Content uploaded by Victor A. Matheson
Author content
All content in this area was uploaded by Victor A. Matheson
Content may be subject to copyright.
Rejecting ‘‘Conventional’’ Wisdom: Estimating the
Economic Impact of National Political Conventions
Robert A. Baade
a
, Robert Baumann
b
and Victor A. Matheson
c
a
Department of Economics and Business, Lake Forest College, Lake Forest, IL 60045 USA.
E-mail: baade@lfc.edu
b
Department of Economics, College of the Holy Cross, Box 192A, Worcester, MA 01610-2395 USA.
E-mail: rbaumann@holycross.edu
c
Department of Economics, College of the Holy Cross, Box 157A, Worcester, MA 01610-2395 USA.
E-mail: vmatheso@holycross.edu
This paper provides an empirical examination of the economic impact of the Democratic
and Republican National Conventions on local economies. Our analysis from 1970–2005
of the 50 largest metropolitan areas in the country, including all cities that have hosted
one of the national conventions during this time period, finds that the presence of the
Republican or the Democratic National Convention has no discernable impact on
employment, personal income, or personal income per capita in the cities where the
events were held confirming the results of other ex post analyses of mega-events.
Eastern Economic Journal (2009) 35, 520–530. doi:10.1057/eej.2009.25
Keywords: conventions; impact analysis; mega-event
JEL: O18; R53
INTRODUCTION
Convention tourism is big business in the United States. According to the Convention
Industry Council, in 2004 the meetings, conventions, exhibitions, and incentive
travel industry generated over $122.3 billion in direct spending and 1.7 million jobs.
These figures are ‘‘more than the pharmaceutical and medicine manufacturing
industry and only slightly less than the nursing and residential care facilities
industry’’ [CIC, 2005]. In hopes of gaining a piece of this lucrative business, cities
compete vigorously to host meetings and conventions, and billions of dollars of
taxpayer money has been directed toward the construction of ever larger and more
elaborate convention centers in cities all across the country.
Perhaps the most sought-after jewels of the convention industry nationwide are
the quadrennial National Democratic and Republican Conventions at which each
party’s presidential candidate is nominated. City and party officials suggest that
these events generate significant economic windfalls for host cities and also serve
to focus national and even international attention on the host city. For example,
city officials of New York City and Boston claimed net economic impacts
of $255 million and $156 million, respectively, for the 2004 Republican and
Democratic National Conventions. These economic impact numbers figured
prominently in press releases promoting the 2008 Republican Convention in
St. Paul/Minneapolis.
The rosy economic impact numbers touted by convention promoters (both for
political conventions as well as other prominent events) are also used to justify hefty
public subsidies for the construction and operation of municipal convention centers.
Eastern Economic Journal, 2009,35, (520–530)
r2009 EEA 0094-5056/09
www.palgrave-journals.com/eej/
Over the past decade, tens of billions of dollars, including significant public funds,
have been spent on new or refurbished convention centers in cities, for example,
the Boston Convention Center ($800 million), D.C.’s Washington Convention
Center ($850 million), and Omaha’s Qwest Center ($291 million) [Malanga 2004].
The question of whether this has been money well spent is one of major
public interest.
Economists tend to be skeptical of the large economic impact numbers touted by
convention facilities and event organizers. Our examination of 18 national political
conventions from 1972 to 2004 fails to support the promoters’ optimistic economic
projections and finds that these events have a statistically insignificant impact on
local economies.
BACKGROUND
Many researchers [e.g. Baade et al. 2008b] typically separate economic impact
analyses into two main categories: ex ante studies and ex post studies. Ex ante
studies estimate the economic effect of an event by predicting the number of visitor
days as a result of an event as well as an average daily expenditure. A multiplier is
often applied to any direct economic impact figures resulting in a total impact
number that is typically about twice as large as the direct economic impact. As noted
previously, ex ante studies of national political conventions routinely ascribe large
benefits to these major events.
