Stovepiping
| Author | John W Patty,Sean Gailmard |
| Published date | 01 July 2013 |
| Date | 01 July 2013 |
| DOI | http://doi.org/10.1177/0951629813477277 |
| Subject Matter | Articles |
Article
Stovepiping
Journal of Theoretical Politics
25(3) 388–411
©The Author(s) 2013
Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
DOI:10.1177/0951629813477277
jtp.sagepub.com
Sean Gailmard
Charles & Louise Travers Department of Political Science, University of California, Berkeley, CA, USA
John W Patty
Department of Political Science, Center for New Institutional Social Sciences, Washington University in Saint
Louis, St Louis, MO, USA
Abstract
In hierarchical organizations lower-level agents can often censor the information that a higher-
level principal has available to make a decision. We present a model of this interaction in which
the principal can also access an independent source of unfiltered but lower-quality information
besides that provided by the agent. This provision of outside information can be thought of as
‘stovepiping,’ the transmission of unfiltered information from analysts directly to decision-makers.
Stovepiping can, in equilibrium, result in the agent passing along more information to the principal,
precisely because the outsider’s information is of lower quality than the agent’s. But it can also
lead the agent to ‘destroy’ information so that there is no basis for any policy change. Accordingly,
stovepiping has countervailing effects on the principal’s utility. We discuss the comparative statics
of equilibrium levels of information transmission with respect to the preferences of the principal,
the outsider, and the probability that the outsider has access to the information held by the agent.
Keywords
Advice; information; transparency; hierarchies
... what the Bush people did was ‘dismantle the existingfiltering process that for fifty years had
been preventing the policymakers from getting bad information. They created stovepipesto get
the information they wanted directly to the top leadership. Their position is that the professional
bureaucracy is deliberately and maliciouslykeeping infor mation from them... Theyalways had
information to back up their public claims, but it was often very bad information...’. (Hersh,
2004, pp. 223–224, quoting Kenneth Pollack, a former National Security Council expert on
Iraq)
Corresponding author:
John Patty, Department of Political Science, Center for New Institutional Social Sciences, Washington Uni-
versity in Saint Louis, 1 Brookings Drive, Campus Box 1063, St Louis, MO 63130, USA.
Email: jpatty@wustl.edu
Gailmard and Patty 389
Hersh’s description of the decision making within the Bush Administration in the
months leading up to the invasion of Iraq in 2003 is clearly disturbing in ex post terms:
the certainty with which Donald Rumsfeld and his colleagues apparently believed that
Saddam Hussein’s regime possessed weapons of mass destruction contrasts starkly with
the lack of evidence discoveredafter the militar y intervention. But this point merelyhigh-
lights a potential argument in favor of political decision makers1possessing control over
the processes by which they are presented with decision-relevant information. In insti-
tutional terms, such control was exercised by Rumsfeld himself through his redirection
of the Northern Gulf Affairs Office. Retitled the Office of Special Plans (OSP) in 2002,
this office reviewed information that had been collected through both public and confi-
dential channels, although it collected little to no new intelligence on its own, as it was
constituted as a policy group.2
The OSP was responsible for forwarding intelligence reports to the Secretary of
Defense that were pertinent in its members’ opinions to the planning for how best to
deal with Iraq, particularly the possibility of direct military intervention. In this sense
it illustrates stovepiping: an information transmission process in organizations in which
decision makers obtain direct access to (a subset of) the same data that their bureaucratic
subordinates possess, but without the data filtering and processing of data that subordi-
nate analysts typically perform. Dreyfuss and Vest (2004) provides a direct illustration of
how the OSP operated as a stovepipe:
raw intelligence is usually subject to a thorough vetting process, tracked, verified, and checked
by intelligence professionals. But not at OSP—the material that it produced found its way
directly into speeches by Bush, Cheney, and other officials. (Dreyfuss and Vest, 2004)
In the case of the debate leading up to the invasion of Iraq, it was clear that a stovepipe
did not lack for smoke. As an unnamed former CIA official is quoted by Hersh (2003),
‘[w]e collect so much stuff that you can find anything you want.’ After the Department
of Defense’s Inspector General had proffered their postmortem report about the handling
of intelligence in the lead-up to the invasion of Iraq, Senator Jay Rockefeller (D, WV),
the then Chairman of the US Senate Select Committee on Intelligence, described the
activities of the OUSD(P) in the following very clear terms:
[i]ndividuals in that office produced and disseminated intelligence products outside of the reg-
ular intelligence channels. These intelligence products were inconsistent with the consensus
judgments of the Intelligence Community. This office did this without coordinating with the
Intelligence Community and as a result policy-makers received distorted intelligence. (Senator
Jay Rockefeller, press release, 8 February 2007)
Of course, while captivating, the events of 2002–2003 are not the first time that the
importance of information in determining public policy has been observed. Indeed, it
has long been acknowledged that a critical source of bureaucratic power is a monopoly
over the provision of decision-relevant information to a principal (Weber, 1946; Niska-
nen, 1971). For instance, in regulatory agencies, line bureaucrats acquire information
from notice and comment proceedings, investigations of complaints, petitions for action,
offeror provisions, incidental learning through the enforcement of existing regulations,
analytical studies of regulated industries, discovery during judicial review and enforce-
ment of agency actions, reviews by the Officeof Management and Budget (OMB), audits
by the Government Accountability Office (GAO), and congressional hearings. In policy-
making agencies, staff bureaucrats acquire information about policy options and their
To continue reading
Request your trial