Debates over U.S. congressional involvement in foreign military
engagements trace back to the Republic's founding. But
irrespective of the normative issues at stake (for example, should
Congress shape foreign policy?), a basic point of fact remains
unresolved: Does Congress effectively constrain presidents from
exercising force abroad? While illustrative case studies abound,
few quantitative investigations systematically address the topic.
In principle, Congress should contribute to the politics surrounding
the use of military force. The U.S. Constitution vests Congress with
the power to raise and support armies, to provide and maintain a navy,
and to regulate the military forces. Congress can use its budgetary
powers to limit the scope and duration of troop
deployments—setting strict reporting requirements and sunset
provisions, targeting money for certain military activities, and
restricting the use of appropriated funds for others.1
Members of Congress can sway public
opinion against the president; just as they can impede other aspects of
his (someday her) legislative agenda.
2 While presidents retain considerable
discretion to use force as they please, Congress may be able to
increase the marginal costs of doing so. And when facing particularly
risky foreign conflicts, congressional opposition may ultimately
dissuade a president from deploying the military.
The existing quantitative literature on the use of force does not
examine whether presidents balance their “preference for
action” against congressional interests.3
Rarely are measures of congressional relations with the
president included in statistical models on the use of force; and when
included, they are crudely specified, typically nothing more than
indicator variables for divided government, the post–War Powers
Resolution era, or periods of “cold war consensus.”
4 We revisit the event-count models used to predict uses of force,
adding appropriate measures of congressional support for the president.
Our findings are unambiguous and run directly against the notion that
politics stop “at the water's edge.”5
While
Congress does not appear to constrain the president's capacity to
initiate low-level military maneuvers, sizeable effects are observed
for major military ventures—the very events that can have
electoral consequences for presidents and members of Congress.
The Quantitative Use-of-Force Literature
Military deployments short of war (such as the Berlin Airlift or the
Cuban Missile Crisis) represent potent expressions of executive
authority. Not coincidentally, they have garnered a large academic
following, beginning with the pioneering work of Blechman and Kaplan in
1978. Blechman and Kaplan were principally concerned with the
international conditions (for example, whether the Soviet Union was a
party to a crisis) that lead presidents to initiate lower-level
military ventures, which they termed “force without war.”
They identified 226 such incidents between 1946 and 1976 and tracked
when, and whether, U.S. presidents achieved their strategic
objectives.
Beginning in the mid-1980s, scholars built on Blechman and
Kaplan's data base to test a confluence of international relations
theories about interstate conflict and political psychology insights
into executive decision making. According to Ostrom and Job, U.S.
presidents use simple decision rules to balance the competing demands
placed before them. As commander-in-chief, chief executive, and
“political leader,” presidents “monitor salient
dimensions in the domestic, international, and political arenas”
before committing U.S. forces abroad.6
Domestic politics, however, retain special significance. In Ostrom and
Job's empirical analysis, the substantive impacts of domestic
variables (weighted economic misery index, presidential approval, and
national elections) on the frequency with which presidents exercise
military force consistently match or exceed those of their
international counterparts.
7 The bulk of studies following Ostrom and Job have examined how the
economy and public opinion influence presidents' decisions to
deploy troops abroad. James and Oneal introduced a new variable tapping
international threats to U.S. interests, yet found results consistent
with Ostrom and Job.8
Fordham subsequently
argued that economic factors and public opinion do not directly shape
presidential choices, but instead influence how presidents view their
environment.
9 Presidents, according to Fordham, only
“see” opportunities to use force when the domestic economy
is poor—otherwise they have few incentives to imperil their
reelection prospects with military ventures.
Other scholars have reached different conclusions, holding that
purely external factors drive decisions to use force. Meernik finds
that domestic economic forces played little to no role in predicting
U.S. uses of military force.10
Gowa contends that neither the
partisan nature of Congress, nor electoral cycles, nor the state of the
economy predicts the regularity with which presidents exercise military
force.
11 In a slightly different vein, Mitchell and Moore,
as well as Fordham, raise important issues of data comparability
(scholars use different years in analyzing their hypotheses) and
temporal dynamics (uses of force may be clustered together in time)
that may compromise previous statistical findings.
12 While much divides the protagonists in the use-of-force literature,
one assumption is consensual: Congress is weak.13
For a partial exception, see Morgan and
Bickers 1992; and Morgan and Campbell
1991. Morgan and Campbell 1991 examine
cross-nationally “legislative constraints” on executive war
powers. Because executives are coded as being constrained only when
“legislatures have the ability to overturn
[them],” Morgan and Campbell do not assess the varying
capacities of individual legislatures to constrain their executives
over time. Morgan and Bickers 1992 examine
the influence of partisanship and public opinion on decisions to use
force in the context of diversionary war.
