Professional Documents
Culture Documents
Göran sundström
To cite this article: Göran sundström (2006) Management by Results: Its Origin and Development
in the Case of the Swedish State, International Public Management Journal, 9:4, 399-427
INTRODUCTION1
In recent years ‘‘public management policy’’ (PMP) has become an increasingly
popular research area in the academic study of public administration (Barzelay
2001, 2003; Pollitt and Bouckaert 2004). The term refers to government-wide policy
ideas and measures concerning government’s way of organizing and working. A part
of this research has focused on the policymaking process and the dynamics of
change. Scholars working within this tradition are interested in what triggers changes
in beliefs about problems and potential solutions, the mobilization of resourses to
accomplish change, the engagement of politicians, civil servants, and other stake-
holders, and the rise and resolution of conflicts (Barzelay 2003).
Often this research focuses on specific reforms. Through case studies covering
rather short time periods—usually one or a few years—researchers try to demon-
strate different mechanisms that can explain the reforms. This is the case both for
rational choice-oriented studies, where reforms are explained as rational adaptations
to changes in the economy and in technology, and for sociologically oriented studies,
International Public Management Journal, 9(4), pages 399–427 Copyright # 2006 Taylor & Francis Group, LLC
DOI: 10.1080/10967490601077178 ISSN: 1096-7494
400 International Public Management Journal Vol. 9, No. 4, 2006
where they are explained as imitation due to states’ disposition to follow rules and
act appropriately (March and Olsen 1989; Powell and DiMaggio 1991).
These kinds of studies have undoubtedly contributed to the development of a pro-
cess understanding of PMP reforms. However, just as within other policy areas,
there are cases of institutional stickiness within the PMP area. Some institutions
seem to be marked more by continuity than change, and case studies covering just
a few years are unable to capture such broader and more enduring patterns. Here,
historical institutionalists have provided a powerful correction by focusing our atten-
tion on forces that promote continuity and by developing analytical tools enabling
researchers to conduct longitudinal studies of particular practices.
However, historical institutionalism is a rather recent perspective.2 During the last
ten years, it has been discussed and elaborated. The discussion regards all sorts of
questions, from what ontological points of departure the perspective has, to the
way in which different analytical tools within the perspective are to be understood
and applied. Further, the perspective has been criticized for failing to illuminate
the mechanisms that keep institutions in place and of offering too few tools for ana-
lyzing how established institutions lose support and become de-institutionalised. It
had also been criticized for being too deterministic, giving too little room for actors
(Weir 2006). Also, studies where central analytical tools within the perspective are
used in a more systematic way, as well as more general discussions about how
to use the tools systematically in concrete empirical studies, are quite hard to find
(Thelen 2003).
A general aim of this paper is to contribute to the discussion on how to regard and
adopt the historical-institutional perspective in empirical studies of PMP reforms. I
will do so by analyzing the origin and development of what I analytically refer to as
the ‘‘rationalistic steering model’’ in Swedish government. The rationalistic steering
model is an interesting object of study because it has been one of the most important
ingredients in Swedish PMP. Today, labelled management by results (MBR), it is
meant generally to permeate Swedish public administration. Sweden is also con-
sidered to be among the leaders regarding both how far MBR has been developed
and in spreading MBR ideas and techniques to other countries. Responsible actors
in Sweden are quite proud of their achievement.
The rationalistic steering model is also an interesting object of study, since there
are different stories common in Sweden regarding how this institution was put in
place. The dominant story is a rationalistic one. According to this account, it all
started with the financial crises in the 1970s. In this connection the public sector soon
came under fire, criticized for being too big, too expensive, uncontrollable, inaccess-
ible, and generally difficult to understand. The government responded by initiating
two major ad hoc commissions, which investigated and discussed the problems at
hand and presented a solution—the rationalistic steering model, labelled MBR. In
1988, the MBR reform was launched in a budget bill, and since then the government
has been working hard implementing the model. MBR methods and techniques have
continuously improved through pilot projects and evaluations.
However, when talking to civil servants and listening to scholars—especially those
who have been around for a while—one can find an alternative story. Here, two
MANAGEMENT BY RESULTS 401
interesting points tend to be stressed. The first is that the origin of the steering model
isn’t to be found in the mid-1980s but in the early 1960s, when program budgeting
(PB) was tried out in Sweden. It is also stated that responsible actors since then—
obstinately, if not to say foolhardily—have been trying to develop and implement
the model. The second point is that actors responsible for the steering model—in
terms of development of ideas and methods, education, information, evaluation,
etc.—have been few and have formed a rather homogenous and secluded group,
dominated by career officials (Pollitt and Bouckaert 2004; Premfors 1999). These
officials have been accused of being unable to raise their eyes above the technicalities
occupying them daily and, thus, being unable to learn from experience.
