University of Utah
Rooting Out Waste, Fraud, and Abuse: The Politics of House Committee Investigations, 1947
to 2004
Author(s): David C.W. Parker and Matthew Dull
Source: Political Research Quarterly, Vol. 66, No. 3 (SEPTEMBER 2013), pp. 630-644
Published by: Sage Publications, Inc. on behalf of the University of Utah
Stable URL: http://www.jstor.org/stable/23563170 .
Accessed: 11/01/2015 04:29
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
.
Sage Publications, Inc. and University of Utah are collaborating with JSTOR to digitize, preserve and extend
access to Political Research Quarterly.
http://www.jstor.org
This content downloaded from 140.232.1.111 on Sun, 11 Jan 2015 04:29:52 AM
All use subject to JSTOR Terms and Conditions
RegularArticle
Out
Rooting
The
Waste,
of House
Politics
C.W.
Abuse:
Committee
sagepub.com/journalsPermissions.nav
DOI: 10.1177/1065912912459566
prq.sagepub.com
1947 to 2004
Investigations,
David
and
Fraud,
Political Research Quarterly
66(3) 630-644
© 2012 Universityof Utah
Reprintsand permissions:
Parker1
and
dSAGE
Dull2
Matthew
Abstract
Scholars
have
oversight
majorities,
executive
branch
characterized
and
malfeasance
by broad
government
contingent
theories
dampen
theory.
Congress'
together
1947
discretion
statutory
or the
government
oversight
of delegation
between
and
faced
show
We
how
structure
why, and
government,
of oversight
by individual
how
long
distributive
investigated
and
committees
committees
on more
than
to
committees,
Congress
committees,
partisan
depends
members
attention
congressional
of congressional
for
whereas
investigations,
conduct
varied
explain
the
when,
Divided
more
vigor.The
structures
We
explain
2004.
generate
investigatory
incentive
in oversight.
disinterest
congressional
the
by advancing
partisan
good
bemoaned
long
and
a desire
unified
to produce
of Congress.
Keywords
oversight,
Congress,
divided
president,
committees
government,
government, or constitutional design, contingent over
sight theory shows how party majorities, committee
Introduction
Committee
have
hearings
sional
power
ment,
but the pursuit
been
from the early
of congres-
instruments
decades
of effective
of American
and efficient
structures,
and
in determining
govern-
"watchful
government
when
eye"
discretion
statutory
and
are far more
for how
on the executive
long
important
turns
Congress
its
branch,
is only one incentive for Congress and its members to
oversee the executive branch.1 The House Committee on
Sustained periods of divided party rule coupled with
intensifyingpartisan and institutional conflict have raised
Un-American
the
nist
Activities
infiltration
hearings
in the late
1940s
commu-
investigating
are
attributed
variously
to legitimate concerns about Soviet espionage, partisaninspired Republican attacks on a Democratic administration, and the political ambitions of committee members.
Sixty
years
Committee
later,
Government
Chairman
Henry
the
scrutinizing
of military
use
and
Oversight
Waxman's
(D-CA)
contractors
Reform
were
attrib-
contingent model of oversight politics, by which we
that the exercise
of congressional
oversight
between
Subsequent
uted at once to the public interest, partisanship, and personal ambition (Weisman 2007). We explain the decision
to launch and sustain congressional investigations with a
mean
have
varies
activities
extra-legislative
become
venues
for political
such
combat
as
(e.g.,
Ginsberg and Shefter 2002). David R. Mayhew's (1991)
empirical challenge to this perspective inspired scholar
ship refining how political scientists think about the inter
play
hearings
that
possibility
oversight
divided
partisan
research
conflict
and
examining
shows
government
institutional
the
congressional
politics.2
of
consequences
investigations
correlate with partisan institutional conflict, contingent on
factors such as party strength,variations between House
and Senate rules, and shifting institutional arrangements
(Kriner and Schwartz 2008; Parker and Dull 2009). We
extend
this research
by demonstrating
that partisan
institu
jointly with institutional-, committee-, and individuallevel dynamics. We study the frequency and duration of
tional conflict is not only correlated with investigations
"fire
'Montana State University, Bozeman, USA
Virginia Tech University, Alexandria, USA
alarm"
executive
style
branch
committee
investigations
misdeeds
between
1947
of
and
alleged
2004,
organizing the analysis around three sets of dynamics:
institutions,
theories
function
committees,
positing
and
reputations.
'
that congressional
of member
incentives,
oversight
the
pursuit
Contrary J
to
is mostly
a
of efficient
Author.
Corresponding
.
.,
David C.W. Parker, Department of Political Science, Montana
State
University, 2-143 Wilson Hall, Bozeman, MT 59717, USA.
[email protected]
This content downloaded from 140.232.1.111 on Sun, 11 Jan 2015 04:29:52 AM
All use subject to JSTOR Terms and Conditions
631
Parkerand Dull
over
from
time,
reveal
one
but also
another,
how
or presidential
congressional
with
particular
"cartel"
partisan
term
of committees.
types
committees
style
to
We
are
espe-
cially sensitive to the presence of divided government in
the decision to investigate the executive branch for accuCongressional oversightis also contingenton committeeand member-level dynamics in the context of delegated
bureaucratic discretion. Contemporary scholarship on conoversight
gressional
around
dynamics
efforts
organizes
two
distinct
to
these
explain
The
purposes.
first purpose
is publicity. Committee hearings create particularly prominent
venues
for performance
and
which
position-taking,
shapes individual and partisan brands. The second purpose is monitoring. Oversight is an ex post instrument of
bureaucratic control, an institutional response to asyminformation
metric
between
and
principals
agents
(which is a problem inherent in all formal delegations of
authority; Miller 2005). These two purposes, by-products
of
constitutional
American
exist
design,
in
tension.
Although Woodrow Wilson (1885, 303) writes that "the
informingfunctionof Congress should be preferredeven to
its legislative function," publicity from that monitoring
can create a considerable threatto state building. The policies and capacity created by legislating provide oversight
opportunities where the goal can be improving the administrative state or contesting state expansion and explora-
tion of its retrenchment. In either case, the publicity
has
generated
individual
for
consequences
that may
reputations
party,
and
committee,
encourage
or not aggres-
sive monitoring of agency compliance.
We find that Congress has not abdicated oversight, but
its motivations for oversight are not driven by a desire to
produce efficient and effective governance alone.
form
in the
oversight
Congressional
of fire-alarm-style
of better
and
member
public
the
agendas,
constitutional
prerogatives,
administrative
state's
protection
the
and
and
capacity
rationale
government"
the advancement
policy,
is but one
of the
partisan
motivation
As
of
the
the "good
for initiating
and continuing investigations of executive branch malfeawe
sanee,
ings
evaluate
by considering
the broader
the recent
of our
implications
lament
with
find-
congressional
abdication of oversight responsibilities.
Congressional
and the Powers
have long bemoaned lack of congressional
oversight (Ogul 1976; Scher 1963; but see Aberbach
2002). One theoretical response is the congressional
dominance literature, which suggests that Congress preScholars
fers
ex
ante
controls
over
agencies
to active
oversight
of ex ante
administrative
particularly
in asserting
expense
control
ex post
oversight
through
hearings, passing new legislation, or reducing agency
appropriations. Instead, enacting coalitions write into ini
tial legislation procedures requiring the agency to seek
comment
public
on proposed
rules,
information
or report
interests,
consult
with
specified
the development
regarding
of mies prior to implementation. McCubbins, Noll, and
(1999)
Weingast
uses
that Congress
argue
such
procedures
to hardwire the preference of the enacting coalition,
"stacking the deck" to enhance the influence of favored
interests and increasing congressional and judicial influ
ence
over
implementation.
policy
conserves
Congress
scarce oversight resources and, by formulizing procedural
mies and setting the groundwork forjudicial intervention,
administrative
the
increase
procedures
costs
of bureau
cratic drift.One common critique of the "deck stacking"
thesis is the assumption of a single, unitary principal
(Ferejohn and Shipan 1990; J. Q. Wilson 1989).
Suggesting the work of positive theorists over the next
decade, Morris Ogul (1976) observed "latent" congressio
nal influence in venues where Congress had been critiqued
for its lack of activity.Reformulating Ogul's "latent" over
sight, McCubbins and Schwartz (1984) set out the now
classic
distinction
"fire alarm"
between
Police
oversight.
patrol
takes
oversight
and
"police
patrol"
the form of Congress
walking an agency beat, sampling agency procedures and
actions "with the aim of detecting and remedying any viola
tions of agency goals and, by its surveillance, discouraging
such violations" (McCubbins and Schwartz 1984,166). By
fire-alarm
and
is intermittent
oversight
reactive,
"Instead of sniffing for fires," McCubbins and Schwartz
explain,
"Congress
builds
(1984,
166).
fire-alarm
places
neighborhood
its own
patches
fire-alarm
boxes
fire houses,
and
that members
argue
because
oversight
street
cor
dis
to an alarm"
in response
hook-and-ladder
They
on
sometimes
of Congress
the opportunity
costs
prefer
of police
patrol are high, police patrols can miss violations, and fire
alarms
efficiently
the time
place
of members
the costs
is scarce.
of patrols
Congress
on others
does
when
not abdicate
its ex post managerial responsibilities when institutingex
ante
.