Indeed, an
assumption of legislative impotence has achieved the status of
conventional wisdom. According to Meernik:
The literature on U.S. foreign policymaking unambiguously
demonstrates that because of his constitutional prerogatives and
political incentives as well as congressional weaknesses in foreign
policy, it is the president who exercises supreme control over the
nation's military actions.14
Because the president also serves as commander-in-chief of the
military, Congress cannot (or will not) try to constrain the
president's freedom to pick battles, define the scope and duration
of conflict, or set the terms by which a conflict ultimately is
resolved. While Congress may direct domestic policymaking, its hold
over foreign policy is quite tenuous; and when the president decides to
exercise military force abroad, members of Congress can only complain
on Sunday morning talk shows. For the most part, the president's
authority over military matters is beyond reproach.
Consider, by way of examples, work on two of the causal mechanisms
that underlie the use-of-force literature: diversionary war and
“rally around the flag” effects. The diversionary war
hypothesis suggests that heads of state deploy troops abroad in an
effort to distract attention away from domestic strife, most commonly
economic downturns.15
Advocates of the
theory typically assume that Congress and the public are blind to a
leader's true intentions and, as a consequence, regularly accept
on faith proffered justifications for conflicts.
16 By sending troops abroad, it is
supposed, presidents can shift public attention away from a failing
economy and rally widespread support, as members of Congress
automatically fall behind their chief executive.
Congress is also largely absent from most quantitative tests for
“rally around the flag” effects.17
Congress's
stance on military ventures conducted abroad, it is assumed, does not
mediate the size or direction of changes in the president's public
approval ratings.
18
“Aggressive foreign behavior is a useful tool for dealing with
domestic political problems”—especially because domestic
political institutions, presumptively, cannot temper the
president's efforts to distract and, occasionally, deceive.
19 Quite the contrary, members of Congress are
just as susceptible to the rally phenomenon as is the general
public.
20Accounting for Congress
Presidents, to be sure, retain profound informational and tactical
advantages over Congress in formulating and implementing U.S. foreign
policy. As Peterson correctly observes, “presidents remain the
most potent political force in the making of foreign policy,”
while Congress remains “a secondary political player.”21
Nonetheless, we question the “unambiguous
demonstration” that domestic political institutions do not, or
cannot, impede the presidential use of force. Presidents cannot easily
and automatically dupe political opponents, especially when doing so
entails putting American troops in harm's way. Indeed, there are
at least three ways in which Congress can interfere with presidential
plans to deploy troops abroad; and hence, three reasons why presidents
should take Congress seriously when contemplating military action.
1. Dismantling the President's Military Venture.
Congress can actively work against the president, materially affecting
the course of a military campaign. It can refuse to appropriate needed
funds, call for the return of troops sent on ill-conceived foreign
missions, or raise concerns about the efficacy of an intervention.
Grimmett has documented numerous instances since 1970 when Congress cut
off military funding to compel the withdrawal of forces, typically
using the appropriations powers to restrict military operations.22
In their study of the War Powers Resolution,
Auerswald and Cowhey show that Congress regularly places obligations on
presidents (reporting requirements, budgetary limitations) that can
prove burdensome.
23 Having to
stave off a mobilized opposition party within Congress during the
course of a military campaign may dissuade presidents from initiating
force at all.
2. Conveying Political Resolve. Whether lower-level
military deployments can achieve strategic objectives in short order
depends, in part, on the president's ability to credibly convey
political resolve. Congress plays an important role in this regard. When
facing considerable opposition within Congress, presidents will have a
more difficult time signaling the nation's willingness to see a
military campaign to its end. As Schultz shows, an “opposition
party can undermine the credibility of some challenges by publicly
opposing them. Since this strategy threatens to increase the
probability of resistance from the rival state, it forces the
government to be more selective about making threats”—and,
concomitantly, more cautious about actually using military force.24
Domestic political strife weakens the credibility of
signals of resolve, and as such, undermines the effectiveness of those
shows of force designed to “influence … specific behavior
of individuals in another nation without engaging in a continuing
contest of violence”—the very kinds of lower-level
deployments the use-of-force literature means to explain.
25 To the extent that presidents want to avoid
protracted military entanglements abroad, they may be wary of deploying
troops when they do not have their domestic house in order.