The discussion above raises a number of questions worth posing:
. When was the rationalistic steering model introduced in the Swedish state? To
what extent has the development been marked by stability?
. How can the learning process be characterized—has the model been developed
and adjusted on the basis of experience with how it works in practice?
. If the model shows a high degree of stability, and if the learning process has failed,
how can this be understood?
The research presented here is gathered mainly from studying documents, fore-
most collected from agencies that have been responsible for developing and imple-
menting the rationalistic steering model. A large number of policy documents,
investigations, government bills, experimental work, evaluations, and publications
have been scrutinized. These have been complemented by some 40 interviews. The
interviewees have been strategically chosen and embrace both officials (top level
and lower) and politicians.3
This paper is divided into five sections. After this first—in which I also discuss his-
torical institutionalism and MBR—three sections follow where I discuss the three
questions posed above. In a final section I present my conclusions.
HISTORICAL INSTITUTIONALISM
As it is applied in this study, historical institutionalism fundamentally differs from
its two neo-institutional ‘‘cousins’’: rational-choice institutionalism and sociological
institutionalism.4 The latter two approaches start from fundamental and well
worked-out ideas about actors’ behavior: rational-choice institutionalism from
‘‘economic man,’’ acting from a logic of consequences, and sociological institution-
alism from ‘‘sociological man,’’ acting from a logic of appropriateness. These two
‘‘models of man’’ imply that the perspectives can be characterized as two more
full-blown social scientific theories (Premfors 2002). The overall goal of the research
within these theoretical perspectives is also to elaborate further and refine these
‘‘models of man.’’
Historical institutionalism, on the other hand, lacks a ‘‘model of man’’ of its own
(Pierson 1996). It has been stated, and I concur, that the intention is in fact not to
create a separate ‘‘model of man’’ or to decide which existing model is the best.
402 International Public Management Journal Vol. 9, No. 4, 2006
And for the juncture to be critical, the new way of organizing and regulating must
generate a new incentive structure influencing important actors to behave in such
a way that the political institution under study is maintained and reinforced.
Thus, criticality is one important feature of a juncture. Contingency is another.
That early historical events are contingent occurrences means that they cannot be
explained on the basis of prior historical conditions; the final outcome is unintended
and unpredictable (Mahoney 2000; Thelen 2003).
A third important feature of self reinforcing path dependencies is sensitivity.
Because events during a critical juncture set into motion institutional patterns that
have reinforcing properties, timing and sequentially are important features of path
dependency processes. As Paul Pierson notes: ‘‘Small events early on might have
big impacts, while ‘large’ events at later stages may be less consequential’’ (Pierson
1998, 6).
Critical junctures can be, and often are, assessed by counterfactual analyses (Sina-
tra 1996; Tetlock and Belkin 1996). Here, the investigator imagines that an alterna-
tive option had been selected and tries to rerun the course of events. The aim is
simply to illustrate the importance of the option actually selected by showing that
another option would have had dramatically different effects. As Mahoney notes,
this kind of analysis is especially persuasive when the investigator discusses options
predicted by theory to be selected but were not (Mahoney 2000, 513). Hence, the
specification of path dependence is always a theory-laden process. According to
Michael Sinatra, ‘‘Counter-factual arguments cannot really be made in the absence
of good theory, and attempts to do so are counterproductive’’ (Sinatra 1996, 34).
committees publish their findings in a final report. A commissioner has the right to
object to proposals in the report. Such an objection is to be attached to the report. A
committee proposal is circulated for comment before it is drafted as a government
bill. This gives government agencies, interest groups, and the general public an
opportunity to express their views on the matter.
MBR embraces two main ideas:one about delegation, the other about information.
According to the first, politicians should leave ‘‘smaller’’ decisions and those of a
more administrative character, such as concerning an agency’s internal organization
and staffing, to the agency. Such delegation will unload the politicians so they can
devote their precious time to politics, which is considered as involving formulating
goals and guidelines for the administration and following up results (Wilson 1887;
Goodnow 1900). Delegation will also increase creativity and efficiency, since the
agencies more freely can choose means in efforts to fulfill their goals.
