland,and
WashingtonD.C.), the presenceof a sports stadium was significantly negative.
Regional Impact. Although individual cities fail to exhibit a statistically significant
impact, it may be that professional sports has a positive effect when a number of metro
areaswithin a region are considered.It may be, for example, that New York, Philadelphia,
and other large cities in the easternpart of the United Statescollectively export stadium and
sports servicesto other regions of the country. Large conventionsutilize stadiumsand
arenas,and super-stationsand cable television broadcastgamesnationally, increasing the
likelihood that stadiumsand teamsare successfullymarketed to the rest of the country.
Results for tests conductedregarding the impact of stadiumsand teams on regions
are recorded in Table II. Detailed results from the regressionanalysis for eachregion can
be found in Appendix B, Table B (page 38).
15
TABLE I
The Impact of New Stadiums and Professional
Sports Teams on Real, Trend-Adjusted, Per Capita Personal
Income Growth For SelectedU.S. Cities (MSAs)
City/Result
City/Result
Atlanta
Hartford
NO
NO
Baltimore
Houston
NO
NO
YES
POSITIVE
NO
Buffalo
Kansas City
NO
NO
Charlotteb
Los Angeles
NO
NO
NO
NO
NO
Chicago
Cincinnati
NO
Milwaukee
NO
NO
MiMeapolis
NO
Dallas
New Orleans
NO
Denver
New York
NO
NO
I
Cleveland
NO
I
NO
NO
I
Detroit
NO
N.A.
Orlando
NO
I
I
Green Bay
N.A.
NO
I
N.A.
Philadelphia
16.
NO
I
NO
TABLE I
New
Stadium
Presence
Statistically
Sienificant’
Pittsburgh
1
NO
1
Team
Presence
Statistically
SignificanV
NO
San Antonio
I-
NO
San Diego
1
City/Result
NO
New
Stadium
Presence
Statisticall:
Significant
NO
1
NO
YES
NEGATIVE
Sacramento
I
NO
I
NO
NO
YES
NEGATIVE
N.A.
4LLd
.
b
e
d
I
NO
I
YES
NEGATIVI
NO
“Statistically significant” meaosthat the probability is five percent or less dnt the correlation is the
result of random events.
The Charlotte Hornets and their stadium cameon line together in 1988. Only the team is considered
in the model, since the model cannot separatethe economicimpact of the team and the stadium
when the two are introduced simultaneously.
It was not possible to match Portland data with the general array of teams and stadiums.
“ALL” includes all 48 cities in sampleexcept Portland, Des Moines, and Waco.
“N.A.” signifies that there was no change in the number of teams or new stadiums and arenas
throughout the sampleperiod.
The following 12 MSAs that did not host sportsteamswere included in the analysis: Anchorage, Boise,
Des Moines, Duluth, Roanoke, SanJose, SantaBarbara, Savannah,Springfield, IL, Springfield, MO,
Waco, and Wichita.
- 17-
TABLE II
The Impact of New Stadiums and Professional
Spurts Teams on Real, Trend-Adjusted, Per Capita Personal
Income Growth For the Eight Major Regions in the
Continental United States, 1958-1987
Far West
Great Lakes
*
I
1
NO
NO
YES
NEGATIVE
)
NO
Plains
Rocky
Mountaius
NO
I
No
NO
1 PO%‘“,,
111
“Statistically significant” meansthat the probability is five percent or less that the correlation is the
result of random events.
The MSAs in eachregion are given below.
Far West:
Great Lakes:
Mideast:
New England:
Plains:
Rocky Mountains:
Southeast:
Southwest:
Los Angeles, Portland, Sacramento,SanDiego, San Francisco/Oakland, SanJose,
SantaBarbara, Seattle.
Chicago, Cincinnati, Cleveland, Detroit, Green Bay, Indianapolis, Milwaukee,
Springfield, IL.
Baltimore, Buffalo, New York, Philadelphia, Pittsburgh, Washington, D.C.
Boston, Hartford.
Des Moines, Duluth, KansasCity, Minneapolis, St. Louis, Springfield, MO,
Wichita.
Boise, Denver, Phoenix, Salt Lake City.
Atlanta, Charlotte, Miami, New Orleans, Orlando, Roanoke, Savannah.Tampa
Bay.
Dallas, Houston, San Antonio, Waco.
18
note:
The regressionanalysis used to generateTable II suggestsseveral things worthy of
n
No region demonstrated a statistically significant relationship between
professional sports teams and real, trend-adjusted, per capita personal income
growth. It does not appearthat professionalteamsexport their servicesbeyond the
region to an extent that matters economically.
n
Four of the eight regions demonstrated no significant relationship between
sports stadiums and real, trend-adjusted, per capita personal income growth. Of
the four other regions, stadiumsappearto have a significant positive impact in two
(the Southwestand the Rocky Mountains), and a significant negative impact in two
(New England and the Far West).
n
While there are more instances of a significant relationship between regions and
stadiums than there were when teams or MSAs were tested, it should be noted
that the percentage of trend-adjusted regional changes in real, per capita
personal income growth explained by changesin the number of sports stadiums
or teams is uniformly small. Inspection of the adjustedcorrelation coefficients
calculatedin the regressionanalysis indicatesthat only a small percentageof the
variation in regional, real, trend-adjusted,per capitapersonal income growth is
explainedby changesin the sports environment(see ’ g * ” values in Appendix B,
Table B). In no case is more than five percent of the variation in growth
explained.