Oversight
^
of Investigation
in the con
is the modeling
discretion,
agency
innovation
significant
literature
procedures. McCubbins, Noll, and Weingast (1987, 1989,
1999) argue that Congress and its members face difficulty
ners,
legislature's
contestation
direction.
on
contrast,
investigations serves a variety of purposes, including the
creation
A
1988).
dominance
controls
and
of wrongdoing.
sations
Bendor
(see
gressional
controls
on
agencies
legislatively;
rather,
Congress
only seems to ignore oversight because it prefersto respond
to fire alarms instead of initiatingpolice patrols
Even if one accepts the argument that Congress prefers
fire alarms over police patrols (there is evidence that
Congress engages in police patrols—see Aberbach 2002;
Baila and Deering 2001), this picture of oversight and
investigation based largely on an individual opportunity
costs
of member
or committee
This content downloaded from 140.232.1.111 on Sun, 11 Jan 2015 04:29:52 AM
All use subject to JSTOR Terms and Conditions
incentives
is incomplete.
PoliticalResearch Quarterly66(3)
632
The
of
willingness
is
branch
costs
by the institution,
the
groups—they
provide
opportunity
to individual
members
majorities,
individual
of
party
hence,
committees, and individual members together. These
costs
opportunity
the
competing
and
are shaped
pressures
desire
to produce
a normative
Ambition
govemment.
because
need
investigations
consuming,
are
taking
by
not
which
effective
fires
in an allegedly
ambition
sive
bureaucracy
for
nity
cost
and
committee
benefits
clear.
moments
precious
and
counter
and
always
are
They
from
away
the
timepoten-
tially more valuable credit-claiming activities. They can
bolster
an
an express
'
,
do
individual
member's
tarnish
but they may
brand,
mandate
tu
These
not.
or
them.
to launch
,
the
have
while
others
opportunity
costs
investigations,
and
■
suggest that where, why, and when Congress chooses to
engage in police patrols or respond to fire alarms is not
motivated just by some objective desire for good government
or
to
balance
the
of
needs
competing
individual
and
sometimes
competing
costs
opportunity
affect
the willingness and ability to initiate and sustain an investigation of the executive branch. These opportunity costs
are a function
level
of institutional-,
dynamics,
the foundation
three broad
of contingent
oversight
individual-
and
committee-,
and we make
claims
theory:
act as agents
Executives
of Congress,
tions
and
goals.
Congress
the transaction
than
lower
the benefits
Members
Reputation-.
tion based
_
Committee-Based
gressional
of Congress
and
majorities
if their
partisan
if engaging
1993).
branch
will
in
engage
can
reputation
in oversight
be
tarnish
may
the
party
vigorously.
Committee: The type of benefits obtained by indimembers
from
committee
To
test
the
effects
of these
various
dynamics
violating
makes
congressional
initial
the
be
motivated
an
by
charge:
external
Baila
wrongdoing.
Like
Congressional
Information
goals"
or
Congress
LexisNexis
and
of
use
Index
(CIS)
a measure
House
by
or charge
we
Deering,
Service
to develop
initiated
event
the
available
of fire-alarm
committees
between
1947 and 2004.3 Unlike Baila and Deering, we focus on
charge
from
emanating
David
using
R.
"exposure"
which
definition
Mayhew's
of
a
as a guide,
probe
sweep
the congressional
of the CIS
Index
resulted
were
coded
based
then
in a large
on the fol
lowing decision rule: A hearing is included if the CIS Index
record
includes
ment,
or abuse
or more
reference
federal
out
to violation
of discretion
mismanage
or past
or an organization
agency
functions.
agency
of law,
in current
Violation
conduct
tasked
of law
in
with
refers
to
activities that contradict formal discretion as interpretedby
Congress, the judiciary, a specified investigator, or a hear
ing witness. Mismanagement is included if reference is
made
assignments
affect the propensity to engage in oversight.
not
who
to be
carrying
brand, oversight is either ignored or pursued less
vidual
"is
seems
one
enhanced;
Alarms
the fire alarms established by Congress? According to
Baila and Deering (2001), a fire-alarm investigation
list of hearings,
oversight
of
and Schwartz 1984, 166). A key distinction
An electronic
Con-
proclivities
(McCubbins
congressional
and partisan
McCubbins
or
How does one distinguish fire-alarm investigations from
police patrols? The objective of both is to make sure the
committee,
for reelec-
reputation,
dampening
,
of Fire
Investigations
1999).
run
as
credit-claiming
_
#
the substantive
by agency
O'Halloran
on their individual
(Cox
reputations
if
are
by Congress
to be obtained
and
(Epstein
compliance
in oversight
engages
to be paid
costs
seen
foundation,
theoretical
investigations
but may exhibit drift from congressional inten-
natural
abu
members, this is powerful evidence that our claims rest
through
Institutional.
the
institutional,
be
put out
and
If the opportu
advantage.
by
can
"No
eagerness
to and
unresponsive,
to our
measured
respond;
The
235).
to respond
ill-managed,
works
they
"ever lost votes by
(1989,
incentives
constraining
should
as
serving
when
test of our theory.
want
may
structure
executive
members. The contingent oversight theory recognizes that
various
members
on a strong
electoral
party's
committees
Some
incentives
complex
the bureaucracy"
denouncing
with
efficient
costs
politician"
benefits
credit-claiming
writes J. Q. Wilson
individual
members,
most
of Congress
a conservative
they present
than the Constitution,
not always
yield
which
investigators,
by more
faced
the
executive
the
oversee
function
a
fundamentally
faced
to
Congress
to specific
added
costs,
delay,
poor
planning
or anal
ysis, inconsistency, improper influence, or negligence in
the administration of agency mandates. This includes inad
on
the
equacies
in agency
and
procurement
contract
management
opportunity cost structure facing Congress and its willingness to engage in oversight of the executive branch,
we examine one kind of oversight: McCubbins and
or any of the above resulting from deficiencies in intergov
emmental coordination. Mismanagement is not included if
it is attributedto the structureof formal discretion or if the
Schwartz's
reference
fire alarms.
These
of the execu-
investigations
is
such
ambiguous,
as
a
reference
only
to
tive branch are not as costly to initiate when compared
"improving efficiency." Abuse of discretion refers to spe
with
cific
ated
police
by
patrols,
alarms
and—because
pulled
from
they
constituents
are
often
and
generinterest
accusations
scrutiny
engaged
that
the
agency
in improper,
covert,
This content downloaded from 140.232.1.111 on Sun, 11 Jan 2015 04:29:52 AM
All use subject to JSTOR Terms and Conditions
or
organization
or deceitful
under
activities
Parker
Dull
and
633
in defiance the congressional wishes—even
violation
of law
Each
was
investigation
committee.
priate
if a specific
is not alleged.4
with
638
yields
the appro-
committee-based
investigations undertaken by the House of Representatives
between 1947 and 2004. Any investigations undertaken
by select committees (e.g., Aging) or those committees
without clear executive oversight responsibilities (e.g.,
House
Administration
are
Committee)
excluded.5
example, in 2004, the House Financial Services committee conducted a daylong hearing titled "Allegations of
and
Failure
Management
at
Fannie
for "waste,
gram
seven
data,
or abuse
We
and
fraud,
House
contain
abuse"
hearings
at least
for three
from
captured
one
of the terms
days.
2004
waste,
draw
platforms
Out
for member
two
develop
committee.
Hearing
committee
measures
and
cal,
partisan,
the
costs
the
investigating
spends
whereas
and
number
charges
Hearing
institutional
Pages
it goes
or
off. It also
can
patrols
of
Congress
the
oversee
1974)
and
a committee
as
a platform
for
on
depends
the
chair's
thus
distance,
the
political
more
likely
the executive
investigate
The
leadership.7
the
chair
will
greater
vigorously
branch,
inten-
or intensely
be lazy
or it can
can
be ignored
be
while
focused
the building
a similar
effect
.
.
Politics.
Committee
Information,
and
on investigatory
shapes
face
when
branch
affairs.
at the hint
burns
of
down.
intensity.
Reputation,
reputations—of parties, of committee chairmen, and of
also
important.
Investigative
hearings provide committee chairmen and congressional
with
venues
to build
their
reputations
President
tance
longer
branch
investigations
waste
and
the
execu
alleging
malfeasance,
by contrast-
ing the failings of the executive branch with their own
which
Distance,
Ideological
between
is the absolute
chair's
the committee
DW-NOMINATE
and
dis
the president's
in that Congress.
score
There is good reason to presume that the autonomy of
committee chairmen in launching investigations is not as
as might
standing.
the role
be supposed,
Over
decades,
of
has
congressional
notwith
Dingell
a robust
committees
theories
competing
Congressman
the last two
of legislative
literature
on
around
developed
as
organization
a
solution to the collective dilemma rooted in the individual
choices
of goal-oriented
mittees
serve
members
.