26
3. Moving Public Opinion. In three ways, scholars
have shown that Congress influences media coverage of, and public opinion
on, the president's foreign policy initiatives. First, Brody presents
evidence that rally effects are conditional on support for the
president within Congress, especially among party leaders.27
Facing foreign crises, presidential approval
ratings spike when members of Congress stand in defense of the
president; when critical voices predominate, rally effects disappear.
Relatedly, a number of scholars have stipulated that media outlets key
off Congress when covering military deployments, a phenomenon they
label “indexing.”
28 “The spectrum of debate in the news, the
indexing hypothesis asserts, is a function of the spectrum of debate in
official Washington. If there is debate inside the American government
over U.S. policy, critical perspectives appear in the news. If
government policy has bipartisan support in Washington, however,
critical perspectives expressed outside the government are not well
reported.”
29 In the early stages of a
military deployment, therefore, Congress helps shape the scope and
content of debate covered in print and television news. Finally,
Congress may directly inform the public's understanding of past
deployments. Should a military venture go awry, or simply prove to be
more protracted than expected, the opposition party may level due
criticism at the next election; and if the opposition can speak with
one voice while holding large majorities in Congress, it may sway a
significant number of voters to vote against the president and his or
her party. Anticipating that Congress will push public opinion against
them, presidents may be less likely to engage the military, especially
on ventures that pose considerable risks.
Two features of these politics deserve emphasis. First, the
probability that Congress employs any of these tactics surely depends
on the perceived costs of a military initiative. While lower-level
deployments are unlikely to attract widespread attention, larger
initiatives (measured in terms of sheer manpower or financial
investment) stand some chance of inviting congressional opposition. To
the extent that interbranch politics surround the use of military
force, they should intensify as the scope of a military venture
escalates.30
Second, much of the action is
anticipatory. Rather than going head to head with Congress at the front
end of a military campaign, presidents anticipate what members of
Congress will do once they have had a chance to observe evidence of the
campaign's success or failure and the public's reaction to
it.
31 When Congress appears likely to extend sympathy
and support throughout the course of a military venture, presidents
should be more willing to deploy the troops; but when early
disappointments are likely to trigger a congressional requital,
presidents, from the beginning, should be wary of proceeding.
Partisan Support/Opposition
That members of Congress can work against the president does not mean
that they necessarily will. Just as they can erect roadblocks, so too
can they lend critical support to a president attempting to deal with a
foreign crisis. If they take any public stand or introduce any
substantive legislation, copartisans in Congress are likely to
demonstrate solidarity with their president, authorizing the use of
force or appropriating the funds needed to carry it out. Interference,
if it is to occur, usually originates from within the opposition
party's ranks.32
We
suggest three reasons why this is so.
1. The Content and Credibility of Signals. With
regards to the use of force, profound informational asymmetries define
relations between presidents and Congress.33
Presidents have at their disposal considerable amounts of private
information about developments abroad and their impacts on the
nation's strategic interests. Congress, meanwhile, has few formal
mechanisms for corroborating this information. When presidents make a
case for deploying troops abroad, Congress must decide whether or not
to take them at their word.
34 Information
asymmetries assuredly decline over the course of a military campaign, as
interest groups, the media, and members of Congress collect and disseminate
relevant information. But when military force is initiated (the subject of
this article), the information available to presidents from the Central
Intelligence Agency, the National Security Council, the Department of
Defense, and the State Department is unparalleled.
A number of game-theoretic models demonstrate that the
credibility of signals improves as political actors' preferences
converge.35
The first principle
of Krehbiel's information theory of congressional committees, for
instance, states that “the more extreme are the preferences of a
committee specialist relative to preferences of a nonspecialist in the
legislature, the less informative is the committee.”
36 When senders and receivers in incomplete information games
share a common worldview, the signals conveyed are especially
informative. Conversely, when preferences diverge, signals can be
indecipherable.
If presidents are to make an effective case for deploying
troops, members of Congress must believe the information presidents
transmit. Because shared ideological orientations facilitate belief,
presidents should garner trust from members of their party, and suspicion
from the opposition party. As their party grows in size and strength,
therefore, presidents ought to benefit from greater legislative discretion
(formal and otherwise) to exercise force abroad.
2. Shared Electoral Fortunes. When their electoral
fortunes are linked to the president's, congressional members of the
president's party have a vested interest in the president's
success—just as members of the opposition party do in the
president's failure. If presidential approval ratings increase
when presidents exercise force abroad, as the “rally around the
flag” literature suggests; and if high presidential approval
ratings boost members' electoral prospects, as the literature on
coattails suggests; then, all else equal, members' willingness to
grant the president broad discretion to exercise force abroad should
critically depend on the president's partisan identification.37
Presidential uses of force
redound to the electoral benefit of members of the president's
party and, by implication, to the detriment of the opposition party. As
such, members of the president's party, all else equal, ought to
actively support the president's plans to exercise force abroad,
as members of the opposition party, at a minimum, raise cautious
reservations.