Critics of ex ante economic analysis, however, point out that these studies often
suffer from three major shortcomings that lead to an overestimation of the total net
impact of these events. First, ex ante reports often fail to account for the substitution
effect that occurs when local residents spend their money on convention-related
activities rather than on other goods and services in the local economy. As the
Democratic and Republican national conventions primarily draw delegates from
across the country rather than from local areas, the substitution effect in these cases
is likely to be relatively small compared with, for example, a county or state political
convention.
The second concern in ex ante studies is the crowding out effect. The large crowds
and congestion associated with ‘‘mega-events’’ like the national conventions may
deter people not associated with the convention from engaging in economic activities
in the host city. While hotels, bars, and restaurants, may do well during the
convention, other retailers and service providers may not benefit from the event and
potentially could lose sales. This issue is of particular concern during a national
political convention that necessitates a high degree of security and also may generate
large crowds of protesters, both of which dissuade casual shoppers and diners and
result in major disruptions for local residents. During the week of the 2004
Republican National Convention in New York City, for example, attendance at
Broadway shows fell more than 20 percent compared with the same week a year
earlier despite the presence of tens of thousands of visiting conventioneers and
journalists. Similarly, police in St. Paul, Minnesota during the 2008 Republican
National Convention had to resort to firing tear gas into protesters to break up
demonstrations, an action that clearly dissuades other visitors from spending time
and money in the city.
Many economists are also wary of the multipliers used to generate indirect
economic benefits in ex ante studies. Often the multipliers used are too high,
Robert A. Baade et al.
Economic Impact of Political Conventions
521
Eastern Economic Journal 2009 35
but even more conservative estimates of multipliers may be suspect. Inter-industry
relationships within regions based upon an economic area’s normal production
patterns are used to calculate multipliers. These inter-industry relationships
may not hold during mega-events, however. Therefore, any economic analyses
based upon these multipliers may be highly inaccurate, since there is no
reason to believe the usual economic multipliers apply during major events
[Matheson 2004].
In particular, national conventions may result in large windfalls to national
restaurant and hotel chains and provide employment opportunities for hospitality
workers and journalists from across the country, but may not result in significant
wage gains for local employees. In this situation, the economic gain from the event
does not accrue to the host city but rather benefits the bottom line back at corporate
headquarters. It is local taxpayers, however, who are often asked to foot the bill for
convention center expansions and who suffer from the disruptions associated with
the event.
Finally, convention promoters often suggest that prominent events such as the
Republican and Democratic National Conventions give cities immeasurable benefits
in terms of national and international exposure by being placed in an intense media
spotlight. While this contention may be true, it is possible that the publicity may not
portray the city in a positive light. In the realm of sporting events, the Summer
Olympic Games in 1972 in Munich and in 1996 in Atlanta were marred by terrorist
incidents, and Salt Lake City’s reputation suffered after the bribery scandal
surrounding its bid for the 2002 Winter Olympics. Host cities for political
conventions are similarly not immune from bad publicity. For example, the chaos
and protests surrounding the 1968 Democratic Convention in Chicago are still
noteworthy even 40 years later. It is hard to imagine that the city of Chicago
benefited from its ill-fated moment in the sun.
Due to the difficulties associated with ex ante estimation, numerous scholars
estimate the effects of mega events on local economies by ex post estimation, which
examines the actual economic performance of local areas that host large events.
While few ex post studies of conventions are found in the existing literature, many
authors have examined major sporting events such as the Olympics [Baade and
Matheson 2002; Jasmand and Maennig 2007] or World Cup [Baade and Matheson
2004; Hagn and Maennig 2007a, b], the Super Bowl [Porter 1999; Baade and
Matheson 2006; Coates 2006], All-Star Games [Baade and Matheson 2001; Coates
2006], and post season play in general [Coates and Humphreys 2002; Coates and
Depken 2006; Baade et al. 2008a, b]. The overwhelming majority of ex ante studies
of mega-sporting events finds little to no significant positive economic impact from
hosting these events. If the Republican and Democratic National Conventions are
truly the ‘‘Super Bowl’’ of the convention business, then based on the evidence of the
actual economic impact of the Super Bowl, cities hosting national political
conventions have every reason to be concerned about the real magnitude of the
economic windfall they can expect.