Exchange
members—are
between
Ideological conflict between the committee chair and
the president is operationalized with the variable Chair
great
The
Why does Congress choose to initiate a fire-alarm-style
investigation? To help make good public policy, to nip
bureaucratic drift in the bud, and to investigate charges
of wrongdoing while responding favorably to interest
group allies are certainly key motivational factors. But
committee
and
more
tive
The same factors affecting the propensity to investigate
have
conflict
Ideological
committee chair and the president will generate
of days
of the response.
immediate
1:
Hypothesis
is the total
executive
the vigor
shapes
to a fire alarm
response
parties
(Mayhew
investigation
to the agency's
and
(e.g., Ogul 1976; J. Q. Wilson 1989). This alters the propensity to patrol agencies or to even respond to a firealarm
should
also
ideological proximity to the presidential administration
of executive
structure
opportunity
members
individual
to investigate
deciding
smoke
an
prolong
the
of investigatory
is simply
Days
malfeasance,
Police
are
sometimes to the great discomfort of testifying agency
administrators. If the committee's agenda-setting prerog
atives rest largely with the chair, the willingness to initi
of
page count of published hearings associated with investigâtions undertaken by each committee. The larger politi-
when
by
hearings
Hearings
in its title.6
also
branch
fire-alarm
attention.
position-taking
of using
example
sity: hearing days and number of pages produced by each
each
because
outside
reputation building. As chair of a powerful committee,
Dingell is described by David King (1997,49) as a "clas
sic policy entrepreneur" who used committee hearings to
exert and expand his political and policy influence—
in our
fraud,
more
"strategic disagreement" (Gilmour 1995). Investigations
may yield new information with the potential to disrupt
stable policy subsystems (Jones and Baumgartner 2005).
Representative John Dingell (D-MI) offers a quintes
ate
Mae,"
whereas Energy and Commerce examined the E-Rate prothe seventeen
definition
sential
How do we know that our investigations capture firealarm-style investigations by Congress? First, the patterns
in our data match those identified by Baila and Deering.
Second, the titles suggest discrete hearings initiated to
investigate a specific violation of congressional intent.For
Accounting
is magnified
reputations
then associated
This
commitment to good government. This ability to build
depends
members
individual
the
How
of Congress.
and
collective
framework.
on the particular
com
interests
of
of the
Each
threepredominant theories—gains fromexchange (Shepsle
an(l Wcingast 1987), informational (Gilligan and Krehbiel
1994; Krehbiel 1991), and parties as cartels (Cox and
McCubbins 1993)—has been generally applied to under
stand the generation of legislation and the final disposition
of bills. The theories also point to differentpropensities for
initiating investigations.
Gains from exchange theories emphasize that members
of Congress
pursue
a diverse
range
of ambitions
and
inter
ests, many derived from their constituencies. Legislative
influence
is allocated
according
vidual legislators (Gilligan
This content downloaded from 140.232.1.111 on Sun, 11 Jan 2015 04:29:52 AM
All use subject to JSTOR Terms and Conditions
to the preference
and Krehbiel
of indi
1994).
The
PoliticalResearch Quarterly66(3)
634
rationale for committees, in sum, is to serve the individual
of members.
needs
Committees
to the chamber
relative
in
geneous
their
are
and
median,
preference
are
tect party
homo-
To
relatively
the
therefore,
preferences;
the committee
words,
system
as a logroll-
operates
reputations?
evaluate
the
effect
to investigate,
decision
ideological
variance among such committees is low relative to the
chamber as a whole (Krehbiel 1990). Committees also
retain special influence in their particular policy area. In
other
seek to provide distributional goods, information, or pro
outliers
of committee
committees
incentives
must
the
on
be classified
prop
erly.First, a key consideration is committee membership. Is
the committee designed to represent individual member,
chamber, or party interests?In rational choice parlance, the
committee
is an agent
of which
commit
Second,
principal?
ing arrangement (Fiorina 1987).
The informational perspective holds that committees
overcome informational uncertainty in policy production.
Gilligan and Krehbiel (1990) note that legislators lack
tees produce benefits and policies that have "external
effects" forthe membership. Some policies have only a nar
row effecton a few congressional districts:thinkagricultural
certain
have
about
information
the
outcome
prospective
of
subsidies. Other policies affect the members broadly and
for the party
consequences
reputation:
think
tax
and
selecting a particular policy and that legislators are risk
averse, preferringpolicies that have a high degree of out-
fiscal policy. Cox and McCubbins (1993) define committees
by the externalitiesthey produce. Uniform committees pro
come
duce
(Krehbiel
certainty
consisting
Because
1991).
of preference
outliers
committees
few net informa-
provide
informational
with
should
uncertainly
outliers.
preference
few
generate
"The
novel
commit-
that
"authorize
affect nearly
every
on
'projects'
member
a national
and congressional
district(Cox and McCubbins 1993, 192). Examples include
Veterans
of the
insight
which
scale,"
tional benefits, a legislature organized around reducing
tees
externalities
Affairs,
mittees
produce
Appropriations,
and
a narrow
of policies
subset
Rules.
corn
Targeted
and
programs
informational approach," observe Shepsle and Weingast
(1995, 15), is that "institutional arrangements may reflect
the need to acquire and disseminate information in addition to (or instead of) the need to solve distributional
affectingonly a select subset of congressional districtsand
members. Agriculture is the most obvious example, but the
issues." Committees distribute expertise, encouraging the
of policies, which have both particularistic and universal
of members.
specialization
In return,
committees
receive
Cartel
to
theories
and
protect
brand.
claim
Powerful
enhance
the
are used
party's
such
agenda,
by parand
reputation
committees
gatekeeping
the party's
advancing
that committees
critical
as Ways
and
to
Means
and Budget, are stocked with party loyalists reflecting the
party's
chamber
median.
control
in committees
In
and
the
on
arena,
legislative
the
floor
enables
party
the pas-
sage of bills that define partisan priorities, build a reputation for effective governance, and draw distinctions with
partisan opponents (Cox and McCubbins 1993, 269).
This
structure
members
the
protects
of the
majority
party's
agenda
from
party
and
of the chamber,
and
the advancement
on
the
of the party
brand—which are central considerations in the production of legislation—are all key to the decision to launch
and sustain an investigation. The process of investigation
can variously disrupt the flow of distributive policies,
provide information to guide the chamber in making better public policy, or hurt/help the majority party craft a
for good
reputation
fraud,
and
abuse
governance
by the exposure
in the executive
branch.
termed
as
mixed
committees.
Armed
Services,
Judiciary,and Foreign Affairs are all examples of this type
of committee.
To
determine
ber, party,
of externality
uniform.
whether
or member
a committee
we
Second,
DW-NOMINATE
committee
ence
a committee
first what
Poole
and
type
mixed,
Rosenthal's
of waste,
Critical
to the
in each
tests
Wilcoxon's
Congress.
determine
whether
a party
differ
or cham
ber median is significantly differentfrom the committee
median (see Cox and McCubbins 1993). This method
yields the following committee variables:
Cartel
Committee
takes
a value
of
1 if they
pro
duce uniform externalities with no significant
difference between the committee's majority
party median and the chamber's majority party
median.8
Distributive Committee takes a value of 1 if they
produce targeted externalities with the commit
tee's median significantly different from the
chamber's median.
Informational Committee takes a value of 1 if they
produce mixed externalities with the committee
median
mirroring
Miscellaneous
the chamber's
Committee
met
none
median,
of the criteria
decision to investigate, then, is the fundamental purpose
outlined for informational, distributive, or cartel
of
committees.
each
congressional
committee.
Does
the
committee
or
(1997)
scores to calculate chamber, party, and
medians
of median
ask
produces—targeted,
use
cham
represents
we
preferences,
prevents
"cheating]
deals that have been struck" by the majority party's leadership (Cox and McCubbins 1993, 2).
Individual member preferences, the informational
needs
are
effects,
deference fromthe chamber in theirjurisdictions (Gilligan
and Krehbiel 1990).
ties
House Interiorand District of Columbia committees also fall
into this category. Finally, committees that produce a blend
This content downloaded from 140.232.1.111 on Sun, 11 Jan 2015 04:29:52 AM
All use subject to JSTOR Terms and Conditions
Parkerand Dull
635
Government
Operations,
Budget,
Commerce, Ways and Means,
defined
as cartel
committees
to these
according
at least
criteria.
and
Energy
and Appropriations are
Banking,
75 percent
of the time
Education
and
five are coded as miscellaneous, including Agriculture
(86%), Interior (100%), Merchant Marines (100%), and
District of Columbia (96%). Miscellaneous committees,
produce
speaking,
externalities
targeted
but
do
not have committee medians significantly differentfrom
the chamber's median. We considered various operationof the
alizations
committees
but
of how
regardless
we
defined cartel, distributional, information, or miscellaneous
the
our findings
committees,
Whether
a committee
of
benefits
do not change.9
investigates
divided
the
committees
to engage
in
control,
party
Hypothesis 4\ Cartel committees will produce the
more and longer investigations alleging the
executive branch waste and malfeasance under
divided government.