3. Currying Presidential Favor. In a variety of
ways, presidents can influence the electoral prospects of members of
Congress. By visiting their districts, appearing in their television
advertisements, or introducing them to potential funders, presidents
can significantly affect members' electoral returns. Precisely
because constituents pay more attention to domestic than foreign policy
matters when deciding how to vote, members of Congress often have
little to lose, and much to gain, from actively supporting a president
interested in exercising force abroad. Of course, presidents work on
behalf of members of their own party. Republican members of Congress
rarely have anything to gain from currying the favor of Democratic
presidents; as Democratic members have little reason to go out of their
way to support Republican presidents. To the extent that broad grants
of authority in foreign affairs help members remain in the
president's good graces, members of Congress have cause to support
presidents of their party when they consider using force
abroad.
There are, then, a variety of reasons why members of the
president's party should be more likely to support the president.
But while resting on different first principles, all generate the same
prediction: congressional members of the president's party
reliably will line up behind the president during times of
international crisis, while members of the opposition party may not.
Thus discretion to exercise force should critically depend on the level
of partisan support a president enjoys within Congress.
Empirical Tests
To test whether congressional politics play a role in the
president's calculus to deploy military troops abroad, we conduct
several statistical examinations of the U.S. use of force between 1945
and 2000. Our data draw from Fordham, Fordham and Sarver, and Zelikow,
all of which update the original Blechman and Kaplan time series that
ended in 1976.38
The dependent variable is a count of the number
of times each quarter that the president initiates military force
abroad.
39 In two ways, we amend the
existing use-of-force data sets. First, we extend the time series
through 2000. To identify uses of force post-1995, we replicated the
procedures described in Fordham 1998a.
Second, we correct several mistakes in both the original
Blechman/Kaplan time series, as well as amendments made to it by
subsequent scholars. Specifically, we remove two classes of events:
those uses of force that were scheduled (for example, Operation Team
Spirit) and those carried out by troops already deployed to a crisis
area (for example, attacking Iraqi antimissile batteries after being
fired on in the no-fly zone). Note, though, that none of the findings
presented below change significantly when the 1945–95 data sets
are analyzed.
Blechman and Kaplan ranked demonstrations of force on a five-point
severity scale. Many scholars use only the most severe uses of force in
their analyses, that is, instances that involved the deployment of
nuclear capabilities or the mobilization of multiple aircraft carrier
task groups, battalions, or combat wings.40
Examples include Fordham 1998a
and 1998b. The severity of the use of force
is determined by information available at the outset of the crisis.
Thus while neither Congress nor the president know, by the eventual
severity or duration of a conflict, both have information regarding the
size and nuclear capability of the troop deployment, which
distinguishes major from minor incidents.
As previously
indicated, one would expect Congress to exert the most influence in
major uses of force. Still, we do not want to dismiss the possibility
of broader legislative involvement in foreign policy, and therefore we
estimate separate models for both major uses (1–3 on the severity
scale) and minor uses (4–5).
41 This
definition of major versus minor force is standard in the use-of-force
literature. See DeRouen 1995 and 2000; Fordham 1998a and
1998b; and Mitchell and
Moore 2002. Changing the definition of major uses of force to
include only level 1 and 2 deployments on the Blechman/Kaplan
severity scale yields positive coefficients for all versions of our
independent variable of interest, but these estimates vary in their
statistical significance. This result is not surprising given the large
number of quarters during which no level 1 or 2 deployments
occur.
Figure 1 traces use-of-force trends in U.S.
foreign policy.42
Yearly data is used in
the graphs simply for purposes of visual coherence. All analyses use
quarterly data.
Between 1945 and 2000, presidents exercised
force a total of 383 times, 141 of which qualified as major and 242
minor.
43
Quarterly U.S. uses of force, 1945–2000
Key Explanatory Variables
We construct three measures of congressional support for the
president.44
Consistent with
Edwards's observation that “members of the president's
party almost always form the core of the president's support in
Congress,” all measures of congressional support are partisan
based. Edwards 2003, 10. There is, at
present, a sizeable literature in American politics that examines the
effects of partisan divisions across the legislative and executive
branches on lawmaking. See Mayhew 1991; Coleman 1999; Howell et al.