The second main idea embraced by MBR is that agencies can be controlled better
through a certain kind of information flow. Such a flow is characterized by its cir-
culating form and its intensity. In special budget documents the politicians should
‘‘place an order’’ from each agency regarding both activities (through formulation
of objectives and results requirements) and information about results (through for-
mulation of reporting requirements). Objectives should include both activities and
outcomes (de Bruijn 2002). Further, the objectives should be as precise as possible,
preferably measurable and time specified. The agencies, in their turn, are to report
back their activities as well as their costs and results. On the basis of this infor-
mation, the agency should also make proposals regarding their future activities. This
information should form a basis for new objectives and results requirements.
The actors responsible for the development of MBR have not explicitly made clear
what basic assumptions the model rests on, but is possible to reconstruct them. A
first, and perhaps the most important, is that individual administrative units at a
rather low level can map out the causal connection between their own activities
and the results of these activities. A second is that the units can, without making
value judgments, report such information to higher administrative units, and on
to the central government, and also make impartial proposals regarding direction
and financing of their own future activities. A third assumption is that information
about results from lower levels can be aggregated at higher levels and be made the
base for new decisions on goals, result requirements and resource distribution. These
goals and result requirements can also—and this is a forth assumption—be made
clear and measurable by the politicians. A fifth assumption is that the public sector
can be divided into two relatively clear-cut and stable spheres: one political (where
goals and ends are discussed and stated) and one non-political (where administrative
action and means are discussed and chosen in order to fulfill the goals).
stability and change. This is not a surprising observation. It is hard to imagine any
political institution being totally unchanged during a period of 40 years. But how
profound have these changes been?
Neither the commission, nor RRV or Statskontoret (Swedish Agency for Public
Management)—the two agencies that became responsible for developing, imple-
menting, and evaluating PB—explicitly discussed the basic assumptions on which
the model rested. But by looking more closely at their discussions, and at handbooks
and guidelines produced by RRV and Statskontoret, it is clear that the five assump-
tions mentioned above, which today underpin MBR, also underpinned PB.
In 1977, the government proposed that PB should ‘‘be transformed’’ into what
was called the ‘‘state economic-administrative system’’ (SEA), a system RRV was
developing at the time (prop. 1976=77, 130). This transformation was seen as neces-
sary because of difficulties shown when pilot projects ware conducted with PB in the
early 1970s.
SEA was presented as a circle showing how ‘‘planning and budgeting’’ was to be
followed by ‘‘accounting and analysis of results,’’ which in turn was to be followed
by ‘‘auditing,’’ which led back to ‘‘planning and budgeting,’’ and so on (RRV
MANAGEMENT BY RESULTS 409
1975a). Just as with PB, the agencies were given a central position in SEA, doing
strict and recurring analysis of their own results, and sending proposals to the cabi-
net about their own future activities. On the basis of this information the cabinet was
supposed to make plans (short- and long-term) and to formulate objectives for each
agency. Hence, there were striking similarities between SEA and PB. In fact, RRV
characterized SEA as the Swedish version of PB.
Then, during the 1980s, SEA was replaced by MBR. And even though the govern-
ment tried to distance MBR from PB by claiming that implementation of MBR
should be done ‘‘from another point of departure than when working with PB,’’ it
was very hard to separate them. Here, the reader should compare RRV’s description
of PB from 1970 cited above with the government’s description of MBR from 1988:
The government shall determine the objectives and main directions for
state activities for a period of three years. The government shall also state
the resources available for the activities. Agencies are gradually to be
given increased responsibility for fulfilling their tasks by choosing means
themselves. This implies less detailed central regulation . . . The demands
on information about, and analysis of, results will sharpen, and focus
ought to shift from budgeting to follow-up activities and evaluations
(prop. 1987=88:150 appendix 1, page 69).
When described on this general level, MBR and PB seem to be almost identical
control models. This impression only increases when one reads legal texts, hand-
books, and other documents from different time periods in this study.
My conclusion is that the same five basic assumptions have been basic throughout
the whole time period. Surely some of them have been stressed more than others over
time, but none was ever abandoned. They were guiding officials in the 1960s when
trying to implement PB, in the 1970s when trying to implement SEA, and in the
1980s and 1990s when trying to implement MBR.
From the beginning, however, it has been perfectly clear that the experi-
ment simply has to be successful and that the major principles of PB have
MANAGEMENT BY RESULTS 411
come to stay. Only the forms and details, not the basic principles, have
been open to challenge and change (Andrén 1976, 351–352).