The effect of divided government on the relationship
between cartel committees and investigations is examined
here by dividing the sample into divided and unified party
control and estimating the model for each subsample.10
Informational
This
costs.
committees
exist
somewhere
between
car
tel and distributionalcommittees in termsof the opportunity
costs of investigations. Public hearings broaden general
and,
knowledge
consequently,
weaken
may
a committee's
informational advantage. Nevertheless, investigative hear
ings
on whether
depends
outweigh
investigations
of cartel
aggressive oversight will be amplified during periods of
Labor,
Foreign Affairs, Judiciary,and Intelligence are most often
coded as informational committees. Few committees fit
the distributive definition: Agriculture is defined as distributive 10 percent of the time, whereas Post Office and
District of Columbia are coded as such 4 percent of the
time. Most of the committees scholars viewed as distribu-
generally
that the tendency
expect
an important
represent
serve
sight—and
to
measures
formal
instrument
such
of committee
a public
make
as
case
and
subpoenas
over
for contentious
even
of
charges
opportunity cost analysis depends on the type of exter-
contempt (Fisher 2003). Committees that push the use of
nalities
their powers
the committee
produces
and
whether
the commit-
tee pursues member, chamber, or party goals. Gatekeeping
committees
a prominent
play
role
in shaping
the
party
brand, so the investigations pursued should reflect key
party goals and interests. Committee investigations also
expose wrongdoing and allocate blame, offering few
opportunities to advance the production of goods for distribution back home. Members serving on distributive
committees may actively avoid exposing possible
instances of corruption for fear of disrupting benefits
flowing to the district. Although the allocation of
resources may generate more opportunities for malfeasance, the incentive structure decreases the likelihood of
distributive committees generating investigative hearings.
We
offer the following
therefore
tional
committees
2: Cartel
and
longer
branch
when
or informational
3:
Hypothesis
fewer
investigations
malfeasance
tributive
and
Distributive
shorter
will
committees
more
produce
the executive
alleging
with
compared
dis-
committees
investigations
generate
alleging
execu-
tive branch malfeasance when compared with
cartel or informational committees.
because
the
presidency
manages
the
executive
branch,
periods of divided government amplify partisan and individual
incentives
for committees
to expose
incompetence
and corruption in the executive branch (Moe
1999). We
risk
majority
should
be more
to pursue
likely
aggres
under divided government,
Epstein and O'Halloran (1999) develop a model of del
egation to and oversight of agencies that builds on and
extends the partisan, informational, and gains from
traditions.
executive
model
gates
their analysis
Driving
takes
organization
of statutory
to a committee
delegates
informational
of policy
to an
policy.
Both
further
arrangements
asymmetries.
The
agency
dele
chamber
The
specialists.
that
agency
House
of the
a two-stage
to propose
The
delegation.
is a key insight—
in the shadow
shape
them
branch—leading
implements
committee
and
specializes
are
with
wrought
has the most
accu
rate estimate of policy impacts, the committee has a slightly
less accurate estimate, and both use this advantage to influ
ence
Congressional investigations are bounded by the
Constitution (Landis 1926; Taylor 1955), particularly
when it comes to investigating the White House. Yet
the chamber
Hypothesis 5: Informational committees will
undertake more and longer investigation alleg
ing executive branch waste and malfeasance
then
committees.
from
sively oversight during periods of divided government
(Epstein and O'Halloran 1999).
congressional
Hypothesis
far away
losing autonomy (Gilligan and Krehbiel 1994), but informa
exchange
hypotheses:
too
the discretion
granted
by the floor
voter.
Floor
voters
thereforeface two problems when they decide whether to
delegate. They do not know the real-world factors that will
shape a policy's final impact; and both the committee and
the agency
know
more
about
these
real-world
factors
and
will use informational advantages to steer policy toward
their
own
O'Halloran
rather
than
the
floor's
priorities.
Epstein
and
(1999, 48) conceptualize these trade-offs as
This content downloaded from 140.232.1.111 on Sun, 11 Jan 2015 04:29:52 AM
All use subject to JSTOR Terms and Conditions
PoliticalResearch Quarterly66(3)
636
losses
agency
associated
with
or
delegation
"political
instead
becoming
of "political
instruments
Parker
combat."
transaction costs." Policy by legislation avoids agency
bias, but forces the floor to rely in part on informationfrom
and Dull (2009) demonstrate that the relationship between
partisan conflict and congressional oversight shifted
the committee.
Policy
between
but
more
tise
leaves
to influence
committees
offers
by delegation
room
greater
for opportunistic
experor
agencies
policy.
the periods
Among
the
during
before
the most
and
after
significant
in relation
1970s
1975.
of the changes
between
occurring
and
Congress
the
When it is politically efficient—when workloads are
too great or policies are informationallyintense—Congress
will delegate. More importantly,the model treats formal
agency discretion and the exercise of congressional over-
agencies was the dramatic growth of congressional institu
tional capacity. Committee staff and hearing days more
than doubled between 1969 and 1977, before leveling off
in the 1980s and 1990s (Aberbach 2002). Committee staff
sight as related
plays
of delegation.
instruments
occurs
Oversight
in the shadow of statutory delegation to the agencies.
Oversight
authority,
this
take
hearings
operating
context,
in the context
place
as a constraint
and
Epstein
of delegated
on agency
O'Halloran
contend
(1999)
trade-offs
In
discretion.
that
committees operate as "contrary outliers," accentuating
conflict
between
House
the agency
as a whole.
Their
and
relative
committee
analysis
to the
that where
shows
com-
committees
in more
engage
aggressive
moni-
in the conduct
role
associated
on investigative
a count
of oversight
(West
1995,
with
hearings.
of staff assigned
expanding
committee
resources
We include
Committee
Capacity,
to the committee
Ornstein,
(see
Mann, and Malbin 2008). Deering and Smith (1990, 154)
that broad
observe
with
mittee jurisdictions delegate broad policy decision making
to agencies,
a central
131) and staff capacity defines the scope and the relative
and
numbers
large
diverse
are associated
jurisdictions
of staff.
Finally, three variables relate to the political environ
ment.
the number
of Control,
Margin
of seats
controlled
toring of agency activities, including oversight hearings
(Epstein and O'Halloran 1999, 213). Consistent with this,
by the majority party minus seats held by the minority
party in the House, measures the strength of the party
we
majority.
that
expect
areas
characterized
by
broadly
delegated
research
Contemporary
on
Congress
empha
authority will be subject to greater committee oversight,
particularly when these issues are subject to conflict
sizes the conditional nature of partisan institutional poli
tics (McCarty, Poole, and Rosenthal 2006; Rohde 1991).
between
A majority
rooted
tion
dedicate
the executive
and
branches.
legislative
Hypothesis 6: Committees that have granted agencies broad statutory discretion will undertake
more and longer investigations during divided
should
Delegated Discretion is a trichotomous high-mediumvariable
derived
from
Epstein
and
O'Halloran's
(1999, 200-203) analysis of delegation in significant legislation
between
should
be positive
The
and
during
committee-level
ate through
ful
1946
of which
1995.
periods
dynamics
changing
institutional
are
by
shown
prior
Discretion
Delegated
of divided
government.
specified
so
research
far opera hand-
arrangements,
to
influence
committee investigations. We include here five variables
measuring a few factors shaping committee oversight.
First, we include a dichotomous variable Post-1974,
which
takes
a value
of 1 for all congressional
committees
after the 93rd Congress (1973-1974). Partisan polarization in Congress (McCarty, Poole, and Rosenthal 2006)
and the reform of House rules in the wake of Watergate
shifted power to the party leadership in the House
(Aldrich and Rohde 1997). Set amid broad political and
institutional
changes
during
the second
half
of the twen-
tieth century, Zelizer (2004, 178) describes the changing
nature
of
committee
investigations
in
the
post-reform
era, noting that during the 1970s scandals were, "no Ionger
part
of a vibrant
movement
to reconstruct
politics,"
coali
legislative
resources
to poli
through legislation. For a slim partisan
majority, however, investigations may be particularly
useful devices for influencing policy and defining reputa
as Margin
Thus,
committees
low
cohesive
committee
greater
cymaking
tions.
government.
in a strong,
will
of Control
more
generate
we
decreases,
expect
investigations.
Gridlock is the absolute ideological distance between
the filibuster
House—in
and
pivot
the furthest
Rosenthal's
Common
scores.
Space
trois
for the structural
impediments
that
can
production
contribute
tions,
veto
or
pivot—Senate
each Congress, calculated using Poole and
frustrate
the
to a willingness
when
particularly
to use
the
This
variable
in the political
of legislation
congressional
Gridlock
con
system
and
may
investiga
Interval
is wider
(see Krehbiel 1998).
Presidential Approval represents the public's approval
of the president's job and is derived from monthly Gallup
polls
averaged
over
the
first three
months
of each
year,
Because of their public nature, the incentives for congres
sional committees to initiative and sustain investigative
hearings are shaped by changing perceptions of the presi
dential administration as a whole. Allegations of corrup
tion and inefficiency leveled
against a popular
administration
may
fail
to produce
electoral
benefit
for
individual members or the congressional parties, or even
prove
costly.
When
the president
is popular,
congressio
nal committees will initiate fewer and shorter investiga
tions
alleging
executive
branch
This content downloaded from 140.232.1.111 on Sun, 11 Jan 2015 04:29:52 AM
All use subject to JSTOR Terms and Conditions
malfeasance.