Discussion
of the Findings
The findings summarizedin Table I indicate that professionalsports teams generally
have no significant impact on a metropolitan economy. Only in the caseof Indianapolis did
teams exert a statistically significant, positive impact on the metro area economy. This
positive result may be attributable to a combination of things, including the relatively small
size of the Indianapolis economy, the Hoosier State’s tradition of sports involvement, and
the city’s commitment to defining itself as America’s sports capital, particularly for amateur
sports. The citizens of Indianapolis appearto have embracedthis sports theme. The Colts
and Pacersmay have a stronger statefollowing than is usual relative to national standards.
The findings summarizedin Table II indicate that professional sports teamsgenerally
haveno significant impact on a region’s economy. They also indicate that, to some extent,
the servicesof stadiumsand arenasare exportedor imported on a national basis. In
particular, the Rocky Mountains and Southwestregions are net exporters of stadium
services,while the Far West and New England regions are net importers of stadium and
arena services.
19
Nevertheless,after accountingfor
national trends, stadiumsand arenasdo not
explain much of the variation in a region’s
real, trend-adjusted,per capita personal
income growth. Becausenone of the
adjustedcorrelation coefficients exceeds
five percent (see “R * ” figures in Appendix
B, Table B), stategovernmentsare likely
to derive little benefit from developing
plans to exploit their stadium “advantage”or to ameliorate their stadium “disadvantage.”
State governments are likely to
derive little benefit from
developing plans to exploit thehstadium “advantage” or to
ameliorate their stadium
“disadvantage.”
Negative Effects of Sports Stadiums.Stadium investmentsdo appear, however, to
have some economic effects. In 63 percent of the MSA tests, the correlation coefficient
associatedwith stadiumswas negative(see “&” values in Appendix B, Table A, page 34).
This suggestsa tendencyfor stadiumsto push rates of economicgrowth below the average
defined by trend lines for cities nationally. In the Washington, D.C. and Oakland-San
Franciscometro areas, stadiumscontributednegatively to real, trend-adjusted,per capita
personal income growth. It is noteworthy that both Washington, D.C. and the Bay Area
have relatively high per capita incomes. In light of this, the negativeresults of stadium
investmentsmay be attributable to the fact that sports-relatedjobs pay below the
metropolitan norm.
These results support a conclusionregarding professional sports and economic
growth reachedby the author and Robert Dye in 1990 < 3 1> :
The impact of stadium constructionor renovationon the metropolitan area’s
shareof regional income is negativeand significant. This result is consistent
with the kind of economicactivity that stadiumsand professional sports
spawn. Professional sports and stadiumsdivert economic developmenttoward
labor-intensive, relatively unskilled (low-wage) activities. To the extent that
this developmentalpath diverges from less labor-intensive, more highly
skilled labor (high-wage) characteristicof other economieswithin the region,
it would be expectedthat the sports-mindedareawould experiencea falling
shareof regional income. < 32 >
The types of jobs induced by stadium activity are typically low-wage and seasonal:
ticket takers, ushers, vendors, restaurantand bar workers, guards, parking lot attendants,
and so on. City developmentsubsidiesthat concentrateon thesetypes of jobs could lead a
city to gain a comparativeadvantagein unskilled and seasonallabor. Long-term future
growth in sports communities will be concentratedin low-income jobs.
But regardlessof the type of job, wouldn’t the spendingstill be of some economic
consequenceto a city? After all, sports entertainmentinvolves spending, and it is axiomatic
that one person’s spendingis anotherperson’s income.
- 20 -
As explainedpreviously, spending
spurredby sports eventsmay represent
spendingthat would still occur elsewherein
the area. Attending a sporting event is but
one possibleuse of an individual’s leisure
time and money. It is possible that no
connectionbetweenprofessional sports and
per capita income growth emergedbecause
sports spendingsimply substitutesfor other
forms of leisure spending.
It is possible that no connection
between professional sports and
per capita income growth emerged
because sports spending simply
substitutes for other forms of
leisure spending.
Who Benefits from Sports Subsidies?If professional sports teams and stadiums
haveno significant economic impact, where does the money go? The data suggestthat
stadium subsidiesand other sports subsidiesbenefit not the community as a whole, but
rather team owners and professionalathletes.
As economistDean Bairn has
The data suggest that stadium
pointed out, teams typically receive
subsidies and other sports
subsidiesfor their fiied costs(most
subsidies benefit not the
commonly, the cost of the stadium), and
professional sports must competewith
community as a whole, but rather
other entertainmentbusinesses.As a
team owners and professional
consequence,“there is little likelihood that
athletes.
a significant amount of the subsidy will be
passedon by the team” to the fan in the
form of lower prices. < 33 > The failure of professional sports to demonstratea significant
impact on economicgrowth suggeststhat either team owners and athletesspendlittle of
their revenuesin the area, or that professionalsports doesnot generateeconomic growth as
substantialas alternativeuses of the funds investedin sports. In either event, the team, not
the area, becomesthe beneficiary of the subsidy.
Public Subsidies to Sports Reconsidered. This study fmds no support for the
notion that there is an economicrationale for public subsidiesto sports teams and stadium
and arena construction. Professional sports doesnot appearto createa flow of public funds
generatedby new economicgrowth. Far from generatingnew revenuesout of which other
public projects can be funded, sports “investments”appearto be an economically unsound
use of a community’s scarcefinancial resources.
