Atl Econ J (2017) 45:45–57

DOI 10.1007/s11293-016-9526-2

Swing States, the Winner-Take-all Electoral College,

and Fiscal Federalism

Christopher M. Duquette 1 & Franklin G . Mixon Jr 2 &

Richard J. Cebula 3

Published online: 31 January 2017

# International Atlantic Economic Society 2017

Abstract The impact of the Electoral College on U.S. presidential elections is often
criticized by academics and political commentators. One facet of its impact, the
winner-take-all allocation of states’ Electoral College votes, serves in practice to
overweight some votes in some states relative to other votes in other states. These
disparities in the relative impact of votes in a presidential election can be large. Here, a
metric is introduced to quantify the magnitude of these disparities in each presidential
election. Using that metric, we show that states whose votes were overweighted in a
presidential election subsequently received higher levels of federal grant spending
under the newly-elected (re-elected) administration.

Keywords Swing voters . ElectoralCollege . Fiscalfederalism . Intergovernmentalgrants

JEL D72 . D78 . D70

The authors are grateful to an anonymous referee for providing helpful comments on a previous version. Any
remaining errors are our own. Lastly, the first author’s affiliation with The MITRE Corporation is provided for
information purposes only and is not intended to convey or imply MITRE’s concurrence with, or support for,
any positions, opinions, or viewpoints expressed by the authors.

* Franklin G . Mixon, Jr
[email protected]

Christopher M. Duquette
[email protected]
Richard J. Cebula
[email protected]

1
The MITRE Corporation, McLean, VA 22102, USA
2
Center for Economic Education, Columbus State University, Columbus, GA 31907, USA
3
Davis College of Business, Jacksonville University, Jacksonville, FL 32211, USA
46 Duquette C.M. et al.

Introduction

In the U.S., a fundamental principle of voting rights is that everyone is equal in the eyes
of the law and hence all votes carry “equal weight.” This principle underlies the U.S.
Supreme Court’s rulings on “one person, one vote.” As a practical matter, however,
some votes do count more than others. A robust literature concludes that during the
presidential primary and caucus process, voters in states such as Iowa or New
Hampshire effectively have a greater voice in the election than those in other states
(Schier, 1980; Bartels 1988; Brady, 1989; Mixon and Hobson, 2001; Mayer and Busch,
2004; Hull, 2008). This conclusion is attributable to the number of voters in these states
and the strategic importance of having their primaries and caucuses positioned at the
beginning of the presidential primary/selection process. Moreover, the Electoral College
is sometimes criticized as giving disproportionate influence to some voters or states, or
as otherwise impacting the results in presidential elections because of its
winner-take-all method of allocating votes in 48 of the 50 states (Cebula and Murphy
1980; Pomper 2001, p. 150). But these assertions notwithstanding, can the impact that
swing-voters have in some states compared to others, in terms of their relative influence
on presidential elections, actually be quantified? Relatedly, does the Electoral College
actually affect the impact of swing voters? This study presents a new method to assess
the impact of swing voters within the winner-take-all method that states use to allocate
electoral votes. By looking at several recent U.S. presidential elections, we quantify how
the winner-take-all method of allocating electoral votes produces disparities in the
voting power of citizens across states. This study also provides an empirical test
involving an important policy implication pertinent to this issue.

Presidential Elections and the Electoral College System

The Electoral College is an intriguing institution in American politics. The framers of the
Constitution defended it as critical to producing “extraordinary persons” as Presidents
because they would be selected by “men most capable of analyzing the qualities adapted
to the station” of the Presidency (Madison 1937, p. 444). Others have noted that, with an
Electoral College, recounts would not need to be done nationally, but only in specific
jurisdictions where there were disputes (Posner 2001, pp. 224–227).
The Electoral College has also had its detractors. It has been criticized as anachro-
nistic, having outlived the purpose for which it was created (Glennon, 1992). Following
the 2000 presidential election, where George Bush lost the national popular vote to Al
Gore but won the Electoral College vote, those criticisms intensified (Bugliosi, 2001;
Dershowitz, 2001). Others find that the Electoral College disproportionately weights the
votes of smaller states relative to larger states (Banzhaf, 1968), that it serves to depress
and/or alter voter turnout (Keyssar, 2001; Cebula, 2002), or that it creates a system of
wasted votes (Cebula and Meads, 2008; Edwards, 2004).
Of special interest here is the effect of the winner-take-all allocation of each state’s
Electoral College votes. At present, the Electoral College votes in all but two states,
Maine and Nebraska, are allocated to each state’s popular vote winner. This winner-
take-all allocation is not mandated by the U.S. Constitution. The U.S. Constitution
provides that each state’s electors shall be appointed in a manner to be determined by its
Swing States, Winner-Take-All Electoral College 47