Parker
and
Dull
Table
I. House
637
Committee
Investigations
of the Executive
Branch,
1947 to 2004.
variables
Dependent
variables
Independent
Gridlock Interval
Margin of Control
Chair-President
Informational
Ideological
Distance
Committee
Distributive Committee
Cartel Committee
Discretion
Delegated
Presidential Approval
Post-1974
Committee
Capacity
Constant
n
Wald
Investigations
Pages
Days
-0.4310(0.6204)
-0.0028 (0.0022)
0.6261** (0.2155)
0.4433** (0.1731)
-0.8696 (6.5736)
0.7527*** (0.1992)
0.2399* (0.0994)
-0.0041 (0.0053)
0.2921 (0.2290)
0.0060* (0.0027)
-1.8052*** (0.4441)
923
-0.8804 (0.8244)
-0.0026 (0.0027)
0.5177 (0.2902)
1.0935*** (0.2714)
-1.1908 (7.6086)
1.1726*** (0.2780)
0.2606 (0.1380)
-0.0086 (0.0078)
-0.0020 (0.3349)
0.01 15** (0.0041)
4.0302*** (0.6986)
-1.1566 (0.8144)
-0.0014 (0.0027)
0.4670 (0.2806)
0.7558** (0.2505)
-1.3200 (6.2865)
1.0640*** (0.2646)
0.3838* (0.3023)
-0.0060 (0.0080)
-0.5372 (0.3142)
0.0041 (0.0039)
0.1297 (0.6825)
923
43.50***
Unit of analysis is the committee
are reported in parentheses.
*f> < .05. **p <.01. ***p < .001.
We
model
and
three
87.61***
dependent
1 seem not to support
The results in Table
Hypothesis 3; the Distributive coefficient is not signifi
Results
and
cant.
variables—Investigations,
for House
Pages—observed
committees
annu-
specify
mial
models
regression
distribution.11
Table
the
using
1 reports
bino-
negative
estimates
committee
2
Table
reports.12
model
reports
and
divided
first,
government.
Hypothesis
each
Taking
1 is
weakly
only
indicates that while ideological
men
and
committee-based
pears
are
is
presidents
in Table
considered.
2 once
associated
the
the effects
Oversight
may
1
Table
conflict between chair-
seemingly
investigations,
in turn,
hypothesis
supported.
with
relationship
of divided
provide
opportunities
chairman's
reputational
needs
than
other
mean
levels
illustrate
investigatory
activity,
constant.13
variables
of investigation
the
effects
we
calculated
Cartel
of
committees
all
across
production
Miscellaneous
whereas
variables,
dependent
three
committees
consistently exhibit the lowest probability of exceeding
investigatory
mean
number
and
factors.
on
independent
mean
for
commit
consistently yield the highest probability of exceeding
disap-
chairmen to build their individual reputations, but the
initiation of investigations seems predicated less on the
To
committees.
type
more
government
is
category
the probability of investigatory activity exceeding the
mean across our three dependent variables while holding
other
estimates separating the sample into periods of unified
committee
that Distributive
means
tees are not significantly differentin their behavior from
committee
for models
base
the
this only
Miscellaneous
predicting total committee Investigations, Days of committee investigative hearings, and Pages of investigationrelated
as
However,
Miscellaneous,
ally. A goodness-of-fit test indicates overdispersion, so
we
standard errors (1,000 repetitions)
session/year. Cell entries are negative binomial coefficients. Bootstrapped
of Estimation
Method
Days,
923
94 43***
x2
26
percent
committees
Cartel
activity.
of Investigations,
of the
are less
time,
active,
committees
Pages,
and
respectively.
exceeding
mean
exceed
Days
the
34,
14,
Informational
Investigations,
Pages, and Days 29, 13, and 23 percent of time, respec
tively. Finally, Miscellaneous committees—the bulk of
which
are
committees
producing
targeted
externalities
The committee-type variables—Cartel, Informational,
and Distributive—generally yield the expected results.
Consistent with Hypothesis 2, Cartel committees produce
more and longer investigations, generating positive and
significant coefficients in all three models in Table 1.
that are not significantly differentfrom the chamber mean
ideologically—are the least active, exceeding the mean
levels of investigatory activity 21 percent of the time for
significant and positive. Under unified government, Cartel
is significant and positive for the Days variable only.
longer investigations under divided government, whereas
under unified government they produce coefficients that
Consistent with Hypothesis 4, when all three models—
Investigations, Pages, and Days—are estimated for periods of divided party control (see Table 2), Cartel is
Investigations, 9 percent of the time forPages, and 16 per
cent of the time for Days.
Consistent with Hypotheses 5 and 6, both Informational
committees and those committees with jurisdictions char
acterized by broad delegated discretion produce more and
This content downloaded from 140.232.1.111 on Sun, 11 Jan 2015 04:29:52 AM
All use subject to JSTOR Terms and Conditions
N O VO CO in — n
LO^LOLOLOOOLnm
oOvocoiocorM
— o — rs^OLoor^oo
— o — OOvOOOOO—
ON CO N CO
CT» LO
— ooco
1
ro -ft lo CO O lo -X- in ro — lo
— co o
N*(>fS(N(N^n
— ^fNOOLOOpOroOlN
~ —
0
fSNO^ffjoton
(NoOCJOO-OOOO
I I
I
1 O
I
C
<u
c
£
L.
>0}
&
XI
J8
-
v,
a)
5P
rF
°-
rj- ro•
n .
^
— — fN — •*}- — Oro
(NOtT
vfl m N 10 n — LO O fN (N —
— "f
ro O ro O In rs
—
LOOroroLOfSI^.0000
— o — 00K0000—
£
""J"O
* ro>OvOfSroOO^vO*
n*cooiflvOO^Nifl(>f
— 5, ro,/
oSvOO^vCCOroOOO
^
—
oo^r^r-i<^omo
og —
— — ooo
OoOOO
1
1
o
Id1
"»
I
—
OfNvO"*rOcNroOO —
NOrovomoN
oo co oo 10 vOOfMsOOOLOCXJOOOCrv
COOCO-fS-(SOtO^
£
c
OOOOO^OOOOOqq
— OvQN*t
vO * TfrONOLO
ro cn CO CO vo — Is ^ 1/1 ?
^
ns n
(NO^-OOcN-TOfSO^
oSoooooooSon
o
I
I
_j
O
ft
J
\D
^
ro
ON>ONCOONSOv(S(St
COnLOfNO^COCOfSOtCO
— sO O 00 ro — — Oro
(NO
— n ^ n O t O —
(NOvO
2004.
^oooo^oooo^
"*■ O — * * *
ro t vO O LO *
^OOirCiSOvOOfS^
£
*
OOO^l'^rooNOOOO
I
I o o
^
fN CO t
t N
i
i
|
N ip rS N t
<n
—
i
to
«3
o
o
m
1947
C
£
E
E
>
V
Government,n
, COO(N(NONr^CX)OroO
—
OfSvOmNOOfStiOfS
lo O ro In ro lo ro
— ro t|- ro
lOOvO
p,
—. P. P. p, O,
N (S t - * * *
o o ro 3 s
*.
—
I
Unified
and
under
Investigations
u
c
o
.82
Q
Committee
11
2
House
Table
|
O
O
p
*
\£>
—
5 s
°
—
rf
h«- O
—
ro (N
lo
oo o
(N o
<
* LO
t N
<n
— lo
Is**
J
O
lo
—;
Z
_
U
.6 Si
s &
8
I
E S
8
S | E !5 L | I g-
75
lo
T
p.
CM
R g s
OOOo^O^OOZ^vOlo (N
;
ro —
(N
I
OOlNvONffvOfStNtfO
vOroiOOONrNlNvOroO
— rN -*r ro O
nOOlo
OOp^r—;(Nro(NO(NOlN
o^o^o^o^o^r^o^o^o^o^o^
ro
CO
o
O
Divided
2.
—
O
O,
00
§
Sv3^o"l^|2
oi-
■25oiL6«E:9o!5^^i/1
a
§.
^
* IS I I I 1 »R 1 S
IS f
is c J
I 3
t S 5 1
IS ?|iUiSNls
U^UQEQUi^UU
~*
"o
c
638
This content downloaded from 140.232.1.111 on Sun, 11 Jan 2015 04:29:52 AM
All use subject to JSTOR Terms and Conditions
|^
—fo.
2$
ví g
o p
VI
5,
Parker
and
are mixed
and
chairmen
mittees
Dull
639
of significance.
short
to opposing
belong
When
parties,
with jurisdictions
and those
presidents
and
com-
Informational
by broad
characterized
statutorydiscretion are more likely to investigate agency
wrongdoing. Once divided government is taken into
account, however, the ideological distance between the
president and the committee chair is no longer a significant
of a committee's
predictor
behavior.
investigatory
under divided government, which empirically is indeed
the case. It is difficult, however, to put much stock in
these findings or the insignificant distributive coefficient
reported in Table 1. Only fourteen committees fitthe deftnition of distributive, and only six operate under the congovernment. In
of the
three-quarters
miscellaneous
fact, more
committees
than
produce
targeted externalities, but do not have medians significantly differentfrom the chamber median. It is not at all
that these
clear
two
are
categories
distinctive
empirically.