Coates and Depken’s [2006] research is of particular interest to our study. The
authors use taxable sales data from individual cities in Texas to measure the
economic gains from hosting a variety of sporting events including the Super Bowl
and the World Series. Houston also hosted the 1992 Republican National
Convention, and Coates and Depken include a control variable for this event.
They find that ‘‘the political convention reduced taxable sales by $19 million and
reduced sales tax revenues by approximately $1.4 million.’’
Robert A. Baade et al.
Economic Impact of Political Conventions
522
Eastern Economic Journal 2009 35
THE MODEL
Two types of data have been used most frequently in the existing ex post studies for
professional sports. Coates and Humphreys [1999; 2002; 2003], Baade and
Matheson [2001; 2004; 2006], Hagn and Maennig [2007b], Jasmand and Maennig
[2007], and Baade et al. [2008b] use annual data on employment, personal income, or
per capita personal income over a large number of cities and years to estimate the
economic impact of sporting events. The use of annual data is clearly not ideal when
examining events such as a political convention with a relatively small duration.
Realizing this fact, other studies such as Porter [1999], Baade and Matheson [2001],
Coates [2006], Coates and Depken [2006], and Baade et al. [2008a] have used higher
frequency data such as taxable sales that are available at a monthly or quarterly
basis. Many of these data sources, such as taxable sales, cannot be used in
nationwide panels of political conventions because of cross-state differences in data
availability and taxation laws leaving researchers with two options: examining any
political conventions that have taken place in a single state using high frequency
data or examining a large panel of conventions using annual data. This paper uses
the panel approach to look at multiple conventions over the period 1972–2004.
As noted by Baade et al. [2008a], there are several approaches to estimate the
impact of an event on a city. Mills and McDonald [1992] provide an extensive
summary of these models, which seek to identify changes in economic activity
through changes in key economic variables in the short run or the identification of
long-term developments that enhance the capacity for growth. Our task is not to
explain metropolitan economic growth but rather to use past work to help identify
any effects of political conventions on economic indicators. To this end we have
selected explanatory variables from existing models to predict economic activity in
the absence of the convention. Estimating the economic impact of a convention
involves accounting for normal activity and determining whether the presence of an
event of such national prominence increases economic activity. Thus, this approach
depends on our ability to identify variables that account for the variation in growth
in economic activity in host cities.
Our model estimates the changes in the growth rates of real personal income,
employment, and real per capita personal income attributable to political
conventions in host cities between 1969 and 2005. We use a sample of 50
metropolitan statistical areas (MSAs) that had at least one million residents in 2005.
This sample includes all 14 of the MSAs that have hosted a national political
convention since 1969 (see Table 1) and a control group of MSAs that have not
hosted such an event. Most of the host cities are relatively large compared to the rest
Table 1 Political convention hosts
Democratic national convention Republican national convention
1972 Miami, Convention Center Miami, Convention Center
1976 Madison Square Garden, New York City Kemper Arena, Kansas City
1980 Madison Square Garden, New York City Joe Louis Arena, Detroit
1984 Moscone Center, San Francisco Reunion Arena, Dallas
1988 The Omni, Atlanta Superdome, New Orleans
1992 Madison Square Garden, New York City Astrodome, Houston
1996 United Center, Chicago San Diego Convention Center
2000 Staples Center, Los Angeles First Union Center, Philadelphia
2004 FleetCenter, Boston Madison Square Garden, New York City
Robert A. Baade et al.
Economic Impact of Political Conventions
523
Eastern Economic Journal 2009 35
of the sample. The smallest host MSA is Kansas City, which had a population of
just under two million in 2005. For this reason and due to the existence of unit roots
in the underlying data, we use growth rates to compare cities of different sizes.
Table 2 presents the summary statistics of real personal income, employment, real
per capita personal income, and population for the entire sample as well as for the
subsets of host and non-host cities.