2000; Rudalevige 2002; Binder 2003; and Lewis
2003. Concurrently, there is an ongoing debate about whether
parties represent mere proxies for members' preferences—see,
for example, Krehbiel 1993—or whether
party leaders independently influence legislative processes—see,
for example, Cox and McCubbins 1993. On this
particular issue, we remain agnostic. Given that the unilateral
presidency literature consistently finds that the partisan composition
of Congress influences executive discretionary authority—see
Howell 2003; Mayer
2001; and Krause and Cohen 2000—however, we choose in this article to
use partisan measures of congressional support.
The first,
unified government, simply indicates whether the congressional
majority party is the same as the president's party in both the
House and Senate.
45 Versions of all key
explanatory variables that focus exclusively on the Senate, which
arguably plays a disproportionate role in defining Congress's
foreign policy positions, generate comparable results.
To
obtain a more nuanced assessment of the level of congressional support
for the president, however, we also compute the average percentage of
seats held by the president's party in the House and Senate and
label this variable
percent president party. As the size of
the president's party increases within Congress, the president
should enjoy additional discretion to deploy troops abroad. Conversely,
as the opposition party increases in size, the president's ability
to use military force should attenuate.
Southern Democrats present obvious problems for partisan-based
measures of presidential support. While Democrats enjoyed large
majorities in the House and Senate in the 1960s, they also faced strong
divisions within their ranks. To address this shortcoming, Brady,
Cooper, and Hurley constructed “legislative potential for policy
change” (LPPC) scores.46
They base
LPPC scores on four factors: (1) the size of the majority party; (2)
the majority party's internal cohesiveness; (3) the size of the
minority party; and (4) its cohesiveness.
47 The LPPC score for either chamber in any given term is
calculated as follows: Chamber LPPC = [(majority party size in
percent) × (cohesion of majority party)] −
[(minority party size in percent) × (cohesion of minority
party)]. Congressional Quarterly's party unity scores are
used.
To generate our third measure of congressional support
for the president, we modify these scores only slightly, substituting
the president's and opposition parties for the majority and
minority parties, respectively. When the president's party is
relatively large and unified and confronts a relatively small and
divided opposition party, the president should be able to use force
with considerable freedom. Conversely, when the president's party
is relatively small and divided, and the opposition party is larger and
more unified, the president's freedom to use force abroad should
decline. We label this variable
president party power.
Background Controls
As previously discussed, scholars have focused almost exclusively on
other domestic and international factors that shape the
president's ability to use force abroad. In order to reduce the
chances that our findings result from omitted variable bias, we
incorporate controls for many of the alternative hypothesized
influences on the use of force. From the beginning, though, it is worth
highlighting that none of the main results presented below depend on
the inclusion of any particular set of background controls. Indeed, all
observed relationships between congressional support for the president
and use-of-force frequencies hold in models that add only presidential
fixed effects.
Consistent with a burgeoning literature on the political economy of
the use of force,48
we incorporate the
quarterly unemployment rate (
unemployment) and the consumer
price index (
cpi)—both of which were taken from the
Bureau of Labor Statistics. Past research finds that poor economic
performance is likely to create incentives for the president to act
aggressively in foreign policy affairs.
49 Because much of the literature on the use of force draws on theories
of diversionary war, we control for the president's public
approval rating (approval). The impetus for much of the
original quantitative work on the subject was Ostrom and Job's
finding that approval ratings were a highly significant determinant of
the use of force—though subsequent research has proven less
definitive on the matter. We measure the first Gallup approval rating
for the president in each quarter.50
See
Ostrom and Job 1986. In the handful of
quarters missing approval data, we fill the data by using a linear
interpolation. We also interpolated by using the last known poll. The
results are consistent regardless of the method used.
A related body of work examines whether elections usher in additional
uses of force.51
This research contends that
“rally around the flag” effects establish incentives for
presidents to use force during the months immediately preceding an
election. Thus we introduce the variable
election, which is
coded “1” during the first three quarters of a presidential
election year, and 0 otherwise.
52 Fordham's election variable differs from ours. While
we isolate the three quarters leading up to the national election,
Fordham 1998a isolates all four quarters in
an election year, as well as the first quarter of the following year
when an incumbent is reelected. Moreover, Fordham specifies two
election variables, separating peace-time from war-time elections.
Doing so makes no difference for our main findings.