This raises the issue of opporhmity cost. Public officials have sought to fnnd
professionalsports stadiumswith everything from salestaxes to seemingly innocuousmeans
such as lotteries and taxes on gambling receipts. But even lottery and gambling revenuesare
limited. Using thesefunds for stadiumsprecludestheir use in other projects that might
better stimulate the local economy.
-21-
Scarceresourcesmight better be targetedto industries that are more clearly engaged
in export salesor import substitntion. The attention of those who allocate economic
developmentresourcesshould be devotedto the types of jobs that are being createdin
alternativedevelopmentscenarios.Indeed, since public expenditureson sports fail to
correlate with changesin real, trend-adjusted,per capita personal income, other things
unique to cities, perhapsincluding other public expenditures,must accountfor positive
changesin real per capita income. These fmdmgs indicate economic developmentstrategists
in governmentwould do well to analyzejust what activities do accountfor economic
growth in their communities.
Alternatively, a sound development
strategymight simply leave these
“development” funds in the private
economy. Private spendinghas spin-off
benefits of its own, and there is reasonto
believe that private individuals and
businesses
often make better spending
:
decisionsin the private sector man public
officials do in the public sector. < 34> If so, governmentinvestmentin professional sports
would be consistentwith a negative economicreturn.
A sound development strategy
might simply leave these
“development” funds in the private
economy. Private spending has
spin-off benefits of its own.
~
Sports subsidiesdo, of course, provide
a benefit to the teams themselves.But care
this claim is beyond the scopeof this
study, it should be noted that a new stadium probably provides only a short-term solution to
a team’s attendanceproblem. As mentionedearlier, the novelty effect of new stadiums
appearsto erode after sevento elevenyears. < 36 > The author’s previous research
indicatesthat the team’s on-field competitivenessis critical in determining
attendance.< 37 > Cities should recognizethat building stadiumsfor mediocre teamswill
probably add little to fan support beyond the short run.
- 22 -
PART 5
Summary
and Concluding
Remarks
For many, professional sports eventsare the transcendentexperiencein
contemporarywestern culture. Sports is filled with myth-wonderful stories of heroic
performancesand courageunder pressure.But increasedgovernmentsubsidiesto sports
teams and stadiums appearto have spawnedan economicmyth that the presenceof sports
teams and new sports stadiumsand arenashas a significant effect on an area’s economic
growth. This study and others find no factual basis for this claim. The author’s examination
of economicdata related to sports and real income growth provides no supponfor the
conventionalargument that profession51sportsstadiumsand teams have a significant impact
on an area’s economic growth.
State and local governmentsnow spendhundredsof millions of dollars eachyear to
subsidizesports teams and stadiums. To the extent that an “economic investmentrationale”
has been cited to justify government subsidiesto sports teamsand stadiums, the evidence
compiled here has important public policy implications.
The findings are particularly clear in
The fmdmgs are particularly clear
suggestingthat public fundiig of
in suggestingthat public funding
professionalsports stadiums is not a sound
of professional sports stadiums is
civic economic investment. If the
opportunity cost is included in cost-benefit
not a sound civic economic
considerations,public investmentsin
investment.
stadiumsmay be more than just
insignificant; they may be negative. Spent
differently-either by public officials on different public setvicesor by private citizens who
are not taxed to subsidizesport-these large sums of money may contribute more to an
area’s economicgrowth.
For professional sports to contribute signiticantly to the local economy, it must
induce large net increasesin spending. The myth is that it does. The reality is that sports
very likely do not expand spending,but serveonly to realign it. The public should view
promises about the economicimpact of professional sports with a healthy skepticism.
###
ENDNOTES
1.
2.
BenjaminA. Okoer, “Subsidiesof Stadiumsand Arenas,” in Roger G. Nell, editor, Govemme~t
and the Sports Business(WashingtonD.C.: Brookings Institution, 1974). pages325-347.
Dean V. Bairn, “Spats Stadiumsas ‘Wise Investments’:An Evaluation,” Heorrland Policy Study
No. 32, (Chicago, IL: The HeartlandInstitute, November26, 1990).
3.
Ibid., page6. Of the fourteen stadiumsin Bairn’s analysis,there was only one that had a net
positive accumulatedvalue for the subsidizinggovernment.Los AngelesDodger Stadium-privately
built, owned, and operated-paid more in property taxesthan it receivedfrom a relatively small
initial governmentsubsidy ($4.7 million in 1958) for site grading and road construction(see
page26).
4.
JamesP. Quirk and Rodney D. FOR,Paydin (Princeton, NJ: PrincetonUniversity Press, 1992),
~pages164-166.
5.
Edward B. Shils, “Report to the PbiladelpbiaProfessionalSportsConsortium On Its Contribution to
the Economyof Philadelphia,” mimeograph,June 17, 1985.
6.
KeonetbR. Clark, “Threat to Move Yanks Angers N.Y.,” Chicago Tribune, July 18, 1993,
Section 1, page 17.
7.
Hal Lancaster, “Tale of Two Cities: why Football MesmerizesBaltimore and Indianapolis,” The
Wall Street Jou.ml, January21, 1986, page24.
8.
For a brief discussionof the assumptionsin Edward Sbils’s study, seeRobert A. Baade, “Is There
an Economic Rationalefor SubsidizingSportsStadiums?”Heartland Policy Shuiy No. 13 (Chicago,
IL: The Heartland Institute, February 23, 1987). page 13.
9.
Robert A. Bade, sopranote 8.