Following closely the outline provided by Baade et al. [2008b], our baseline model
for the estimations is:
Yit ¼b0þb1POPit þb2OTHERit þb3CONit þgtþaiþeit
ð1Þ
There are three different dependent variables (Y
it
): the growth rates of real personal
income, employment, and real per capita personal income in year tand MSA i.To
account for the panel nature of our data, we include controls for each year (g
t
) and
MSA (a
i
). This specification allows MSAs to have different intercepts and also
purges national trends. The vector of city dummy variables (a
i
) allows for fixed
effect differences in growth rates across cities, and the year dummy variable (g
t
)
allows for fixed effect differences in growth rates across time, in effect accounting for
changes in growth rates due to variations in the national business cycle. In other
versions of this model, we also included controls for city-specific trends as well, but
this addition added little explanatory power and did not impact our main results.
POP
it
is the log population of city iin time tand is included to control for
differences in growth rates that can be accounted for simply by the size of the
metropolitan area.
OTHER
it
is a vector of dummy variables that represents important economic
events specific to an area that would not be captured in the national economic
business cycle or overall city growth rate. These events are identified by searching
for outliers in the data that can be clearly explained by obvious idiosyncratic
macroeconomic shocks. The clearest example of such a shock is the effect of
Hurricane Katrina on the New Orleans economy in 2005. Personal income in the
MSA fell by roughly one-third in 2005 resulting in a personal income growth rate
Table 2 Summary statistics (standard deviations in parentheses)
Variable All cities Host cities Other cities
Real personal income $83,025,696 $174,904,375 $47,295,099
($103,984,672) ($156,557,666) ($30,627,925)
Real personal income, growth rate 0.0306 0.0278 0.0317
(0.0308) (0.0342) (0.0294)
Employment 1,458,247 2,913,884 892,165
(1,588,805) (2,239,550) (48,843)
Employment, growth rate 0.0233 0.0200 0.0246
(0.0253) (0.0241) (0.0256)
Per capita real personal income $29,542 $31,542 $28,764
($5,903) ($6,075) ($5,648)
Per capita real personal income, growth rate 0.0159 0.0151 0.0162
(0.0262) (0.0306) (0.0243)
Population 2,612,915 5,282,991 1,574,553
(2,842,968) (4,181,230) (764,231)
No. of cities 50 14 36
No. of city-years 1,750 490 1,260
Robert A. Baade et al.
Economic Impact of Political Conventions
524
Eastern Economic Journal 2009 35
roughly 35 percentage points below that which would have been predicted absent
the hurricane and the resulting devastation of the city. By including a dummy
variable for the disaster, overall model fit is significantly improved. Of course, every
city faces multiple idiosyncratic shocks to its economy each year, so the decision
of whether to include a control for a particular event is, by its very nature, somewhat
ad hoc. Since the econometric procedure detailed below places strict limitations on
the total number of variables that can be included in the model, however, one must
be selective. The general guidelines we used to make decisions about which events to
include were essentially threefold. First, the effect of the shock on the particular
MSA needed to be large enough that a control variable produced a coefficient in the
model that was statistically significant at normal levels. Second, the shock needed to
be unrelated or clearly exaggerated compared to the general business cycle. Third,
the shock needed to be the result of a well-known and newsworthy event.
In the end, we control for the following seven extraordinary events: Hurricane
Katrina in New Orleans in 2005; Hurricane Andrew in Miami in 1992 and the city’s
subsequent recovery in 1993; the September 11 terrorism attack in New York City in
2001; the collapse of oil prices and its subsequent effects on real estate and financial
institutions in the oil patch cities of Dallas-Fort Worth, Denver, Houston, New
Orleans, and Oklahoma City from 1983 through 1987; the financial windfall in
Houston from the first oil crisis in 1974; and the high tech boom in San Jose in 1999
and 2000 and in San Francisco in 2000, as well as the bust in 2001 in both cities. See
Appendix 1 for the exact specification for each OTHER variable included.
Finally, CON
it
, the independent variable of interest for this paper, equals 1 if the
MSA hosted a political convention that year and 0 otherwise. Under alternative
specifications separate convention variables for the Republican and Democratic
conventions were analyzed, but the results were not appreciably different than those
for which a single convention variable was examined. It should be noted that since
the dependent variables are in terms of growth rates, it is reasonable to presume that
if political conventions cause an increase in economic activity in the year they
take place, growth rates in the following year may be below model expectations as
the economy converges back to its long-run trend. In alternative specifications, the
convention variable was set equal to 1 in the year of the convention, 1 in the year
after the convention, and 0 otherwise. The results are not qualitatively different
under this alternative specification, and they are therefore not reported here.