legislature. The only stipulation is that a sitting member of Congress cannot also serve
as an elector. With two exceptions, states have opted for a winner-take-all allocation, in
order to maximize their influence on the outcome of the presidential election.
While Mayhew (2010, pp. 196-198) contends there is no partisan bias to the Electoral
College, there is evidence that it does impact election results (Cebula and Meads, 2008).
The winner-take-all effect ensures that small swings in state-vote margins can dispro-
portionately influence the Electoral College count. In a close election, such swings can
even determine the winner. The extreme case is the 2000 presidential election, where
537 popular votes in Florida represented the difference in awarding the state’s then 25
Electoral College votes, and, ultimately, the election for Bush over Gore. Four years later,
in the presidential election of 2004, the margin of victory for George Bush over John F.
Kerry were the 119,000 votes in Ohio that garnered its 20 Electoral College votes. In the
1976 presidential election, the margin of victory for Democrat Jimmy Carter over
Republican Gerald Ford totaled 175,000 votes in three states. Nearly half of Carter’s
297–240 Electoral College vote margin over Ford was attributable to his winning Ohio’s
then 25 Electoral College votes by a margin of 11,116 popular votes.

Winner-Take-All Impact on Elections: Statistical Analysis

The winner-take-all method of allocating electoral votes serves to overweight

(underweight) votes from the small-margin (large-margin) states. It is even possible to
quantify the magnitude of this issue. The swing votes are the critical element
representing the margin of victory for the winning candidate. They are the votes that
swung the state to the winning candidate. The other votes for each major-party candidate
offset each other. Thus, the swing votes are the ones of interest.
The key is to determine what the Electoral College’s impact is on each state’s
swing votes in an election. That is done here, for each state, by dividing the
number of Electoral College votes at stake (N) by the margin of victory for the
winning candidate (M). Thus, the swing voters’ impact can be expressed as S = N/
M. Invariably, the margin of victory dwarfs the number of Electoral College votes,
and the S-fraction is minute. For ease of interpretation, each state’s S-fraction can
be normalized with respect to the middle-ranking (median) state for that election.
The states can then be ranked, in order of their swing voters’ Electoral College
impact, for each election.1
In the 2012 presidential election, President Barack Obama and Vice President Joseph
Biden defeated Governor Mitt Romney and Representative Paul Ryan by a 332–206
Electoral College margin. Obama and Biden won 26 states (and the District of Columbia)
to 24 states for Romney and Ryan. The popular-vote tally was 65.8 million votes for
Obama and Biden and 60.7 million votes for Romney and Ryan. The state with the
smallest popular-vote margin relative to the number of Electoral College votes at stake
was Florida. Its 29 Electoral College votes went for Obama and Biden by a margin of
73,189 popular votes. Those swing votes, representing the margin of victory for

1
The process of calculating the S-fraction for each state and normalizing the results with respect to the middle-
ranking state follows that of Cebula, Duquette and Mixon (2013). They calculated the metric for the
presidential elections of 2004 and 2008 and examined its relationship to state-level voter turnout.
48 Duquette C.M. et al.