10. Ibid., page 15.
11. Robert A. Bade and Richard F. Dye, “Sports Stadiumsand Area Development:A Critical
Review,” Economic Development Quarter&, August 1988, pages265-275.
12. Robert A. Bade and Richard F. Dye, “The Impact of Stadiumsand ProfessionalSportson
Metropolitan Area Development,* Growth and Change, Spring 1990.
13. JamesP. Quirk and Rodney D. Fort, supranote 4, and Andrew Ziibalist, Baxball md BiNions,
(New York: BasicBooks,HarperCollins Publishers,1992).
14. JamesP. Quirk and Rodney D. Fort, sopranote 4, page 172.
15. Robert A. Baadeand Richard F. Dye, supra note 11.
16. It is conceivablethat somevery real psychologicalbenefitshave no economicimpact. Somesports
enthusiastshave asserted,for instance,that a city’s residentsderive psychologicalbenefits, such as
a senseof pride, from building stadiumsand hosting teams.This claim is part of the “deepspace”
- 24
of the stadiumdebate,sinceassigninga value to such benefitsis a highly subjectivetask.
17. See,for instance,John E. Peck, “An EconomicImpact Analysis of SoothBend’s ProposedClassA
BaseballStadium,” Bureau of Business and Economic Research, May 10, 1985.
18. This information is especiallybard to obtain becausea researcherwould needto determinewhat
would have happenedif therewere oo pro sportsdevelopment.There is no way to know for sure
what ~omume~~would havedone with their dollars in the absenceof pro sports.
19. For a good discussionof an impact study that makessuch unrealistic assumptions,seeJamesP.
Quirk, “The Quirk Study: A Close Look at the Two Proposals,”St. Louis Post Disprch,
January 18, 1987, pages5i-8i.
20. Robert A. Baade,supra note 8.
21. A 1988 study by the Center for Businessand EconomicResearchat the University of Alabama
found that “a significant ocgativerelationshiphas beenidentified and estimatedbetweentax burden
&d economicgrowth for most statesin the United States.”See“A National Study on State
EconomicGrowth, Tax Borden, and GovernmentExpenditures:Phase3 Final Report,” University
of Alabama,June 30, 1988. For a survey of the literature on the negativerelationship betweentaxes
and economicgrowth seeJosephL. Bastand John Beck, “Taxesand EconomicGrowth,” in Joseph
and Diane Bast, editors, Coming Ouf ofthe Ice, (Chicago, IL: The HeartlandI&tote, 1990),
pages15-32.
22. Tax exemptbonds for stadiumconstructionin the United Statesexpired on December31, 1992.
23. For instance,seeJohn E. Peck, supra note 17.
24. For technicalreasons.regressioncalculationscould not be madefor Portland. Portland’s income
growth figures were included, however, in the incomegrowth trends.
25. SeeU.S. Departmentof Commerce,Bureauof EconomicAnalysis, Local Area Personal Income,
(Washington,D.C.: U.S. GovernmentPrinting Office), Various Years. Other data sauces for the
model are discussedin Appendix A. The boundariesof the Metropolitan StatisticalAreas and other
standardlocal areasreportedon by the USDOC have changedover the years. Adjustmentswere
madeto the MSA data to make them uniform over time.
26. “First differences” for real, per capitapersonalincomesfor individual cities were used to more
completelycapturethe changein a city’s economicactivity inducedby a stadiumor team. Had an
averageof city real, per capitapersonalincomeover the sampletime period been deductedfrom the
observedlevel of city economicactivity in year t, the impact of a sportschangein the initial year of
that changewould have been diluted. Using an averagewould have reducedthe likelihood that a
sports-inducedchangewould havebeen found, since the economicimpact of a teamor stadiumin
its first year would have beenaveragedover the stadiumor teamhistory. Therefore, the model
representedby the equationin Appendix A maximizesthe likeliiood of statistically significant
coefficientsfor the stadiumand teamvariables.
27. Roger G. Noll, a StanfordUniversity economicsprofessor,has noted that “a typical stadium’s
economicimpact can be comparedto that of a good-sizeddepartmentstore.” SeeChris Kmul,
“Fields of Dreams,” Los Angeles Tim, July 11, 1993, pageDl.
28. Robert A. Baade, “What Cities Need to Know About Stadiumsand EconomicDevelopment,”
mimeograph,November 1993.
-25
and the
29. Roger G. Nell, “Attendanceand Price Setting,” in Roger G. Nell, editor. Gov@mmenr
SportsBwiness (Washington,D.C.: The BrooldngsInstitution, 1974), pages115157.
30. It may be that private-sectorinvestmentsin sportsare more likely to bring real economicgrowth
than public-sectorsportsinvestments.EconomistDean Bairn has found that the per-seatcost for
public stadiumsis about 50 percenthigher than for private stadiums(seeDean V. Bairn, “City’s
taxpayershave Tiger by the tail,” TheDeiroif News,December23, 1990, page 3B). This finding
suggestsa higher level of economicefficiency when private dollars arc used for stadium
construction. The model employedin this paper. by combining private investmentwith @resumably
lessefficient) public investment,increasesthe likelihood that sportswill look like a good civic
investment.
If, on the other hand, public subsidiesto sportsare more efficient than private sports investments,
the fact that there have beenfew private stadiumsand private subsidiesto sports franchisesin recent
yearsmeansthat theseweakerexampleswill have little effect on the better performanceof the
public invesbnent.In either instance,the model gives every opportunity for the “public-investmentin-sports” rationale to prove itself.