Several tests are used to ensure that the dependent variables do not exhibit a unit
root. First, we perform Dickey–Fuller and Phillips–Perron tests for each city and
each dependent variable. For all three dependent variables, 48 of the 50 cities pass
both tests at 5 percent. Of the other two cities, one passes both tests at 10 percent
(Washington, D.C.), and one fails both tests (New Orleans). We also perform unit
root tests on the entire panel using tests from Levin et al. [2002] and Im et al. [2003],
which allow for panel-specific attributes such as differing time trends and
autoregressive paths. Both tests identify unit roots in the raw data for all three
dependent variables but reject the existence of a unit root for percent changes of all
three dependent variables.
Given the time-series nature of the data, the error term in equation (1) is likely to
be autocorrelated. While ordinary least squares regressions will produce consistent
estimates, the standard errors will be incorrect. We use a test suggested by
Wooldridge [2002] for autocorrelation within each panel, which estimates
^
e
it
¼r^
e
i,t1
þu
it
. Under the null hypothesis of no autocorrelation, r¼0.5, and
all three dependent variables reject this null hypothesis.
Robert A. Baade et al.
Economic Impact of Political Conventions
525
Eastern Economic Journal 2009 35
One method to account for the autocorrelation is to include an autoregressive
component, which changes our estimation model to
Yit ¼b0þb1Yi;t1þb2POPit þb3OTHERit þb4CONit þgtþaiþeit
ð2Þ
Introducing a lagged dependent variable requires the Arellano and Bond [1991]
estimation technique, which is sometimes referred to as a ‘‘difference Generalized
Method of Moments (GMM)’’ model. This model is used by Baade et al. [2008b]
as is described in several works, including Bond [2002] and Roodman [2006]. This
model begins by differencing equation (2), which purges a
i
. Once the city-specific
effect is removed, the model uses higher-order lags of Y
it
to instrument for DY
i,t1
.
Any other independent variables that are believed to be endogenous or pre-
determined (i.e., variables independent to the current error but not previous errors)
can be handled in the same way.
Given T¼35 years of observations, there are 34 observations of the differenced
dependent variable (DY
it
) for each city. Given the first lag of the differenced
dependent variable is endogenous (DY
i,t1
), all of the remaining 32 higher-order lags
can be used as instruments for DY
it
. While the higher-order lags should create
missing values in practice, Holtz-Eakin et al. [1988] show that each instrument
produces a useful moment condition. In other words, consider the moment
condition E[Z
it
0De
it
]¼0, where Z
it
0contains the instruments (i.e., the higher-order
lags) and De
it
is the differenced error term. For the second-order lag instrument,
the moment condition is S
i
y
i,t2
De
it
¼0iftX3, for the third-order lag instrument,
the moment condition is S
i
y
i,t3
De
it
¼0iftX4, and so on.
Consistency of this approach requires that the error terms are independently
and identically distributed, which typically cannot be assumed in dynamic panel
models. For example, it is plausible that the variance of the error term (original
or differenced) may differ across cities. A weighting matrix Wasymptotically
corrects the moment condition W¼1
NSiðZ
*
iDe
*
iDe
*
iZ
*
iÞ, where Z
*
iand De
*
iare
city-specific (T2) vectors. Using this weighting matrix, GMM minimizes
1
NSiDe
*
iZ
*
i
W1 1
NSiZ
*
iDe
*
i
:
To obtain the weighting matrix, it is necessary to have consistent estimates of De
*
i,
which can be obtained using a different weighting matrix W1¼1
NSiðZ
*
iHZ
*
iÞ, where
His a (T2) square matrix with 2 on the diagonal, 1 on all of the immediate off-
diagonals, and 0 elsewhere. Thus, the first-step estimates the model using W
1
to
produce the estimates D^
e
it
, which the second step uses in the weighting matrix W.