one ticket over the other, are of particular interest. Dividing the number of Electoral College
votes at stake (29) by the popular vote margin (73,189) yields the S-fraction, or the Electoral
College impact of each swing vote in that state. In Florida, it is equal to 0.000396. The
number is small because each state has many more popular votes than Electoral College
votes. Relatively speaking, though, Florida’s S-fraction was the largest of all the states
during the 2012 election.
The full slate of state-by-state results for the 2012 presidential election is presented
in Table 1. The states are ranked by S-fraction, with the results normalized relative to
the median state. Rhode Island was the median state for the 2012 election. Its four
Electoral College votes went for Obama and Biden by a margin of 122,473 popular
votes. Dividing the Electoral College votes at stake in Rhode Island by the state’s
popular-vote margin yields an S-fraction of 0.000033 Electoral College votes per swing
vote in Rhode Island. That result was normalized to 1.0 for purposes of comparison
with the other states. The same calculations were performed for the other states, except
for Maine and Nebraska, plus the District of Columbia, with the results expressed
relative to the normalized value for Rhode Island.
As shown in Table 1, the differences between states are sizeable. Florida tops the list
for the 2012 election as the state where the Electoral College impact of a swing vote was
the largest. A swing vote in Florida carried 11.95 times the Electoral College impact of
one in Rhode Island, the median state.2 At the other end of the spectrum is the District of
Columbia. The District’s three Electoral College votes went for Obama and Biden by a
popular-vote margin of 245,689. Dividing the Electoral College votes at stake in the
District by its popular-vote margin yields an S-fraction of 0.000012 Electoral College
votes per swing vote, which equates to 0.37 of the figure for Rhode Island (0.000033). A
swing vote in the District carried just under two-fifths the Electoral College impact of a
swing vote in Rhode Island.
The two battleground states of Ohio and Virginia are near the top of the list. Obama and
Biden won Ohio’s 18 Electoral College votes by a margin of 166,214 popular votes.
Virginia’s 13 Electoral College votes also went to Obama and Biden, by a margin of
149,298 popular votes. A swing vote in Ohio had 3.32 times the Electoral College impact
of one in Rhode Island; one in Virginia had 2.67 times the impact. Ohio was decisive in the
victory of the Republican Bush/Cheney ticket over the Democratic Kerry/Edwards ticket in
2004; it was also hotly contested in 2008 and 2012. Virginia became a battleground state in
2008 and 2012 after having consistently voted Republican in every presidential election
from 1968 through 2004. Those two states received disproportionate attention in terms of
candidate time and advertising resources during the 2012 campaign.
California and New York, the first- and third-most populous states, fall near the bottom
of the list. California’s 55 Electoral College votes went for Obama and Biden by a margin
of 3,014,327 popular votes. New York awarded its 29 Electoral College votes to Obama
and Biden by a popular-vote margin of 2,100,831 votes. A swing vote in California had
0.56 times the Electoral College impact of one in Rhode Island. One in New York had 0.42
times the impact. The election outcomes in both states were treated as almost a foregone
conclusion in those states, and they received hardly any attention from the

2
In other words, each of the 73,189 voters providing the margin of victory in Florida’s 2012 presidential
election was almost 12 times more influential in the Electoral College math than were each of the 122,473
voters providing the margin of victory in Rhode Island’s 2012 presidential election.
Swing States, Winner-Take-All Electoral College 49

Table 1 State S-fractions, 2012 presidential election

Rank State Relative Electoral College impact

1 Florida 11.95
2 North Carolina 4.99
3 Ohio 3.32
4 New Hampshire 3.09
5 Nevada 2.71
6 Virginia 2.67
7 Alaska 2.19
8 Colorado 2.00
Iowa 2.00
10 Pennsylvania 1.98
11 New Mexico 1.92
12 Arizona 1.62
13 Georgia 1.61
14 Wisconsin 1.46
15 North Dakota 1.45
16 South Dakota 1.40
17 Montana 1.39
18 Minnesota 1.36
19 South Carolina 1.34
20 Indiana 1.26
21 Mississippi 1.24
22 Delaware 1.19
23 Missouri 1.18
24 Michigan 1.09
25 Rhode Island 1.00
26 Oregon 0.99
27 Texas 0.92
28 Wyoming 0.90
29 Vermont 0.86
30 West Virginia 0.85
31 Connecticut 0.79
Washington 0.79
33 Kansas 0.73
Arkansas 0.73
35 Louisiana 0.71
36 Illinois 0.69
37 Tennessee 0.67
38 New Jersey 0.66
Hawaii 0.66
40 Kentucky 0.60
Alabama 0.60
42 Idaho 0.59
43 California 0.56
44 Oklahoma 0.48
45 Massachusetts 0.46
46 Maryland 0.43
47 New York 0.42
50 Duquette C.M. et al.