31. Robert A. Baadeand Richard F. Dye, supranote 12.
32. Ibid., page 12.
33. Dean Bairn,,supra note 2, page 12; seealso page 35, endnote11. The important factorsare the
nature of the subsidy and the natureof the market demand.Most teamsreceive subsidiesfor fixed
costsassociatedwith building a stadium.~Tbissubsidy will not affect the variable cost of admitting
fans to the game,and the marketprice that maximizesthe profit realizedby the sale of ticketswill
be unchanged.The net effect of the subsidywill be to increasethe franchise’sprofits, since the
teamwill not have to pay the fixed coststhat were subsidized.
Even in the casewhere the teamreceivesa subsidythat affectsthe variable cost of admitting
additional fans to the stadiummuch of the subsidyis likely to be kept by the team, rather than
passedon in the form of lower ticket prices. This result is due to the “elasticity” of the demandfor
sport.3entertainment.
“Inelastic” demandusually occurswhen a fum has few competitorsfor the provision of a particular
service. At first blush, it might seemthat demandfor sportsentertainmentwould be inelastic, since
a professionalball club would have few-if any-competitors for the provision of their particular
type of sports entertainment.Few areashave more than one baseballteamor more than one football
team.
Nevertheless,it is likely that professionalsportsteamscompetewith professionalteamsin other
sports,and with a wide range of entertainmentservices-movie theaters,restaurants,night clubs,
and so forth. This competitionmeansthat the demandfor sportsservicesis more “elastic”-i.e., in
percentageterms, that the level of quantity demandedchangesmore dramaticallyin responseto
small changesin price. When a firm receivesa subsidyfor variable costsin a market with elastic
demand,it can lower its prices less than the subsidyand increasethe quantity of tickets demanded
sufficiently to once again maximize profits. Charging an even lower price-perhaps “passingon” the
completesubsidy to the fan-will causeprofits to fall, even though the quantity of tickets demanded
increases.
Thus, even when there is a subsidy for variable costs,not all of the subsidywill be passedon to the
fan. In a highly competitivemarket, little of the variable subsidywill be passedon to the fan.
- 26 -
34. For stadiumconstructionin particular, Dean Bairn has found that the per-seatcost for privately
financed stadiumsis less than that of publicly funded stadiums(seeDean Bairn, supra note 30). This
suggeststhat spendingby private interestsin a competitivestadiummarket are mademore wisely.
For a discussionof the problemswith public spendingversusprivate spading, seeJosephL. Bast,
“SelectiveTax Abatementsand Subsidies,”in Josephand Diane Bast, editors, Coming Out of the
Ice, supra note 21, pages3348.
35. Yankeesowner GeorgeSteinbremer, for instance,has suggestedthat a new (subsidized)stadium
would increaseattendanceat Yankee’sgames.SeeKennethClark, supra note 6.
36. Roger G. Noll, supra note 29
37. Robert A. Baade, “An Analysis of Major LeagueBaseballAttendance,1969.1987,” Journal of
Sport md Social Issues, 1990, 14(l), pages14-32.
27
APPENDIX
Discussion
A
of the Methodology
The Equation
This study employs a model basedon the following equation:
Cv,,,
@j,,/k)
cV,,+,
5Yj,t-l’k)
j=1
= pO + ‘lNTLt
+ Bfls,, , + e,
where
yi,, is real, per capita, personal income in MSA i at time 1;
yj,, is real, per capita, personal income in MSA j at time r;
k is the number of Metropolitan Statistical Areas (MSAs) in the sample;
I’IT~,~is the number of professional, major leaguesports franchises(baseball, football,
basketball, and hockey) in city i at time r;
NS,, is the number of stadiumsand arenasten years old or less in MSA i at time t;
e, is the “stochasticerror” at time t;
0, and 0, are coefficients to be determinedfor eachMSA;
and &, is a constantto be determinedfor eachMSA.
The regressionanalysis in this paper estimatesfl, and & for 36 Metropolitan
Statistical Areas (MSAs) that hosted professionalsports during the thirty-year period from
1958 through 1987. These 36 MSAs comprise all of the U.S. metropolitan areasthat hosted
either a professional team in one of the four major team sports-baseball, football,
basketball, or hockey-or a new professional sports stadium or arenaduring this period.
For technical reasons,regressioncalculationscould not be made for Portland, which hosted
sports during the time tested. Portland’s income growth figures were included in the
national income trends.
-
28-
To ensurea broader, more representativesample, cities that have nof hosted a
professional team, ostensibly becauseof their size, also were included in the study. A total
of 48 Metropolitan Statistical Areas were included in the analysis (a list of both setsof
MSAs appearsin Appendix B, page 32). The boundariesof the Metropolitan Statistical
Areas and other standardlocal areasreported on by the USDOC have changedover the
years. Adjustmentswere made to the MSA data to make them uniform over time.
This empirical investigation allows us to estimatethe extent to which changesin each
metro area’s per capita real income from one year to the next, adjustedfor general trends in
the 48 cities’ economic growth, are explainedby changesin either the number of new
stadiumsor professional sports franchisesthe area acquires. In keeping with standard
approachesto urban economics,the model considerseconomicgrowth for the entire MSA,
rather than for the host city alone. Using the MSA increasesthe likelihood of capturing any
positive economic impact that results from indirect expendituresthat occur outside the
immediate neighborhoodin which the stadium is located.