While this correction produces the desirable asymptotic properties, several works
[Arellano and Bond 1991 and Blundell and Bond 1998, to name only two] suggest
the standard errors in the second step are downward biased. We use the Windmeijer
[2005] finite-sample correction to adjust the standard errors. Finally, one concern
with the Arellano and Bond [1991] technique is over-identifying restrictions,
especially given the relatively long time period for each city in our data. We use a
Hansen [1982] test to determine the number of over-identifying restrictions.
Table 3 presents the Arellano–Bond estimation results of equation (2) using each of
the three dependent variables. For brevity, we omit the estimates for the year dummies
although these are available upon request. As noted previously, the city specific dummy
variables are purged by differencing. The Arellano–Bond tests for autoregressive errors
suggest that autocorrelation exists in the first lag, which is expected and justifies the
inclusion of the first difference of each dependent variable. In addition, the same test
suggests that a second lag term is not necessary for any of the dependent variables.
Robert A. Baade et al.
Economic Impact of Political Conventions
526
Eastern Economic Journal 2009 35
We find only the weakest evidence that political conventions increase economic
activity above normal fluctuations. Controlling for other factors, personal income
grew 0.15 percent faster in cities during convention years than non-convention cities
and/or non-convention years. Personal income per capita grew 0.09 percent faster in
convention cities while employment growth in convention cities actually lagged
other cities by 0.05 percent. None of these values are close to statistical significance.
Of course, given the size of these large, diverse metropolitan areas, even small
increases in economic activity in percentage terms may result in a large increase in
activity in dollar terms.
With the average personal income in a host city being roughly $175 billion, point
estimates for personal income suggest that the presence of a national convention
increases local personal income by about $260 million, close to the estimates
promoted by the conventions’ backers. The estimates for the per capita personal
income model point to a much smaller increase of only $150 million from hosting a
convention, while the employment model produces an estimate of job losses of about
1,500 workers, translating into roughly a $90 million loss from hosting the
convention. In all cases the confidence intervals on the coefficients are large enough
such that an ex ante estimate of $150–$250 million benefit from hosting a national
Table 3 Arellano–Bond estimation results (standard errors in parentheses), all cities
Dependent variable Personal
income growth
Employment
growth
Personal income
per capita growth
Dependent variable
t1
0.3914*** 0.5050*** 0.3057***
(0.0621) (0.0620) (0.1001)
Population 2.45e8*** 2.83e8*** 3.98e8***
(6.49e9) (6.67e9) (9.38e9)
National Convention 0.0015 0.0005 0.0009
(0.0028) (0.0042) (0.0028)
Oil Boom 0.0341*** 0.0003 0.0357***
(0.0022) (0.0020) (0.0035)
Oil Bust 0.0358*** 0.0234* 0.0234***
(0.0084) (0.0099) (0.0065)
Hurricane Katrina 0.3690*** 0.0771*** 0.3677***
(0.0037) (0.0014) (0.0033)
Hurricane Andrew 0.0189*** 0.0105*** 0.0179***
(0.0012) (0.0011) (0.0013)
Tech Boom — San Jose 0.0457*** 0.0114*** 0.0459***
(0.0023) (0.0010) (0.0028)
Tech Boom — San Francisco 0.1382*** 0.0379*** 0.1297***
(0.0021) (0.0021) (0.0064)
9/11 0.0173*** 0.0035* 0.0156***
(0.0029) (0.0016) (0.0029)
Arellano–Bond test for AR(1) Z=5.09 z=5.02 z=4.29
p=0.000 p=0.000 p=0.000
Arellano–Bond test for AR(2) Z=0.45 z=1.54 z=0.24
p=0.653 p=0.124 p=0.807
instruments (lags of differenced dep. var.) 2,3,4,5 2,3 2,3,4,5,6
Hansen test for over-identification w
2
=2.09 w
2
=0.96 w
2
=3.58
p=0.553 p=0.327 P=0.466
For brevity, we omit the year dummies. Full results are available from the authors upon request.
***Statistically significant at the 1 percent significance level.
**Statistically significant at the 5 percent significance level.
*Statistically significant at the 10 percent significance level.