Table 1 (continued)

Rank State Relative Electoral College impact

48 Utah 0.38
49 District of Columbia 0.37
n/a Maine n/a
n/a Nebraska n/a

The reference point for the relative Electoral College impact scores above is Rhode Island, which is the median
state. For example, Florida’s score of 11.95 means that each of the 73,189 voters providing the margin of
victory in Florida’s 2012 presidential election was almost 12 times more influential in the overall Electoral
College math than were each of the 122,473 voters providing the margin of victory in Rhode Island’s 2012
presidential election. Data Source: Statistical Abstract of the United States (various issues), Washington, D.C.:
Bureau of the Census

candidates other than for fund-raising purposes. The outcome was somewhat closer in
Texas, the second-most-populous state. Texas awarded its 38 Electoral College votes to
Romney and Ryan by a popular-vote margin of 1,261,719. A swing vote in Texas carried
0.92 times the impact of one in Rhode Island.
The results for Ohio and Virginia relative to those in California and New
York underscore how the winner-take-all allocation of states’ Electoral College
votes serves to underweight votes from the large-margin states. Swing votes
from the large-margin states carry a smaller impact on the election’s outcome
than those from the small-margin states. From a strategy standpoint, the candi-
dates are well-advised to direct their resources toward any of the small-margin
states, where a last-minute campaign swing could make the difference in
moving enough votes to swing the full complement of the state’s Electoral
College votes, and avoid any of the large-margin states, where upwards of a
million or more votes would have to be swung to affect the statewide outcome.
That is what happened in 2012, as the Obama/Biden and Romney/Ryan tickets
both aggressively targeted Ohio and Virginia and largely avoided California and
New York.
Maine and Nebraska are two special cases listed at the very bottom of the
table. No Electoral College impact calculation is reported for either one, the
reason being that neither allocates its Electoral College votes on a winner-take-
all basis. Rather, both award their Electoral College votes by congressional
district, with the plus-two bonus going to the statewide popular vote winner.
Maine has done so since 1972, while Nebraska has since 1996. Previously, both
states allocated their Electoral College votes on a winner-take-all basis, like all
the other states. Both Maine and Nebraska are relatively homogeneous states,
and have seldom split their Electoral College votes.3
Table 2 uses the same methodology to present Electoral College voting
ratios back to 1960. The 1960 election was the first with the Electoral
College at its current 538 electors, which resulted after Alaska and Hawaii
joined the union.

3
Nebraska only once (2012) split its Electoral College votes, Maine has never done so.
Swing States, Winner-Take-All Electoral College 51

Table 2 Impact of a swing vote: Ratio of top (ST) to bottom (SB) ranked states, 1960–2012

Election Top-ranked state Bottom-ranked state ST/SB

1960 Hawaii Massachusetts 832

1964 Arizona Rhode Island 63
1968 Arkansas Massachusetts 71
1972 Minnesota Florida 7
1976 Ohio Utah 88
1980 Massachusetts Utah 288
1984 Minnesota Utah 209
1988 Vermont Florida 16
1992 Georgia District of Columbia 54
1996 Nevada Massachusetts 59
2000 Florida Utah 2905
2004 Wisconsin Utah 65
2008 Missouri District of Columbia 215
2012 Florida District of Columbia 32

Interpretation of the ratio of top (ST) to bottom (SB) ranked states for Florida’s 2000 presidential election
means that each of the 537 voters providing the margin of victory in Florida’s 2000 presidential election was
almost 3000 times more influential in the overall Electoral College math than were each of the 312,043 voters
providing the margin of victory in Utah’s 2000 presidential election
Data Source: Statistical Abstract of the United States (various issues), Washington, D.C.: Bureau of the Census

Table 2 shows that the 2012 election was not unusual in terms of the over- and
underweighting of votes induced by the winner-take-all allocation of states’ Electoral
College votes. The 2000 election had the highest swing-vote impact, due to the razor-
thin margin in Florida for the Republican Bush/Cheney ticket over the Democratic
Gore/Lieberman ticket. In 2000, the Bush/Cheney ticket won Florida by a mere 537
popular votes. That popular vote margin swung the state’s full complement of 25
Electoral College votes, and the election, to the Republican ticket. Florida was the state
with the largest Electoral College impact of a swing vote in 2000, whereas Utah was the
state with the smallest. Utah’s five Electoral College votes went to Bush and Cheney by
a margin of 312,043 popular votes. Comparing these two states, a swing vote in Florida
carried 2905 times the Electoral College impact of a swing vote in Utah.4
The elections of 1960, 1980, and 2008 had the next-largest Electoral College swing-
vote impact differential. For each election, the Electoral College impact of a swing vote in
the highest-ranked state was more than 200 times that of a swing vote in the lowest-ranked
state. The elections of 1960 and 2000 were cliffhangers, unlike the elections of 1980 and
2008. Whether the election was close or not did not matter, as every election produced
significant over- and underweighting due to the winner-take-all allocation of states’
Electoral College votes. This result will hold whenever the election outcomes are not
evenly distributed across all the states.