Technical
Discussion
of the Model
Part 4 of this study discussesvarious aspectsof the model, but for those familiar
with the technicalities of modeling, a few questionsmight arise. First, it might appearthat
local and national trends are not properly included in the analysis. Shouldn’t local and
national income trends be explanatory variables, and therefore on the right-hand side of the
equation with coefficients 0, and fi,? If local or national trends (or both) spurred an MSA’s
economicgrowth when a stadium was introducedor a team was adopted, would the model
be able to separatetheseeffects?
For severalreasons,national and local income variables should not appearas
independentvariables in the model. The most important reasonis that the appearanceof
thesetwo variables on the right-hand side of the equationleads to the problem of
multicollinearity, since it is likely that local and national income variables are highly
correlated. Moreover, using national and local income variables as independentvariables
results in another problem. Consider the following equation:
Y;t
=
PO
+
PINTi,,
+
&
t;
7.,/k
j-1
’
The last term in this equation is the averageof all the cities in the sample, including
the ith city. Thus, the dependentvariable appearson the right-hand side of the equation. In
formulating the model for this paper, the restriction of & = 1 is imposed on the model.
- 29
This approachalso enablesus to give an economicmeaning to the dependent
variable, since it can be interpreted as the deviation of Y from trends. By designing the
independentvariable such that it measuresdeviationsfrom local and national trends, the
model is safeguardedagainstmis-specification. Note that, as the question above suggests,
the model does not separatethe local and national impacts. The model is not designedto
separatetheseimpacts Instead, it factors them out.
“Fist differences” for real, per capita personal incomesfor individual cities more
completely capture the changein a city’s economicactivity induced by a stadium or team.
If, for instance,an average of city real, per capita personal income over the sample time
period had been deductedfrom the observedlevel of city economic activity in year t, the
impact of a sports changein the initial year of that changewould have been diluted. Using
an averagewould have reducedthe likelihood that a sports-inducedchangewould have been
found, since the economic impact of a team or stadium in its first year would have been
averagedover the stadium or team history.
The way in which trends are adjustedfor in the model helps give teams and stadiums
the best possible opportunity to demonstrateany positive impact they might have on
economicgrowth. Amusementand recreation servicesin a large city typically accountfor
less than 2 percent of overall economicactivity. Formulating the dependentvariable as a
deviation from a trend maximizes the likelihood of identifying a significant relationship
betweenthe stadium and team variables and the level of economic activity. Basing the
model on the above equation maximizes the likelihood of statistically significant coefficients
for the stadium and team variables.
Those who do not follow professionalsports might also wonder why teamsand new
stadiumswere representedby different explanatoryvariables, since it would appearthat
eachhost city that has a team would need a stadium. The number of stadiumsand teams
can vary independently.In some cases,new stadiumswere built in an MSA to
accommodateteams occupying an old stadium in the area.
The null hypothesisfor the model is that /3r # 0 and that & # 0, meaning that pro
sports teams and new stadiumshave a statistically significant economic impact. Multiple
regressionanalysis was used, and the Cochrane-Orcuttprocedurewas employed to correct
for autocorrelation.
Data Sources
Data for the model were taken from several sources.Information on the location of
teams and new stadiumswas taken from “Teams and Ball Parks” in The 1992 Sporfs
Ahamc (Boston: Houghton Mifflin Co., 1992), edited by Mike Meserole. Data on MSA
population and nominal, per capita personal income were taken from various years of Local
- 30 -
Area PersonalIncome (Washington, D.C.: U.S. GovernmentPrinting Office), compiled by
the Bureau of Economic Analysis, U.S. Departmentof Commerce. National Consumer
Price Index figures for adjusting nominal income trends were taken from the Economic
Repoti of the President, 1992 (Washington, D.C.: U.S. GovernmentPrinting Office, June
1993).
-31-
Detailed Results of Regression Analysis for
Metropolitan
Statistical Areas and Regions
Detailed results from the regressionanalysisfor the 48 Metropolitan Statistical Areas
(MSAs) included in this study appearin Table A below. A “Key To Understandingthe
Results” is provided for the interestedreaderunfamiliar with econometrics.
The following is an alphabeticallist of the 36 MSAs that hostedprofessional sports
from 1958 to 1987: Atlanta, Baltimore, Boston, Buffalo, Charlotte, Chicago, Cincinnati,
Cleveland, Dallas, Denver, Detroit, Green Bay, Hartford, Houston, Indianapolis, Kansas
City, Los Angeles, Miami, Milwaukee, Minneapolis, New Orleans, New York, Orlando,
Philadelphia, Phoenix, Pittsburgh, Portland, Sacramento,Saint Louis, Salt Lake City, San
Antonio; San Diego, San Francisco/Oakland,Seattle, Tampa Bay, and Washington, D.C.
These 36 MSAs all appearin Table A below (and in Part 4, Table I, page 16).
The following is an alphabeticallist of the 12 MSAs that did not host pro sports
teams: Anchorage, Boise, Des Moines, Duluth, Roanoke, San Jose, SantaBarbara,
Savannah,Springfield, IL, Springfield, MO, Waco, and Wichita. Per capita income figures
for thesemetro areaswere incorporatedinto the analysisof real, per capita personal income
trends.
The MSAs included in eachregion analyzedin the study are listed below.
Far West:
Great Lakes:
Midwest:
New England:
Plains:
Rocky Mountains:
Southeast:
Southwest:
Los Angeles, Portland, Sacramento,San Diego, San
Francisco/Oakland,San Jose, SantaBarbara, Seattle.