Combining the miscellaneous and distributive categories
into a single variable generates results in line with the
expectations outlined in Hypothesis 3: The distributivemiscellaneous category yields the fewest and shortestcommittee
hearings
in divided
and
unified
Presidential popularity is not associated with investigatory activity in the House. Consistent with prior findings
(e.g., Parker and Dull 2009) Margin of Control is significant and negative only in the unified government models
in Table 2, suggesting that large, partisan majorities act to
suppress furtherthe investigatory impulses of House committees when the White House is occupied by a partisan
ally. Finally, the institutional changes in Congress dating
to the mid-1970s do not generate additional investigations
or pages
at the level
of the committee;
however,
investiga-
tions are shorter in Days after 1974. This null finding is
likely the result of controlling for Committee Capacity,
which
and
aged
changed
varies
greatly
dramatically
between
26 staff members
post-reform
between
committees.
in the pre-1974
era, committees
averaged
these
two
periods
Committees
era, whereas
more
than
are
committees
80.
by committee
of a handful
ues
deviation
standard
of any
staff is also
associated
with
more
of key
variables.
explanatory
Scenario
the mean.
below
scenario
This
yields
relatively low levels of activity: Under these conditions,
miscellaneous committees will generate more than one
investigation only 18 percent of the time. Likewise, the
mean number of pages (205) and days (2) is exceeded
8 and
14 percent
of the time,
respectively.
Altering the base category to a Cartel committee and
increasing chair-president conflict to one standard devia
tion above the mean, Scenario 2 shows a substantial increase
in the probability of an investigation emerging and becom
ing more intense. A Cartel committee is likely to exceed the
mean number of investigations by 21 additional percentage
points when compared with Miscellaneous committees,
More days and pages are also more likely to be produced,
with the probability of exceeding the mean number of days
by
13 percentage
points
and
the mean
number
of pages
by
approximately 7 percentage points. Scenario 3 moves com
mittee discretion from low to high, and this increases the
probability
points,
assigned to a committee increases the likelihood it will
initiate and prolong investigations.
Summarizing the results of the analysis, Cartel and
Informational committees are strongly associated with
more and longer investigative hearings. Miscellaneous
Distributive
hearings
1 posits a pre-1974, low-conflict committee categorized
as Miscellaneous with chair-president conflict set at one
in the
Veteran's Affairs averages only 28. On average, committees still employed 75 staffmembers. Committee Capacity,
as anticipated, is positive and significant. More staff
no
ger investigations of the executive branch,
The calculation of predicted probabilities explains
how investigation activity is affected by each indepen
dent variable. Table 3 reports the predicted probability
that each of the three dependent variables will exceed
the mean level of investigatory activity at differentval
tions,
after 1994, when Republicans reduced committee staffby
a third, great disparities existed between committees:
Appropriations averages 149 staff members, whereas
to produce
and longer investigations of the executive branch. Finally,
divided government affects the opportunity costs for
investigations undertaken by cartel and informational
committees—in both instances, leading to more and lon
aver-
Even
where
government,
predicted
length. Committees with jurisdictions defined by signifi
cant delegated discretion are associated with a higher ire
quency and duration of investigative hearings, particularly
under divided government. The capacity to investigate as
only
government.14
divided
under
except
measured
Distributive committees produce the fewest number of
hearings in divided government—the coefficient is negative and significant. In fact, distributive committees have
zero probability of producing any hearing of any length
dition of divided
committeesgeneratethefewestandshortestinvestigations—
days,
the
of exceeding
and
pages
by
mean
another
number
of investiga
5, 3, and
1 percentage
respectively.
Finally,
Table
4 illustrates
the effects
of divided
gov
ernment and delegated discretion on investigatory activ
ity. All variables are constrained to the levels as outlined
in Scenario 2 above. First, note the low levels of investi
gatory activity in unified government regardless of the
amount of delegated discretion: There is no apparent rela
tionship between committee's propensity to investigate
and the amount of delegated discretion granted to agen
cies. Congressional committees are unlikely to initiate an
investigation and, if they do, are likely to keep it short
during unified government even when the institutional
and political environment favors lengthy investigations.
This content downloaded from 140.232.1.111 on Sun, 11 Jan 2015 04:29:52 AM
All use subject to JSTOR Terms and Conditions
PoliticalResearch Quarterly66(3)
640
Table
3. Predicting
Exceeds
the Mean,
the Probability
that the Number
and Intensity
of House
Committee
Investigations
of the Executive
Confidence
Probability of exceeding
the mean (%)
I: A miscellaneous
Scenario
Number
Branch
1947 to 2004.
Lower
bound
interval
(%)
committee
of investigations
18
13
24
Hearing
days
14
7
21
Hearing
pages
8
2
14
2:A
Scenario
Number
cartel
committee
39
32
48
Hearing
days
of investigations
27
20
34
Hearing
pages
15
9
21
44
35
53
22
37
10
22
3:A
Scenario
Number
cartel
committee
with high discretion
of investigations
Hearing
days
30
Hearing
pages
16
Independent variables other than post-1974, committee type, discretion, and chair-president
and confidence intervals are calculated with Clarify.
Table
4. The
1947
to 2004.
Probability
House
Committee
Divided
Investigatory
Activity
Low
Mean
Levels
in Divided
Lower
bound
versus
Unified
Government,
Unified government
government
Confidence
Probability of exceeding
the mean (%)
Exceeds
conflict held to their mean values. Point estimates
interval
(%)
Upper
bound (%)
Probability of exceeding
the mean (%)
Confidence
interval
Lower
Upper
bound (%)
bound
(%)
discretion
Investigations
34
24
45
21
Hearing
22
12
33
18
-1
37
14
8
25
8
-5
22
days
Hearing pages
High discretion
7
45
Investigations
46
35
57
24
8
49
Hearing
days
30
21
39
22
4
40
Hearing
pages
17
8
25
18
Independent variables other than post-1974, committee
and confidence intervals are calculated with Clarify.
type, discretion, and chair-president
Alternatively, the condition of divided government
and high levels of delegated discretion amplifies investigatory activity by House committees. Looking across the
table, the increased probability of investigations or investigatory activity exceeding mean levels between unified
and divided government runs between an additional 4 and
13 percentage points when delegated discretion is low. To
see how discretion increases this effect in divided govern-
ment, examine the bottom half of the table. The probabil
ity of the number and days of investigations exceeding
the mean is dramatically affected by high discretion in
divided government: an additional 22 and 8 percentage
points, respectively. The effect on page production is
-2
22
conflict held to their mean values. Point estimates
similar: High levels of delegated discretion under divided
government yield an additional 7 percentage point likeli
hood of exceeding mean levels of page production by the
committee. In sum, the chance that a committee will
exceed the mean number of investigations is more than
45 percent in divided government when delegated discre
tion is high and the committee is responsible for protect
ing the party brand,
_
.
.
We generate three findings improving our understanding
of congressional investigations. First, divided government
This content downloaded from 140.232.1.111 on Sun, 11 Jan 2015 04:29:52 AM
All use subject to JSTOR Terms and Conditions
Parkerand Dull
matters.
The
641
is that Congress
presumption
to
investigates
Members
on
these
committees,
are
therefore,
chosen
guard jealously its constitutional prerogatives. Ambition
carefully to protect either the chamber or the majority
should
party's
while
counter
a
ambition.
of
system
Our
demonstrate
findings
separated
powers
certainly
that
produces
institutional incentives to oversee and investigate the
executive, partisan considerations alternatively temper and
encourage
and
the
have
you
other
of executive
pursuit
legislative
"When
doing.
one
the
controls
branch
that controls
party
presidency
wrongCongress
harsh, it's all about... discrediting the administration. It's
about exposing all their warts and flaws," former
Government
Reform
Committee
Chairman
Tom
Davis
of the
same
party,
tend
you
to under-investi-
the
Second,
decision
to
is
investigate
a
function
of
much more than the committee chair's own authority or
the individual opportunity costs faced by committee
members. By restricting our analysis to fire-alarm-style
investigations—those oversight moments rife with reputational benefits for credit claiming and position-taking—
we maximize the possibility of uncovering the effects of
individual and chairman incentives on the decision to
investigate. Yet, we find that the external political context
and
committee
are most
structures
in determin-
important
the effects of a second-term presidency, public opinion
and
efficiency
government
the
economic
general
and
launch
In any
and
sustain
case,
Reform
Government
investigations
our research
on
the
to
decision
of the executive
branch.
borne
Third,
congressional
sance.
by individual
institutional
We
that
matter
directly.
a
robust
branch
literature
of
malfeato
developed
Informational
committees,
to help
designed
the chamber
develop sound policy, incentivize investigatory activity
by design. Furthermore, the apportionment of committee
assignments further cements the investigatory roles of
these
committees.
Members
choosing
to pursue
narrow,
particularistic benefits for their districtoften get placed on
distributive committees like Agriculture. Investigations,
by definition, do not furtherthat goal and are unlikely to
be initiated or—if they are—will be short, cursory affairs.