4
In other words, each of the 537 voters providing the margin of victory in Florida’s 2000 presidential election
was almost 3000 times more influential in the Electoral College math than were each of the 312,043 voters
providing the margin of victory in Utah’s 2000 presidential election.
52 Duquette C.M. et al.

Swing States and Fiscal Federalism: an Empirical Testing Application

Empirical studies investigating various political or economic distortions related to the

Electoral College system are not new. Wright’s (1974) analysis of federal government
spending in the states during the “New Deal” indicates that Electoral College votes per
capita across states is the most significant explanatory variable. At the same time, Brams
and Davis (1974) argue that presidential campaign resource allocations exhibit a
large-state bias, based on Electoral College counts. In an interesting analysis of
President Abraham Lincoln’s quest for a second presidential term, Anderson and
Tollison (1991) present statistical results suggesting that

“[. . .] Northern causalities [during the Civil War] were partly determined by
electoral votes in 1864 [. . .] Given that the Northern troops were organized by
states and that President Lincoln sought to be reelected, . . . [t]roops from closely-
contested states were much less likely to suffer causalities . . . [based on the logic
that] . . . dead men cannot vote ” (Tollison, 2004, p. 558).

Grier et al. (1995) find, using a pooled sample (1970–1988) of 325 individual bills,
that the presidential veto is sensitive to winner-take-all electoral voting in the 50 states.
Statistical evidence indicates that presidential veto decisions are influenced by
the floor votes of senators from closely-contested, larger states, where political
payoff is highest.5
The Electoral College-related asymmetries in the impact of a swing vote across the 50
states carries with it public choice implications regarding fiscal federalism. Mixon and
Hobson (2001) show that intergovernmental grants, from the federal level to the states, are
sensitive to the presidential primaries/caucuses calendar during each presidential election
cycle. As Mixon and Hobson (2001) indicate, by frontloading the dates of their primaries
and caucuses, states can play a more important role in shaping the field of presidential
candidates within the two major political parties. Specifically, results in Mixon and
Hobson (2001) suggest that a 10.36 day movement (closer to 1 January) of a state’s
primary or caucus results in an increase of $181 million to $600 million per presidential
election year in federal grants to state governments as the current administration attempts
to shape the primary election’s outcome.
In an earlier, yet related study, Mixon and Ladner (1998) apply the public choice model
of Tullock (1967) and Posner (1975) to events surrounding the 1995 Republican takeover
of the U.S. House of Representatives. As part of the Republican campaign platform of
1994, many candidates adopted the “Contract with America,” which promised to reinforce
adherence to the Tenth Amendment to the U.S. Constitution by sending various fiscal
responsibilities, often accompanied by federal receipts, back to the 50 states.6 As Mixon
and Ladner (1998, p. 31) posit, as the federal government bundles tax receipts in the form
of block grants for states, the value of holding state office would increase, thereby enticing
5
See Tollison (2004) for a concise review of these and other studies related to the Electoral College and
presidential politics.
6
Bills passed by the U.S. House resulting from the “Contract with America” commitment made by
Republican candidates include the Local Government Law Enforcement Block Grant of 1995, the Unfunded
Mandate Reform Act of 1995, and the Personal Responsibility Act of 1995. The second of these three
ultimately became law. Though the other two bills failed to pass in the U.S. Senate, elements of them found
their way into various federal spending bills later in 1995 and 1996.
Swing States, Winner-Take-All Electoral College 53

a larger pool of candidates for those offices. Event study results from the panel data set
employed by Mixon and Ladner (1998) suggest that the block grant movement that
occurred after the Republican victory in 1994 led to a significant increase in the number
of candidates for state house seats.
As an extension of the above studies, the model in eq. (1) is proposed as a test of the
impact of swing states’ votes on federal grants to state and local governments:
RGRANTPOPtþ2 ¼ α þ β1 SWING−S t þ β2 POPU tþ2 þ β3 HOMESTATEtþ2