Chicago, Cincinnati, Cleveland, Detroit, Green Bay,
Indianapolis, Milwaukee, Springfield, IL.
Baltimore, Buffalo, New York, Philadelphia, Pittsburgh,
Washington, D.C.
Boston, Hartford.
Des Moines, Duluth, KansasCity, Minneapolis, St. Louis,
Springfield, MO, Wichita.
Boise, Denver, Phoenix, Salt Lake City.
Atlanta, Charlotte, Miami, New Orleans, Orlando,
Roanoke, Savannah,Tampa Bay.
Dallas, Houston, San Antonio, Waco.
Detailed results of the regressionanalysisfor eachregion appearin Table B. Part 4,
Table II, page 18, summarizesthe results from Table B.
-
32 -
KEY TO UNDERSTANDING THE RESULTS
Table A presentsthe results of the regressionanalysis for eachMetropolitan
Statistical Area (h4SA) and for the entire group of MSAs (listed under “ALL”). Table B
presentsthe results of the regressionanalysis for major U.S. regions. The comments
below concerningMSAs apply equally to the regions in Table B.
The first variable listed, &, measureswhat the MSA’s real, trend-adjusted,per
capita personalincome growth would be if there were no new stadiumsor teams. This
variable is a necessarypart of the equationfrom a mathematicalstandpoint, but doesnot
tell us about the economic impact of pro sport.
Of more interest are the coefficients 0, and &. Thesevariables measurethe
relationship betweenprofessional sports teamsand new stadiumsand an MSA’s real,
per capita personal income growth relative to the other cities in the study and relative to
its own pattern of growth. The sign of the coefficient indicateswhether pro sports are
related to positive or negative (though not necessarilysignificant) changesin
trend-adjustedgrowth. When the “t-statistic”-the number given in parenthesesbelow the
coefficient-is greater than 1.9 or less than -1.9, the observercan be 95 percent
confident that the relationship betweenpro sports and growth is “statistically significant”
(i.e., a relationship stronger than what could be expectedto occur by chance).
When p, is positive (negative)and the t-statistic is significant, it suggeststhat
professional sports teamshave a statistically significant positive (negative)effect on the
MSA’s real, trend-adjusted,per capita personalincome growth. The variable flz and the
correspondingt-statistic are interpreted in the sameway, exceptthat they measurethe
effect and significance of new professionalsports stadiumsand arenas.
Numbers in the adjustedcorrelation coefficient column, E * , identify the
percentageof variation in real, trend-adjusted,per capita personal income growth
explainedby changesin the sports environment.
The Durbin-Watson measurementis of interest to economists.The Durbim-Watson
figures in Table A demonstratethat relationshipsbetweenthe variables that might bias
the fmdings have indeed been accountedfor in the analysis. ln an efficient model, the
Durbin-Watson figure is close to 2.
ln some instances, “N.A.” is enteredin place of a variable value. This occurs
wheneverthere was no changein the number of an MSA’s teams or new stadiumsduring
the test period (1958 to 1987). Computing the relevant variable in thesecasesis not
possible.
33 -
TABLE A
The Impact of New Stadiums and Professional
Sports Teams on Real, Trend-Adjusted, Per Capita Personal
Income Growth For SelectedU.S. Cities (MSAs), 1958-1987:
Regression Analysis Results
City/
Statistic
Bo
(t-statistic)
81
(t-statistic)
I%
(t-statistic)
-2
R
DurbmWatson”
TABLE A
(Continuedl
City/
Statistic
&
(t-statistic)
SI
(t-statistic)
t%
-2
R
(t-statistic)
DurhiiWatson’
Hartford
-42.16
(-.27)
247.24
(1.46)
-95.86
(-.70)
.02
1.67
Houston
292.3
(.33)
-148.2
(-.52)
140.1
(.49)
.22
1.73
(.71)
(-.54)
1 (-1.32)
-176.36
(-1.16)
99.88
(.93)
-84.28
(-.67)
-.04
1.82
-846.31
(-1.41)
93.90
(1.21)
43.39
(.83)
.04
1.99
29.41
-123.22
(-.44)
N.A.
.14
1.86
IIINew Orleans
New York
Orlando
t.28)
Philadelphia
12.48
C.01)
6.79
(.02)
-17.94
(-. 18)
.09
2.00
PlJOeJliX
35.4
(.135)
-85.83
(-.42)
148.03
(.73)
.lO
1.77
Pittsburgh
-188.87
(-.45)
39.37
62.58
(1.23)
-.02
1.91
C.28)
- 35 -
III
TABLE A
(Continued)
City/
Statistic
Seattle
Tampa Bay
Washington,
D.C.
ALL
’
’
d
c
Bo
A
8,
-a
R
DurbinWatson’
(t-statistic)
(t-statistic)
(t-statistic)
497.81’
(1.99)
-132.24
(-1.34)
-227.84
(-1.61)
.33
1.64
28.03
-57.27
(-.34)
.07
1.85
(. 16)
48.54
(.27)
479.26’
(1.93)
-76.79
(-.93)
-204.44’
(-2.11)
.08
1.98
4.17
(.31)
4.24
(.69)
-22.28
(-1.36)
.oil
N.A.
After Cocbrane-Orcutt adjustment.
The Charlotte Hornets and their stadium came on line together in 1988. Only the team is consideredin
the model, since the model cannot separatethe economic impact of the team and the stadium when the
two are introduced simultaneously.