The next four variables capture facets of the international
environment that may impinge on the president's autonomy in
foreign policy. Because of contemporary military commitments, there
should be a tendency for presidents to employ force for bargaining
purposes less often during times of war. We create a variable labeled
war, coded “1” during periods of international
wars in which the United States was involved (Korea, Vietnam, and the
1991 Gulf War). The Cold War was also a period of unprecedented concern
over international engagement of U.S. forces. To control for its
influence, we include the variable cold war, which is coded
“1” during the 1945–89 period.
To account for systemic forces that have been linked to the onset of
both interstate wars and disputes, we include a measure of U.S.
hegemony during the period of analysis.53
The
measure is the percentage of international military capabilities held
by the United States, and derives from the Correlates of War
Capabilities data set.
54 With hegemonic
power may come responsibilities (and incentives) to monitor and
intervene in conflicts worldwide. Thus the variable
hegemony
ought to be positively associated with the use of force. Finally, we
include a measure of the number of world military conflicts beginning
in each quarter of observation. Presumably, a higher number of world
conflicts provides more opportunities for the United States to respond
with the use of force.
55 The variable
world
disputes aggregates non-U.S. militarized interstate disputes
(MIDS) over each period of observation.
56Results
The use of force, then, is assumed to be a function of the partisan
composition of Congress, presidential fixed effects, and background
controls. We estimate the following event count model:57
Autocorrelation function (ACF) plots reveal
negligible temporal dynamics. Three of the first four autocorrelations
of the major and minor time series are statistically significant, but
the largest value of rho observed is only 0.19. Estimates obtained from
general estimation equation (GEE) models that correct for an AR(1)
process yield virtually identical results to traditional event count
models. On GEE models, see Zorn 2001. (The
ACF plots and results from GEE models are available from the authors on
request.) For the sake of simplicity, we report estimates from negative
binomial regressions with robust standard errors that account for
clustering on each president. We use negative binomial models because
Poisson models assume that events within an observation period are
independent. King 1998, 50–51. In this
case, the presence of military action may spur further military action
in close proximity. Negative binomial models do not assume
independence.
If our expectations about the relevance of Congress hold,
β1 should be positive and statistically significant.
Descriptive statistics for all variables are reported in the
Appendix.
Tables 1 and 2
present the estimated regression results for each measure of
congressional support—unified government, percent
president party, and president party power—on minor
and major uses of force, respectively. The results are strikingly
consistent. When minor uses of force serve as the dependent variable,
none of the key explanatory variables are statistically significant.
Major uses of force, however, are affected by domestic political
factors, as the partisan composition of Congress appears to play a
significant role in shaping the president's willingness to deploy
troops abroad.58
Models that pool major
and minor uses of force never generate statistically significant
impacts for congressional support for the president.
With
regards to larger-scale deployments, significant effects are
consistently observed; the positive coefficients suggest that as the
size of the president's party grows, the president's
proclivity to use the military for significant purposes also
increases.
Minor uses of force: 1945–2000
Major uses of force: 1945–2000
Although theory suggests that the size of the president's party
affects the discretion of the president to use force abroad, it is
largely silent with regard to the functional form of the relationship.
There may be good reason, however, to expect the impact of percent
president party and president party power to be
nonlinear. Incremental changes at the tails of the distribution may not
have an appreciable impact on the frequency with which presidents
exercise force abroad. Shifts around the center of the distribution,
meanwhile, may induce large changes in the use of force. To test for
the possibility of nonlinear effects, we take the logistic
transformations of percent president party and president
party power (noted in each table with an “e” prefix)
and reestimate the models.
No matter which functional form is used, every measure of
congressional support for the president serves as a significant
predictor of the quarterly number of major military deployments.
Presidents exercise major force roughly 45 percent more often during
periods of unified government than during periods of divided
government. A one standard deviation increase in the size of the
president's party in the Congress (from a mean of slightly greater
than 50 percent to 58 percent) corresponds with a 17 percent increase
in the use of force by the United States. Conversely, while a one
standard deviation drop leads to a 16 percent decline in the number of
uses of force. When shifting to its maximum value of percent
president party (68 percent), the model predicts a 36 percent
increase in the incident rate.
Similar results arise when using the president party power
scores. A one standard deviation increase in the relative size and
cohesiveness of the president's party yields a predicted 19
percent increase in the number of uses of force, while evaluating the
variable at its maximum generates a predicted increase of 39 percent.