þ β4 FEDLAND%tþ2 þ β5 ALASKAtþ2 þ β6 YEAR þ ε: ð1Þ

In the model above, the subscript, t, represents a presidential election year, where t is
equal to either 2000 or 2004, meaning that the dependent variable, RGRANTPOP, and
several of the regressors are measured either in 2002 or 2006. In (1) above, the
dependent variable RGRANTPOP is equal to real per-capita federal government grants
to state and local governments, both for 2002 and 2006, of the states carried by the
victorious presidential candidates in 2000 and 2004, respectively. These two elections
were won by Republican George Bush, who previously served as governor of Texas.
Thus restriction of our analysis to pooled data covering the 2000 and 2004 presidential
elections avoids any potential statistical complications associated with political party
control that would otherwise be encountered if either of the two more recent elections
of 2008 or 2012 were included in the sample. Moreover, given that data on
RGRANTPOP for 2014, the applicable year for the 2012 presidential election, were
not available at the time of this study, restriction of the sample to the 2000 and 2004
elections, and not also the 2008 election, avoids empirical estimation using unbalanced
panel data.
Federal government grants (i.e., RGRANTPOP) are a function of six independent
variables, including our variable of interest, SWING-S, the ratio of Electoral College votes
(N) to margin of victory (M) in the presidential elections of 2000 and 2004. If swing states
are rewarded by victorious presidential candidates, one such potential reward is through
increased intergovernmental grants in the years after the presidential election. This study
examines year t + 2 in order to allow for lags in the legislative process. The rent-seeking
theory described here predicts a positive sign for β1.7
The variable POPU, the population in each state (in 2002 or 2006) captures
essentially what is referred to in the economic growth literature as the “catch-up effect.”
Here, large states are expected to lag behind smaller states in terms of per-capita
intergovernmental support. A number of studies have shown that presidential candi-
dates perform well in the general election in their home states (Lewis-Beck and Rice,
1983; Kjar and Laband, 2002; Mixon and Tyrone, 2004; Mixon, 2013).8 One would
expect, then, that victorious Presidents want to reward their home states, possibly
through larger intergovernmental grants. As such, HOMESTATE, which is a dummy

7
Summary statistics and results from ordinary least squares (OLS) estimation of equation (1) above are
presented in Table 3. The pooled model is jointly significant (F-statistic =29.71), and produces an R2 of 0.774.
Additionally, all of the parameter estimates are correctly signed, with five (out of six) statistically significant as
well.
8
Kopko and Devine (2016) find that localism in voting carries over, at least marginally, to vice presidential
candidates, while research by Malcová (2012) finds evidence of localism in Senate elections in the
Czech Republic.
54 Duquette C.M. et al.

variable equal to 1 for the home states of the victorious presidential and vice presiden-
tial candidates in the 2000 and 2004 presidential elections, and 0 otherwise, is included
in (1) above. Given this discussion, we expect to find β2 < 0 and β3 > 0.
States with large land masses that are owned and administered by the federal government
are expected, ceteris paribus, to receive fewer federal grant dollars. Furthermore, with its
transportation obstacles and oil and gas reserves, Alaska might receive larger amounts of
federal grants than other states, ceteris paribus. Thus, FEDLAND%, or the percentage of
each state’s land area owned by the federal government, and ALASKA, a dummy variable
equal to 1 for observations on Alaska, and 0 otherwise, are included in (1) above. It is
expected that β4 > 0 and β5 > 0. Lastly, the dummy variable YEAR is equal to one for 2006
observations on RGRANTPOP, and zero otherwise (i.e., for 2002 observations on
RGRANTPOP). It is included in (1) to capture any regime or structural effects present in
the RGRANTPOP data present in the two-year data panel.
Summary statistics for our pooled data are provided in the first column of Table 3.
As noted there, about 7 % of the states in the sample represent the home state of either
the winning presidential or vice presidential candidate, while Alaska represents about 3
% of the observations in the sample. Additionally, almost 20 % of the average state’s
land mass is owned by the federal government, while the average state in the sample
has just over five million residents. The mean value for the variable of interest, SWING-
S, is 20.7. Next, the results from OLS estimation of (1) are provided in the final two
columns in Table 3. The first of these provides estimates for the pooled data (i.e., the

Table 3 Swing votes and fiscal rewards: OLS results (dep var = RGRANTPOP)

Pooled results 2000 results

Variable Parameter estimates Parameter estimates

Constant 926.60 943.76

SWING-S 0.29* (1.78) 0.27* (1.92)
[20.7, 145.0]
POPU −0.04** (−7.74) −0.04** (−6.01)
[5030, 4926]
HOMESTATE 724.75** (7.67) 564.60** (5.27)
[0.07, 0.25]
FEDLAND% −3.32** (−3.22) −3.97** (−3.59)
[19.5, 24.3]
ALASKA 1166.80** (9.16) 1298.70** (9.36)
[0.03, 0.18]
YEAR 49.11 (1.14)
[0.51, 0.50]
n 59 29
F-statistic 29.71** 27.87**
R2 0.774 0.858