The probability is five percent or less that the correlation is the result of random events.
It was not possible to match Portland data with the general array of teamsand stadiums.
“ALL” includes all 48 cities in sample except Portland, Des Moines, and Waco.
36
The “t-statistics” appear in the parentheses.
“N.A.” signifies that there was no changein the number of teamsor new stadiumsand arenas
throughout the sample period.
- 31-
TABLE B
The Impact of New Stadiums and Professional
Sports Teams on Real, Trend-Adjusted, Per Capita Personal
Income Growth For the Eight Major Regions in the
Continental United States, 1958-1987:
Regression Analysis Results
III
Region/
Statistic
I
B”
(t-statistic)
‘*
B,
(t-statistic)
I
I
82
(t-statistic)
I
-2
R
I
-4.23
(-.37)
-67.54
(-2.37)
.04
Great Lakes
-39.39
(-1.28)
3.86
(.35)
-31.57
C-.93)
-.Ol
Mideast
-1.79
t-. 14)
5.38
c.901
-20.56
C-1.28)
.oo
128.59
(1.76)
18.83
.04
C.81)
-141.75
(-1.82)
Plains
-48.44’
(-2.26)
17.63
(1.18)
-39.23
(-1.04)
.oo
Rocky
Mountains
-45.98
(-1.50)
-14.86
(-.50)
171.83”
(2.07)
.04
Southeast
-3.61
(-. 14)
8.08
(.34)
-31.66
(-.68)
-.Ol
Southwest
-55.33
(-1.29)
-32.73
(-1.22)
168.39”
(2.88)
.05
I
New England
’
I
The probability is five percent or less that the correlation is the result of random events
The “t-statistics” appear in parentheses.
The MSAs in each region are given below.
Far West:
Great Lakes:
Mideast:
New England:
Los Angeles, Portland, Sacramento,San Diego, San Francisco/Oakland.SanJose,
SantaBarbara. Seattle.
Chicago, Cincinnati, Cleveland, Detroit, Green Bay, Indianapolis, Milwaukee.
Springfield. IL.
Baltimore, Buffalo, New York. Philadelphia, Pittsburgh, Washington, D.C.
Boston. Hartford.
- 38
Plains:
Des Moines, Duluth. KansasCity, Mimeapolis. St. Louis. Springfield. MO,
Wichita.
Rocky Mountains: Boise, Denver, Phoenix, Salt Lake City.
Southeast:
Atlanta, Charlotte, Miami, New Orleans. Orlando, Roanoke, Savannah.Tampa
Bay.
Southwest:
Dallas, Houston. San Antonio. Waco.
Robert A. Baade is Vail Professorof Economicsar Lake Forest College in Illinois. He is a former
Fellow at The London School of Economics and was the recipient of a 1992 Fellowship at the University
of Chicago. His work on sports economicshas been published in such academicjournals as the Sewn
Hall Joumal of Sportsand Law, the Journal of Sport and Social Issues. and Growth and Change. and he
has written for popular publications like C&4 Today Eawball Weekly,Chicago Enterprise, and News&y.
In 1985 and 1986 Baadeserved as chairman of the Chicago Metropolitan Planning Council’s Committee
to Determine Costs and Benefits of a New Sports Stadium in Chicago. He is a former Wisconsin State
University Conference “Outstanding Scholar-Athlete.”
The extensive statistical analysis in this paper required gathering. adjusting. and entering a vast amountof
data. The author would like to thank his resourceful and tireless senior researchassistant,Sheila Somers,
and his colleagues, ProfessorsJeffrey Sundberg.Richard Dye, and Nader Nazmi for their help with this
enormous task. This paper could not have been completedwitbout their assistance.Of course. all errors
and omissions in this paper are solely the author’s own.
l
l
l
Publishedby The Heartland Institute as one of a series of in-depth studies on important state and local
public policy issues. Nothing in a Heartland Pohcy Stu& should be construedas necessarilyreflecting tbr
views of The Heartland Institute or as an attempt to aid or hinder the passageof any legislation.
ABOUT THE HEARTLAND INSTITUTE
PUrpCSe:
The Heartland Institute is an independent,nonprofit. researchandeducationorganization that
applies cutting-edgeresearch to state and local public policy issues.
History:
The Heartland Institute was incorporated in Chicago in April 1984.Since then, it has released
60 studies of public policy issues, over 200 opinion essays, five books, and one video.
Heartland has sponsored scores of conferencesand otber events on public policy issues
throughout the Midwest. A Detroit office of the Institute opened in May 1990 when a
Michigan Board of Governors and Board of Policy Advisors was established.
Program:
The Heartland Institute solicits manuscripts from academics and public policy experts.
circulates those manuscripts amonga Board of Policy Advisors for review and revision, and
distributes the resulting publications to local and national media, legislators, academics,and
opinion leaders.Heartlmd Policy Studiesand Heartkmd Perspectivesare distributed to media
outlets and state legislators across the country. Heartland also publishes /tttellectual
Ammttition, a bimonthly public policy magazine.
Funding:
The Institute receives support from individuals. businesses,and foundations. It acceptsno
grants from governmentagenciesand does not conductcontract researchfor private interests.
Contributions to Heartland are tax-deductibleunder Section501(c)(3) of the Internal Revenue
Code.
offtce:
For additional information, write to The Heartland Institute/615 Griswold, Suite
15091Detroit. Michigan 48226: or call 313/961-1950. Facsimile requests can be dialed to
313/961-1420.