Robert A. Baade et al.
Economic Impact of Political Conventions
527
Eastern Economic Journal 2009 35
political convention cannot be rejected; however, these results do little to bolster
claims of large positive economic impacts from major conventions due to the
inconsistency of the signs on the coefficients and magnitude of the confidence
intervals.
CONCLUSIONS
This paper provides an empirical examination of the economic impact of the
Democratic and Republican National Conventions on local economies. Confirming
the results of other ex post analyses of mega-events, particularly sporting events, this
paper finds no statistically significant evidence that these huge conventions
contribute positively to a host city’s economy. Our analysis from 1970–2005 of
the 50 largest metropolitan areas in the country, including all cities that have hosted
one of the national conventions during this time period, finds that the presence of a
national political convention has no discernable impact on employment, personal
income, or personal income per capita in the cities where the events were held.
While the conventional wisdom regarding national conventions is that they bring
fame and fortune to host cities, our results suggest that any economic benefits are
quite elusive. Indeed, as noted by Matheson [2006] and Mondello and Rishe [2004],
instead of bidding to host these premier conventions, cities may be better served by
pursuing a larger number of small and medium-sized events that result in less
crowding out, lower hosting and security costs, and less leakages of visitor spending
from the local economy. Above all, people should view promises of economic
windfalls from hosting national political conventions in the same way they should
view the campaign promises of the candidates at these very conventions — with
skepticism.
Acknowledgements
This research was supported by a grant to Holy Cross from the May and Stanley
Smith Charitable Trust. The authors thank Kim Makuch and Jim Doyle for
excellent research assistance.
Appendix
See Table A1.
Table A1 Data used in OTHER vector
Event MSA(s)Values and years
Oil Boom Houston 1974=1
Oil Bust Dallas/Fort Worth, Denver,
Houston, New Orleans,
Oklahoma City
1983=1, 1984=1, 1985=1,
1986=1, 1987=1
Hurricane Katrina New Orleans 2005=1
Hurricane Andrew Miami/Fort Lauderdale 1992=1, 1993=1
Tech Boom — San Jose San Jose 1999=1, 2000=1, 2001=1
Tech Boom — San Francisco San Francisco 2000=1, 2001=1
9/11 New York/Newark 2001=1
Robert A. Baade et al.
Economic Impact of Political Conventions
528
Eastern Economic Journal 2009 35
References
Arellano, Manuel, and Stephen Bond. 1991. Some Tests of Specification for Panel Data: Monte Carlo
Evidence and an Application to Employment Equations. Review of Economic Studies, 58: 277–297.
Baade, Robert A., Robert Baumann, and Victor A. Matheson. 2008a. Selling the Game: Estimating the
Economic Impact of Professional Sports through Taxable Sales. Southern Economic Journal, 74(3):
794–810.
_______. 2008b. The Economic Impact of College Football Games on Local Economies. Journal of Sports
Economics, 9(6): 628–643.
Baade, Robert A., and Victor A. Matheson. 2001. Home Run or Wild Pitch? Assessing the Economic
Impact of Major League Baseball’s All-Star Game. Journal of Sports Economics, 2(4): 307–327.
_______ . 2002. Bidding for the Olympics: Fool’s Gold?, in Transatlantic Sport: The Comparative Economics
of North American and European Sports, edited by Carlos Pestana Barros, Muradali Ibrahimo and Stefan
Szymanski. London: Edward Elgar Publishing, 127–151.
_______ . 2004. The Quest for the Cup: Assessing the Economic Impact of the World Cup. Regional
Studies, 38(4): 343–354.
_______ . 2006. Padding Required: Assessing the Economic Impact of the Super Bowl. European Sports
Management Quarterly, 6(4): 353–374.
Blundell, Richard, and Stephen Bond. 1998. Initial Conditions and Moment Restrictions in Dynamic
Panel Data Models. Journal of Econometrics, 87: 115–143.
Bond, Stephen 2002. Dynamic Panel Data Models: A Guide to Micro Data Methods and Practice.
Portuguese Economic Journal, 1: 141–162.
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(4): 239–252.