The pooled results employ data from both the 2000 and 2004 elections. The numbers in brackets beneath the
variables are means and standard deviations, respectively. The mean and standard deviation for RGRANTPOP
are 769.8 and 325.8, respectively. Terms in parentheses beside the parameter estimates are t-statistics, where
** denotes statistical significance at the 1% level and * denotes statistical significance at the 10% level
Data Source: Statistical Abstract of the United States (various issues), Washington, D.C.: Bureau of the Census
Swing States, Winner-Take-All Electoral College 55

2000 and 2004 elections) whereas the second provides parameter estimates for only the
2000 cross section. As shown in Table 3, 10 of the 11 estimated coefficients exhibit the
expected signs and are statistically significant at the 10% level or better. Indeed, in eight
of these 10 cases, the coefficient is actually statistically significant at the 1% level. The
R2 is 0.77 in the first estimation and 0.86 in the second; thus, the model in (1) explains
in excess of three-fourths of the variation in the dependent variable in each case.
Furthermore, the F-statistics are both statistically significant at the 1% level, attesting
to joint significance of the parameter estimates.
In terms of individual results, it appears that real per capita federal grants to state and local
governments are positively and significantly (at the 1% level) associated with both
HOMESTATE and ALASKA, while they are negatively and significantly (at the 1% level)
associated with POPU and FEDLAND%. In fact, only the variable YEAR fails to reach
statistical significance. As for the swing-state variable, SWING-S, its coefficients are both
positive and statistically significant. Indeed, the coefficient of SWING-S from the pooled-
data estimation (i.e., the 2000 and 2004 elections) is significant at approximately the 8%
level, whereas its counterpart from the cross-sectional model (i.e., the 2000 election) is
significant at approximately the 6% level. Hence, there is at least preliminary evidence that
“swing state” status for a state brings with it the potential reward/benefit of being associated
with the receipt of greater real per capita federal government grants.

Concluding Remarks

This study quantifies how the winner-take-all method of allocating electoral votes produces
disparities in the voting power of citizens across states, and, in doing so, also provides
empirical evidence that such disparities are rewarded in the fiscal policy process. More
specifically, we provide at least preliminary empirical evidence that “swing state” status for a
given state in a given presidential election brings with it the potential reward/benefit of being
the recipient of greater real per capita federal government grants to state and local
governments.
Clearly, the implication of swing-state status for the U.S. political and economic
landscapes warrants further investigation, some of which might address limitations of this
study, such as its lack of emphasis on the role played by political party affiliation. Future
research could parse the impact of political party affiliation on the process examined here by
examining a balanced panel of the four most recent presidential elections (i.e., 2000, 2004,
2008 and 2012). While the first two of these were won by Republican George Bush, the
latter two were both won by Democrat Barack Obama. Additionally, future research might
also investigate alternative methodologies for quantifying swing votes across the 48 states
that employ winner-take-all vote counts, as well as how to integrate the remaining two
states, those that award their Electoral College votes by congressional district, with the plus-
two bonus going to the statewide popular vote winner, into statistical analyses.

References

Anderson, G. M., & Tollison, R. D. (1991). Political influence on civil war mortality rates: the electoral college
as a battlefield. Defense Economics, 2, 219–234.
56 Duquette C.M. et al.