The level of partisan support for the president within Congress, it
seems, stands out as a major determinant of the U.S. propensity to use
force abroad. That congressional effects uniquely apply to major uses
of force, further, corroborates Auerswald and Cowhey's observation
that:
Congress has acquiesced to the use of force so long as the
administration's action is swift or small scale. Such deployments
raise little likelihood of future Vietnam-style conflicts and do not
necessitate congressional action. Without exception, however, long-term
conflicts were met with either congressional threats or legislative
action with the potential to start the war powers clock.59
The distinction between major and minor uses of force, then, is
important, as the president's inclination to consider
congressional interests appears to depend upon the size and expected
costs of a military venture.
For the most part, the remaining model estimates are consistent with
previous findings. For instance, estimates in Tables 1 and 2 for unemployment
mirror the original Ostrom and Job findings that higher levels of
“economic misery” correspond with higher levels of military
activity.60
Inflation rates also
contribute positively to minor deployments; although they do not
influence major uses of force.
Similar to those scholars who do not differentiate war-time and
peace-time elections, we find no statistical relationship between the
election cycle and the use of force.61
In
addition, the approval variable never attains statistical significance
in our estimates. This is consistent with the findings of both Meernik
and Fordham, but counters Ostrom and Job's claim that declining
approval ratings posit incentives for presidents to use force.
Other variables are of the predicted sign. During the Cold War, the
expected number of uses of force more than doubles. Ongoing wars
depress the propensity of presidents to deploy U.S. troops, yet their
statistical significance varies based on whether minor or major uses of
force serve as the dependent variable. The level of military hostility
in the world and the level of U.S. hegemony, however, appear to have no
consistent or statistically significant influence on the propensity of
the United States to engage the military.
Robustness Checks
Do congressional effects hold for the entire post–World War II
era? A number of scholars have observed that during the 1950s and early
1960s, a broad consensus about national security policies and
priorities transcended party lines in American politics; the Vietnam
War, however, reestablished partisan cleavages with regards to foreign
policy.62
If true, then the partisan composition of
Congress should profoundly affect the presidential use of force during
the post-Vietnam years, but have a negligible impact before.
Table 3 examines the effects of percent
president party and president party power before and
after the War Powers Resolution (November 1973).63
Estimates are not sensitive to this choice of year. When
using adjacent years, virtually identical results are
observed.
This resolution makes for a useful dividing line, as
it stood as the centerpiece of the congressional resurgence of the
early 1970s and, according to many scholars, marked Congress's
reassertion of its constitutional prerogatives over foreign
policymaking.
64 If the War Powers
Resolution reestablished Congress's rightful place in
deliberations over the use of force, then congressional effects may be
concentrated in the post-1974 period. One mitigating factor, however,
is worth noting. By establishing strict reporting requirements, the
resolution may have reduced the informational asymmetries between the
executive and legislative branches. As we note in the second section
above, copartisans within Congress are especially likely to support the
president because, in part, they can better trust his assessment of a
military intervention's efficacy. If the resolution achieved its
intended objectives—a subject of considerable controversy; see
Fisher 1995, 2000— then information may replace trust, and
the importance of partisanship may decline.
Still, we do not
find any evidence that Congress's impact on the quarterly number
of major uses of force initiated by the president is confined to the
second half of the postwar era. Quite the contrary, whether using
percent president party or
president party power,
impacts appear more pronounced between World War II and the advent of
the War Powers Resolution. We cannot reject the null hypothesis that
the slopes for the two periods are equivalent.
65 Without taking logistic transformations, significant
impacts are not observed between 1974 and 2000. These null findings are
due to Carter, who exercised force relatively infrequently even though
he enjoyed large Democratic majorities in Congress. Taking the logistic
transformations of the key explanatory variables effectively attenuates
the influence of this outlier, which leverages considerable influence
on the shortened time series.
Major uses of force, with period effects: 1945–2000
Another concern is that these findings are peculiar to the particular
data set analyzed. Virtually all quantitative work on the use of force
in U.S. foreign policy relies on the original Blechman and Kaplan data.
Indeed, these data were designed especially for this kind of
investigation, and various updates have kept them current.
Traditionally, however, cross-national studies of military conflict use
the Militarized Interstate Dispute Data (MIDS).66
Gowa also uses the MIDS data in her contribution
to the use-of-force literature.
67 We prefer the Blechman and Kaplan data over MIDS. Employing the
Blechman/Kaplan data facilitates comparisons with a larger body of
past statistical work on the use of force. In addition, as Fordham and
Sarver have shown, the MIDS data overlook many uses of force included
in the Blechman/Kaplan data.68
Still,
Gowa's rejection of the idea that partisan politics play a role in
decisions to use force abroad was based on analyses of MIDS data.