Coates, Dennis, and Craig Depken. 2006. Mega-Events: Is the Texas-Baylor Game to Waco What
the Super Bowl is to Houston? International Association of Sports Economists Working Paper Series,
No. 06–06.
Coates, Dennis, and Brad Humphreys. 1999. The Growth Effects of Sports Franchises, Stadia and
Arenas. Journal of Policy Analysis and Management, 14(4): 601–624.
_______ . 2002. The Economic Impact of Post-Season Play in Professional Sports. Journal of Sports
Economics, 3(3): 291–299.
_______. 2003. The Effect of Professional Sports on Earnings and Employment in the Services and Retail
Sectors in US Cities. Regional Science and Urban Economics, 33(2): 175–198.
Conventions Industry Council (CIC). 2005. Meetings Industry is 29th Largest Contributor to the Gross
National Product. http://www.conventionindustry.org/aboutcic/pr/pr_091305.htm. Posted September
13, 2005 (accessed June 1, 2009).
Hagn, Florian, and Wolfgang Maennig. 2007a. Labour Market Effects of the 2006 Soccer World Cup in
Germany, International Association of Sports Economists Working Paper Series, No. 07–16.
_______ . 2007b. Short-Term to Long-Term Employment Effects of the Football World Cup 1974 in
Germany, International Association of Sports Economists Working Paper Series, No. 07–21.
Hansen, Lars. 1982. Large Sample Properties of Generalized Method of Moments Estimators.
Econometrica, 56: 1371–1395.
Holtz-Eakin, Douglas, Whitney Newey, and Harvey S. Rosen. 1988. Estimating Vector Autoregressions
with Panel Data. Econometrica, 56: 1371–1396.
Im, Kyung So M., Hashem Pesaran, and Yongcheol Shin. 2003. Testing for Unit Roots in Heterogeneous
Panels. Journal of Econometrics, 115(1): 53–74.
Jasmand, Stephanie, and Wolfgang Maennig. 2007. Regional Income and Employment Effects of the
1972 Munich Olympic Summer Games, International Association of Sports Economists Working
Paper Series, No. 07–12.
Levin, Andrew, Chien-Fu Lin, and Chia-Shang James Chu. 2002. Unit Root Tests in Panel Data:
Asymptotic and Finite Sample Properties. Journal of Econometrics, 108(1): 1–24.
Malanga, Steven. 2004. The Convention Center Shell Game. City Journal, 14(2).
Matheson, Victor. 2004. Economic Multipliers and Mega-Event Analysis, College of the Holy Cross
Working Paper Series, 04–02.
_______. 2006. Is Smaller Better? A Comment on ‘‘Comparative Economic Impact Analyses’’ by Michael
Mondello and Patrick Rishe. Economic Development Quarterly, 20(2): 192–195.
Mills, Edwin S., and John F. McDonald eds. 1992. Sources of Metropolitan Growth. New Brunswick, NJ:
Center for Urban Policy Research.
Robert A. Baade et al.
Economic Impact of Political Conventions
529
Eastern Economic Journal 2009 35
Mondello, Michael, and Patrick, Rishe. 2004. Comparative Economic Impact Analyses: Differences
across Cities, Events, and Demographics. Economic Development Quarterly, 18(4): 331–342.
Porter, Phil. 1999. Mega-Sports Events as Municipal Investments: A Critique of Impact Analysis, in
Sports Economics: Current Research, edited by John Fizel, Elizabeth Gustafson and Larry Hadley.
Westport, CT: Praeger Press.
Roodman, David. 2006. How to do Xtabond2: An Introduction to ‘‘Difference’’ and ‘‘System’’ GMM in
Stata, Working Paper 103, Washington: Center for Global Development.
Windmeijer, Frank 2005. A Finite Sample Correction for the Variance of Linear Two-Step GMM
Estimators. Journal of Econometrics, 126(1): 25–51.
Wooldridge, J. 2002. Introductory Econometrics: A Modern Approach, 2nd ed., New York: South-Western
College Publishers.
Robert A. Baade et al.
Economic Impact of Political Conventions
530
Eastern Economic Journal 2009 35