Informational committees and cartel committees by design
require
more
oversight,
and
these
activities
have
conse-
quences reaching beyond the committee membership.
and
Oversight
is
mandate,
in terms
of
because
committee,
its broad
far the
by
most
In its various
of investigations.
of testimony
pages
and
documents
and 2004. The consequences
cartel
powerful
1947
between
of finding wrongdoing by
committee
can
be
and
widespread
devastating, which is why the investigatory activities of
this committee are particularly sensitive to the presence
of divided government and its membership carefully
reflects
medians.
party
affirm
We
the
notion
that
over
sight is a collective responsibility involving more than
individual member decisions.
Recent evidence suggests that Congress has abdicated
its oversight role, and this has been a cause for concern
(Mann and Omstein 2008). Our findings explain the
oversight retreat: Partisan majorities in unified govern
ment do not investigate a president from their own party
because paying the transaction costs is inefficient. It dam
the party
and
brand,
with
agencies
wide
ranging
dis
cretion are unlikely to develop policies far from the party
majority's preferences. The arrival of divided govern
ment during the final years of the Bush administration,
a
brought
dramatic
in
increase
investigatory
activity (Binder et al. 2009). Our findings, however, sug
gest
A "mended"
caution:
does
not
Congress
in divided
performing
serve
necessarily
committees
the public
government
pursue
vigorous
interest.
investi
gâtions aggressively and intensely, yes, but such activity
DC,
serve
only
given
partisan
explain the legislative role of committees provides important insights into the nonlegislative operations as well.
congressional
on
iterations, Oversight and Government Reform has pro
duced 198 investigations totaling 645 days and more than
may
for the conduct
of executive
investigations
show
members
structures
not)
Committee
oversight
committee
Cartel
investigations are the product of more than the opportunity costs
This
and
oversight
that congressional
demonstrates
does."15
however,
mood, and the special role of the House Committee on
Oversight
active
ages
ing whether an investigation is launched and the intensity
with which it is pursued. Future research might consider
about
ernment
this
gate." To amend Madison's famous phrase, ambition
counters ambition especially when the president's party
and the majority party in the House are different.
(or
pursuing
House
Government Reform is a clear example: "Our primary
responsibility... is oversight of virtually everything gov
91,000
(2009; R-VA) observes, noting that by contrast,"When the
president's
The
jurisdiction
is pretty
oversight
when
reputation
investigations.
such
purposes.
ter government.
to increase
partisan
are
investigations
More
Indeed,
investigations
given
rancor
in Washington,
potentially
need
driven
not imply
the consternation
by
bet
expressed
by some politicians and the public at large about pork bar
rel spending, our findings on investigations should give
pause.
Yes,
Congress
may
be
more
willing
to investigate
the executive branch under divided government; how
ever, it will not disturb the logrolling arrangements so
important to the reelection prospects of individual mem
bers.
Agricultural
subsidies
and
transportation
dollars
will flow unimpeded back home, but the possibility of
graft,corruption, and executive malfeasance in these pro
grams being uncovered is low given the unwillingness of
these committees to engage in vigorous oversight. In
short, committees may be more willing to investigate
intensively
in divided
government,
but whether
is better governance is an open question.
This content downloaded from 140.232.1.111 on Sun, 11 Jan 2015 04:29:52 AM
All use subject to JSTOR Terms and Conditions
the result
PoliticalResearch Quarterly66(3)
642
occurred
investigations
Acknowledgments
fied government.
The
authors
David
thank
Hilary
Girshick,
Keith
Cole,
Patrick
Houweling,
Mike
Anne
Franz,
and Pavielle
7.
Khademian,
Karen
Haines.
earlier
Huit,
understates
8.
of this article
Midwest
Political
was
(MPSA)
Chicago.
Declaration
The
of Conflicting
declared
author(s)
respect
no potential
to the research,
9.
Interests
of interest
with
10.
and
Stewart
and Jonathan
The
.001.
Funding
no financial
received
and/or publication
authorship,
for the research,
support
of this article.
Notes
used
in this
Dataverse
[IQSS]
and Tatelman
White
to
British
sight
the wake
review
the divided
Dull
from
were
2008)
4.
See
5.
Although
with
the other
ment
in the model's
traces
and
Dull
branch
not be
committees
The
in
Dickey-Fuller
Congresses
syntax
are
in select
in the analysis
can
be
more
apt to be
not alter our results;
indeed,
in more investigations
This
mittees
are
not
of investigatory
creation
of select
provides
check,
investigations
This tension
select
government.
committees.
the
Including
ing divided
starved
in
pulled
goals—investigating
committees.
does
they can
in little over
new legislation.
tend to engage
validity
use
and investigations,
by multiple
or proposing
found
from the analysis
we
suggests
Excluding
a conservative
that standing
oxygen
select
dur
com
by
the
committees
test of our theory.
we only miss four of Mayhew's
using
our
method.
Two
13.
with
each
of those
selection
session
did
criteria.
We
bino
negative
and the results
of autocorrelation,
unit
root
were
sub
we performed
an
using
1 lag.
We rejected
augmented
the null that a unit root
time in favor of the alternative
Confidence
in the series
intervals
and probabilities
are calculated
and King
(Tomz,
14.
Results
available
15.
http://oversight.house.gov/about-the-watchdogs/
Wittenberg,
hypothesis
that
is stationary.
Clarify
(2005
and
test. We ran the test on all of our dependent
at least one panel
matters and engage
committees
select committees
As another
exists
are often formed for the specific
oversight
directions
executive
110th
(2009).
legislative
Standing
original
officials
(2008)
in a given
our
a zero-inflated
using
test for panel
Schwartz
it
improve
for days
performance
not shown)
the possibility
as a third of the committee
and
(data
the Fisher
appendix.
of conducting
investigate
sight.
109th
and
test indi
and interacting
a significant
committees
To address
literature.
as many
the
select committees
competing
Kriner
government
similar.
variables
and
government
in the online
purpose
of government
the models
stantively
12.
between
Ratio
government
satisfying
of
identifies
alleging
A Likelihood
predictive
misuse
and
tradition,
a hearing
mial regression
over
congressional
in unified
provided
at p <
significant
(p < .08 and .09).
of all House
hold
estimated
of legislative
variables
conflict
distance
and the president
divided
of alternative
with ideological
of .667,
DW-NOMINATE
that including
A majority
Charles
types.
correlation
cated
not
not yet published.
Parker
found
6.
Oleszek,
investigations
(2009)
of this writing,
hearings
may
Kaiser,
the codification
11.
of the War of 1812.
and
also
on publication.
Parliamentary
Parker
As
Science
99-101)
(1951,
at Harvard
deposited
Social
funds and the misconduct
public
3.
be
Quantitative
) discuss
committee
congressional
2.
for
Network
(2011
oversight.
will
article
Institute
University's
the
Nelson
from
for a discussion
it is .800.
investigations
1. Data
Garrison
110th
the
through
in divided
is .283,
from
is correlated
average
prin
1993).
Woon.
of committee
chairs
by political
for the 80th through
obtained
appendix
government
committee
author(s)
103rd
with a Spearman's
article.
The
the
the online
Divided
of this
and/or publication
authorship,
Congresses
operationalizations
conflicts
A long tradi
that this approach
Woolley
membership
were
(2009),
See
contends
of control exercised
1990;
of
that the preferences
assumption
and Shipan
on committee
102nd
in
meeting
the degree
(Ferejohn
Data
at the 2010
presented
Association
Science
appendix
tion in the study of bureaucracy
cipals
version
uni
for an expanded
are defined by those of the president.
the agency
Any errors
Note
Authors'
We admit the problematic
of the authors.
are the sole responsibility
and two under
divided
the online
discussion.
Goodman,
Craig
Smith,
Roberts,
Parker, Jeffrey Lazarus,
remaining
An
Wade
Canon,
Rachel
Van
Robert
under
See
from the authors
with
2001).
on request.
References
Joel. 2002.
Aberbach,
Congress
Aldrich,
"What's
John H., and David
to Republican
Rule
ries of Congressional
to the Watchful
Happened
& the Presidency
Eye?"
29:3-23.
W. Rohde.
in the House:
Parties."
1997.
"The
Transition
for Theo
Implications
Political
Science
J. Deering.
2001.
Quarterly
112:541-67.
Baila,
Steven
over Time
J., and Christopher
and across
of Police
Patrols
Midwest
Political
Bendor,
Jonathan.
British
Binder,
and
Journal
Sarah
Molly
A.,
Committees:
and Fire Alarms."
Science
1988.
of Political
Thomas
Reynolds.
Paper
Association,
"Formal
E.
Mann,
Mending
This content downloaded from 140.232.1.111 on Sun, 11 Jan 2015 04:29:52 AM
All use subject to JSTOR Terms and Conditions
Measure
presented
at the
IL.
Chicago,
Models
Science
2009.
"Oversight
An Operational
of Bureaucracy."
18:353-95.
Norman
J. Omstein,
the Broken
Branch:
Parker
and
Dull
643
the
Assessing
Cox,
110th
DC:
Washington,
Leviathan:
Mathew
Tom. 2009.
Branch:
1993.
in the House.