Finding an insignificant effect for divided government on MIDS counts
for the period between 1870 and 1992,
69 Taking the use-of-force time series back to 1870 would
appear to be a strength of Gowa's analysis. Unfortunately, her
statistical models do little to account for the marked changes in
presidential power, presidential-congressional relations, and the
nation's standing in the international order before and after the
administration of Franklin D. Roosevelt, even though she recognizes the
marked increase in the number of MIDS observed in the modern era. Gowa 1998, 318. Moreover, as Fordham has pointed
out, Gowa fails to account for temporal dynamics in her dependent
variable that, uncorrected, potentially generate spurious results.
Fordham 2002.
Gowa concluded that
with regards to the use of military force, “partisan politics has
… remained muted,” indicating what she deemed “a
subversion of the checks-and-balances system.”
70 When extending the MIDS time series through 2000 and using more
informative measures of partisan dynamics, very different findings
emerge. As the results in Table 4 indicate,
unified government, percent president party, and
president party power all serve as significant predictors of
the frequency of MIDS involving the United States; only the logistic
transformation of president party power, though positive,
generates an estimated coefficient that falls below conventional levels
of statistical significance.71
When
examining the 1945–92 period, we replicate Gowa's null
finding on unified government. Effects for percent
president party and president party power, however,
remain positive and statistically significant.
During periods
of unified government, military interventions increase by nearly 25
percent; and a one standard deviation jump from the mean values of
percent president party and
president party power,
meanwhile, generates a 19 percent and 13 percent increase in military
involvements, respectively. If these relationships measure the health
of the U.S. system of separated powers, as Gowa suggests, then there is
more cause for optimism than previously recognized.
U.S. militarized interstate disputes (MIDS): 1945–2000
Conclusion
For too long, literature on the use of force has ignored domestic
political institutions. Indeed, most research equates domestic politics
in the United States with election cycles, public opinion, and the
state of the economy, and nothing more. Scholars have assumed that
because it is not an integral component of the domestic politics of
international engagement, Congress can effectively be ignored. Finding
that presidential approval ratings are unrelated to the use of force
between 1953 and 1978, Moore and Lanoue offer the sweeping conclusion
that “international politics, not domestic politics, [must
be] the primary determinant of conflictual U.S. foreign policy
behavior.”72
We argue otherwise and show empirically that Congress critically
affects presidents' proclivity to use military force. When we
isolate minor deployments of troops, we do not observe any correlation
between the composition of Congress and the frequency with which
presidents generally use force abroad. But when presidents consider a
major use of force abroad, the level of partisan support within the
Congress appears to matter greatly. As their party's share of the
Congress increases, presidents conduct major military campaigns abroad
with increasing frequency. Conversely, as the size of the opposition
party increases, presidents act with major force less and less often.
This finding is robust across various model specifications, data sets,
and time periods, and using multiple measures of congressional
support.
These findings clash with existing statistical studies of the use of
force in two ways. First, few studies control for potential
congressional involvement in foreign policy, but rather assume away its
importance. The few models that incorporate measures of divided
government do not find substantial effects. Second, when incorporating
the influence of Congress in shaping use-of-force dynamics, some
traditionally important variables (for example, public opinion and
inflation rates) cease to hold much explanatory power.
Important questions, nonetheless, remain unanswered. That the use of
force covaries with the partisan composition of Congress is consistent
with multiple micro- explanations of executive-legislative relations.
Copartisans within Congress have strong incentives to support their
president for reasons having to do with interbranch signaling, shared
electoral fortunes, and the benefits of currying presidential favor.
Meanwhile, because of the importance of staying the course in ongoing
military ventures, conveying political resolve abroad, and sustaining
public support, presidents have good cause to be wary of strong
partisan opposition within Congress. Nothing in the data presented here
distinguishes among these explanations. Answers, nonetheless, are
critical if one hopes to further disentangle the domestic institutional
politics of military action.
The proper question should no longer concern whether domestic
political institutions matter. Rather, the challenge is to specify when
they matter, exactly how their constraining influence manifests, and
what this entails for the ultimate conduct of U.S. foreign relations.
Scholars would do well to study the particular causal mechanisms that
allow Congress to influence foreign policy. Moreover, these mechanisms
may influence much more than just the frequency with which presidents
initiate force—they may critically affect the choice of foreign
crises presidents respond to; the timing, scope, and duration of the
response; and the terms of the conflict's ultimate resolution.
Future research, we hope, will further explore the institutional
politics surrounding the use of force. For while presidents certainly
command considerable discretion to deploy troops when, and as, they
please, Congress deserves greater recognition that is has heretofore
received.