Banzhaf, J. F. (1968). One man, 3.312 votes: a mathematical analysis of the electoral college. Villanova Law Review,
13, 304–332.
Bartels, L. M. (1988). Presidential primaries and the dynamics of public choice. Princeton: Princeton
University Press.
Brady, H. E. (1989). Is Iowa news? In P. Squire (Ed.), The Iowa caucuses and the presidential nominating
process (pp. 89–119). Boulder: Westview Press.
Brams, S. J., & Davis, M. D. (1974). The 3/2’s rule in presidential campaigning. American Political Science Review,
68, 113–134.
Bugliosi, V. (2001). The betrayal of America: how the supreme court undermined the constitution and chose
our president. New York: Nation Books.
Cebula, R. J. (2002). The electoral college and voter participation: evidence on two hypotheses from the 1996
presidential election. Atlantic Economic Journal, 30, 67–75.
Cebula, R. J., Duquette, C. M., & Mixon Jr., F. G. (2013). Battleground states and voter participation in US
presidential elections: an empirical test. Applied Economics, 45, 3795–3799.
Cebula, R. J., & Meads, H. (2008). Voter participation and political party dominance. Atlantic Economic
Journal, 36, 53–64.
Cebula, R. J., & Murphy, D. (1980). The electoral college and voter participation rate: an exploratory note.
Public Choice, 35, 185–190.
Dershowitz, A. M. (2001). Supreme injustice: how the high court hijacked election 2000. New York: Oxford
University Press.
Edwards III, G. C. (2004). Why the electoral college is bad for America. New Haven: Yale University Press.
Glennon, M. J. (1992). When no majority rules: the electoral college and presidential succession. Washington,
D.C.: CQ-Roll Call Group Books.
Grier, K. B., McDonald, M., & Tollison, R. D. (1995). Electoral politics and the executive veto: a predictive
theory. Economic Inquiry, 33, 427–440.
Hull, C. C. (2008). Grassroots rules: how the Iowa caucus elects American presidents. Stanford: Stanford
University Press.
Keyssar, A. (2001). The right to vote and election 2000. In J. N. Rakove (Ed.), The unfinished election of 2000
(pp. 75–100). New York, NY: Basic Books.
Kjar, S. A., & Laband, D. N. (2002). On ‘home grown-ness’ in politics: evidence from the 1998 election for
Alabama’s third Congressional District. Public Choice, 112, 143–150.
Kopko, K. C., & Devine, C. J. (2016). The VP advantage: how running mates influence home state voting in
presidential elections. London: Bloomsbury Academic.
Lewis-Beck, M. S., & Rice, T. W. (1983). Localism in presidential elections: the home state advantage.
American Journal of Political Science, 27, 548–556.
Madison, J. (1937) The Federalist No. 68. New York: NYL The Modern Library.
Malcová, K. (2012). The local aspect of electoral support for candidates to the senate of the Czech Republic.
Czech Sociological Review, 48, 283–313.
Mayer, W. G., & Busch, A. E. (2004). The front-loading problem in presidential nominations. Washington,
D.C.: Brookings Institution Press.
Mayhew, D. R. (2010). The meaning of the 2008 election. In M. Nelson (Ed.), The elections of 2008 (pp. 187–
204). Washington, D.C.: CQ Press.
Mixon Jr., F. G. (2013). Warm springs ran deep: friends-and-neighbors voting in the U.S. presidential elections
of 1940 and 1944. Journal of Politics and Law, 6, 105–115.
Mixon Jr., F. G., & Ladner, J. A. (1998). Sending federal fiscal power back to the states: federal block grants
and the value of state legislative offices. Journal of Public Finance and Public Choice, 16, 27–41.
Mixon Jr., F. G., & Hobson, D. L. (2001). Intergovernmental grants and the positioning of presidential primaries and
caucuses: empirical evidence from the 1992, 1996, and 2000 election cycles. Contemporary Economic Policy,
19, 27–38.
Mixon Jr., F. G., & Tyrone, J. M. (2004). The ‘home grown’ presidency: empirical evidence on localism in
presidential voting, 1972-2000. Applied Economics, 36, 1,745–1,749.
Pomper, G. M. (2001). The presidential election. In G. M. Pomper (Ed.), The election of 2000 (pp. 125–154).
New York: Chatham House.
Posner, R. (1975). The social costs of monopoly and regulation. Journal of Political Economy, 83, 43–69.
Posner, R. A. (2001). Breaking the deadlock: the 2000 election, the constitution, and the courts. Princeton:
Princeton University Press.
Schier, S. E. (1980). The rules of the game: democratic National Convention Delegate selection in Iowa and
Wisconsin. Washington, D.C.: University Press of America.
Swing States, Winner-Take-All Electoral College 57

Tollison, R. D. (2004). Voting in U.S. presidential elections. In C. K. Rowley & F. Schneider (Eds.),
Encyclopedia of public choice. Dordrecht: Kluwer.
Tullock, G. (1967). The welfare costs of tariffs, monopolies and theft. Western Economic Journal, 5, 224–232.
Wright, G. (1974). The political economy of new deal spending: an econometric analysis. Review of
Economics and Statistics, 56, 30–38.