Legislative
Uni
Berkeley:
on Leadership
David
New
Haven:
Mayhew,
David
and Administration."
Yale
shipandAdministration_November2009
Deering,
Christopher
tees in Congress.
and Sharyn
David,
Epstein,
2nd ed. Washington,
ers: A Transaction
under
O'Halloran.
Cost Politics
Powers.
Separate
DC:
New
Commit
Pow
to Policy
Approach
Cambridge
Press.
Influence
Charles
on Bureaucracy."
Shipan.
1990.
Journal
of Law,
&
Economics,
6:1-20.
Organization
Morris.
Procedures."
"Alternative
Journal
Rationales
of Law,
for Restrictive
1999.
&
Organization
dure
Investigations:
Subpoenas
Economics,
3:337-45.
Fisher,
2003.
"Congressional
and Contempt
Power."
Thomas
CRS
and
W.,
for Congress.
Report
American
Ginsberg,
the Press
W.W.
2005.
by
from
Norton.
The Politics
Press.
Virginia
R. Weingast.
Barry
&
Economics,
Proce
Organization
DC:
Washington,
University
of Chicago
Keith.
Political
American
Information
Politics:
Pivotal
University
and Liam
and
Legislative
of Michigan
University
1998.
Com
Science
A Theory
of Chicago
Schwartz.
2008.
Investigations."
of U.S.
Law
"Divided
Govern
Legislative
Studies
James M. 1926.
(2): 153-221.
"Constitutional
of Investigation."
Limitations
Harvard
Law
on the Con
Review
40
ed.
8:203-25.
Role
Station:
College
As Chief
edited
Presidency,
265-84.
Fire
2:165-79.
of Principal-Agent
Science
The President's
2009.
"Committees
80th-102nd
1976.
by James
Texas
A&M
in the U.S.
Congress,
Congresses,
May
15.
Supervision.
J., Thomas
Vital Statistics
David
Shepsle,
and
Seymour.
Kenneth
1963.
2009.
DC:
We
"Divided
Investigations,
34:319-45.
1997.
Call
and Leaders
of Chicago
"Conditions
of Politics
A
Congress:
Oxford:
Voting.
in the Postreform
Press.
for Legislative
Science
of
Review
and
Committee
1987.
Power."
"The
Insti
American
81:85-104.
Barry
of Congressional
of Congressional
Control."
25:526-51.
A., and Barry R. Weingast.
A.,
J. Malbin.
Washington,
Quarterly
of Roll
Parties
Political
Kenneth
Dull.
Rosenthal.
University
Foundations
ries
of
Press.
W. 1991.
Chicago:
Theories
Studies
History
University
David
and Michael
2008.
Congressional
Howard
tutional
Shepsle,
of
Legislative
T.,
E. Mann,
Matthew
Politics
Political-Economic
Scher,
Stud
University
Press.
C. W., and
The
Quarrel:
Keith
the Bureaucracy:
Pittsburgh:
on Congress,
Institution
Brookings
Poole,
Oversees
Congress
Press.
The Journal
Press.
"At Risk:
Science
Evolution
of Political
House,
Norman
2008.
House.
Press.
Political
Review
in Legislative
Pittsburgh
Rohde,
Organiza
33:295-322.
Power
Morris.
ies
Oxford
1991.
Chicago:
Douglas,
Ogul,
1947-2004."
Press.
Committees
of Political
"Congres
versus
http://web.mit.edU/17.251/www/data_page.html#0
Parker,
Committees
Congressional
Journal
1984.
Patrols
Press.
Garrison.
Nelson,
Police
In The Managerial
Manager."
Ornstein,
B. Tatelman.
84:149-63.
Keith.
gressional
and
Schwartz.
C. 1999.
1947-1992."
Politics
and Todd
Congressional
Outliers?"
ment and Congressional
Landis,
in
Service.
"Are
of Preference
Stalemate
York:
Manual.
Turf Wars: How
Chicago:
Quarterly
Con
R. Weingast.
Barry
and Thomas
"The
Annual
P. Pfiffner, 2nd
of Pittsburgh.
and
J. Oleszek,
Jurisdiction.
making.
Kriner,
1987.
3:243-77.
and Policy."
Overlooked:
American
University
of Chicago
University
Oversight
1990.
Gains
Organization."
2002.
R. Baumgartner.
Research
tion. Ann Arbor:
Krehbiel,
Shefler.
3rd ed. New
Keith.
Review
Unequal
of the Administrative
Origins
of Law,
Mathew,
Ronald
Moe,
"The
Disagreement:
Prosecutors,
M., Walter
C. 1997.
1994.
University
Martin
Chicago:
Congressional
posed
Krehbiel,
and
D., and Frank
Congressional
Claim
and
Politics
G. Noll,
Roger
Oversight
Models."
19:181-214.
Strategic
Politics,
Frederick
King, David
2006.
of Political
& Organization
G. Noll,
Roger
Journal
Miller, Gary J. 2005.
"Organiza
Legislature."
of Legislative
Pittsburgh:
to Whitewater.
of Attention.
Krehbiel,
1995.
Means:
Watergate
2011.
and
of Ideology
As Instruments
and Process,
Political
Act."
sional
2,
April
34:531-64.
Krehbiel.
Quarterly
Politics.
Jones, Bryan
Keith
1990.
by a Rational
Science
Hypothesis
Benjamin,
Other
Kaiser,
and
Studies
John B.
Gilmour,
Krehbiel.
of Political
W.,
Exchange
Legislative
Keith
Committees
Journal
Thomas
from
Rosenthal.
Roger G. Noll, and Barry R. Weingast.
Mathew,
Alarms."
tion of Informative
Gilligan,
and Howard
75:431-82.
"The
McCubbins,
RL31836.
American
Haven:
15:180-217.
Louis.
Gilligan,
Control,
New
of Law, Economics,
Review
McCubbins,
1987.
Party
1946-1990.
Press.
Procedures
"Structure
Law
Connection.
We Govern:
The Dance
Mathew,
1989.
"Congressional
T. Poole,
MIT
Mathew,
McCubbins,
and
John,
Ferejohn,
Fiorina,
Boston:
trol." Journal
University
University
Press.
"Administrative
Making
Divided
America:
McCubbins,
to Get
Press.
Press.
Investigations,
Keith
Nolan,
Riches.
Delegating
and
Polarized
Press.
CQ
1999.
York:
1990.
1991.
The Broken
and How
The Electoral
Congress:
University
R.
University
McCarty,
S. Smith.
Yale
2008.
America
Oxford
Cambridge:
R. 1974.
Lawmaking,
J. Omstein.
Is Failing
Congress
on Track.
Mayhew,
http://www.archive.org/details/CPAPRoundtableonLeader
J., and Stephen
E., and Norman
How
It Back
to the Center for Public Administra
Roundtable
Thomas
Mann,
Press.
"Comments
tion & Policy
111th.
Press.
McCubbins.
Government
Party
versity of California
Davis,
D.
the
Anticipating
Institution
Brookings
W., and
Gary
Congress,
R. Weingast.
Institutions."
Institutions,
This content downloaded from 140.232.1.111 on Sun, 11 Jan 2015 04:29:52 AM
All use subject to JSTOR Terms and Conditions
edited
1995.
"Positive
In Positive
Theo
by Kenneth
A.
PoliticalResearch Quarterly66(3)
644
and Barry
Shepsle
sity of Michigan
Telford.
Taylor,
gressional
Tomz,
1955.
Grand
Investigations.
New
Weisman,
York:
Jonathan.
Ann Arbor:
Univer
The Story
Simon
and
of Con
and Shuster.
and Gary King.
for Interpreting
Available
sight Gaze."
Inquest:
Jason Wittenberg,
Software
Results."
5-35.
2001.
"Clar
Statistical
Presenting
"White
The Washington
House
Post,
Feels
Waxman's
October
Over
25. http://www
.washingtonpost.com/wp-dyn/content/article/2007/10/24/
AR2007102402757.html
West, William
Constraints
F. 1995.
Leonard
Controlling
in Theory and Practice.
Armonk:
James
Wilson,
Agencies
Institutional
M.E.Sharpe.
1951.
1989.
Q.
Do
The Jeffersonians:
1801-1829.
and
New
York:
Bureaucracy:
Why
Do
They
It.
A Study in Admin
Macmillan.
What
Government
New
York:
Basic
Books.
Wilson,
Woolley,
Woodrow.
1885.
Politics.
John T.
Hypothesis
1993.
Government:
Congressional
Boston:
"Conflict
of Congressional
Houghton
among
Dominance."
A Study
and Mifflin.
and
Regulators
Journal
the
of Poli
tics 55:92-114.
Zelizer,
the Bureaucracy:
D.
History,
in American
at http://gking.harvard.edu/clarify
2007.
White,
istrative
The
Michael,
ify:
R. Weingast.
Press.
Julian E. 2004.
and Its Consequences,
University
On Capitol
Hill:
1948-2000.
Press.
This content downloaded from 140.232.1.111 on Sun, 11 Jan 2015 04:29:52 AM
All use subject to JSTOR Terms and Conditions
The Struggle
Cambridge:
to Reform
Cambridge