Parts of the SEA reform can be characterized in the same way. In 1978, RRV used
SEA as a point of departure when working out a new, ambitious budget handbook
to guide the agencies in their efforts to write budget requests. A pilot project was
conducted, but when the project was to be evaluated it became clear that the aim
was to ‘‘increase the interest for modernizing the budget system’’ and to ‘‘produce
data and impulses for the next edition of the budget handbook’’ (RRV 1979). The
handbook as such, and the basic ideas that formed its foundation, were not to be
put to the test.
The MBR reform was also accompanied by pilot projects. The commission pro-
posing the reform suggested that pilots should be conducted with its main proposal,
‘‘three year budgeting’’ (replacing annual with triannual budgets). But as soon as
results were collected, three year budgeting was to be ‘‘carried through on a full
scale’’ (SOU 1985:40, 78). Apparently, the results from the pilots had very little to
do with this.
Further, decisions about universalizing reforms have on several occasions been
made without pilot projects being awaited and reported. For example, the Program
Budget Commission finished its work without awaiting the results from pilot pro-
jects initiated by the commission itself. Later, when the projects were running, offi-
cials prepared handbooks and ran information campaign stressing the excellence of
PB.
Another example was the introduction of agency annual reports in the early 1990s.
This reform was preceded by a pilot project with what was called an ‘‘annual report
on results.’’ However, while this project was running, the government appointed an
ad hoc commission to examine introducing higher-quality annual reports, using the
private sector as a model. These two documents—the annual report on results and
the annual report—were very similar. However, the decision to introduce annual
reports on a full scale was taken without awaiting the results from the pilots with
the annual reports on results.
Finally, decisions about going ahead with reforms have, rather often, been made
even though results from pilot projects have been quite discouraging. This was, for
example, the case with the annual reports just mentioned. Here, the experiments
revealed severe difficulties in identifying and measuring activities, effects, and costs.
It was also shown that information about results wasn’t used at the ministerial level.
In the same manner, decisions to let PB morph into SEA in 1977 and to go ahead
with three-year budgeting in 1988 were taken even though the pilots had revealed
severe difficulties.
general character and have often been reports on attitudes, captured through sur-
veys, for example:
. that the agencies have become more conscious about the importance of results;
. that there is a growing interest for this steering model and its various techniques;
. that the interaction between ministries and the agencies has improved; and
. that the ministries have a better general view of the agencies.
After this discussion of successes the evaluators have, as a rule, discussed a num-
ber of setbacks. Unlike the successes, these setbacks have had a more concrete econ-
omic or administrative character, captured through studies of documents, for
example:
. that the politicians tend to meddle in administration (and are not able to uphold
the desirable distinction between the political and the administrative spheres);
. that the politicians state too few and too indistinct objectives and reporting
requirements;
. that the agencies fail to identify activities and to map out the causal connection
between activities and outcomes;
. that the agencies are disinclined to present proposals about savings to question
their own activities; and
. that information from the agencies about performance against targets isn’t used
by the ministries.
Finally, the evaluations have contained a set of solutions. Here, the uniformity has
been very high over the years. The solutions have been four in number:
1. Refine and specify methods and techniques.
2. Increase and improve information and education.
3. Increase engagement among politicians and officials.
4. Show patience — the reform is new and complicated.
A common feature of the evaluations is that these solutions haven’t been discussed
in relation to the setbacks. Their importance has been taken for granted. However,
by looking more closely at the solutions one may draw some conclusions about the
evaluators’ understandings. The first solution—refine and specify methods and tech-
niques—implies that the steering model is not yet fully built, or that there are (minor)
problems, which in turn means that politicians and officials are not able fully to meet
demands the model makes. The second solution—increase and improve education
and information—implies that the model and the methods are unclear, which in turn
means that politicians and officials are not able fully to understand demands placed
on them. The third solution—increase engagement—implies that methods and tech-
niques are in place but that politicians and officials are not interested in doing what
is required. The fourth solution—show patience—also implies methods and techni-
ques are in place but that politicians and officials need time to learn how to use them
(see Table 1):
As the reader notices, the solutions suggested have been quite shallow. Problems
have been regarded as temporary and have not generated more fundamental
MANAGEMENT BY RESULTS 413
TABLE 1
The Relation Between Proposed Solutions and Understanding of the Problems
analyses or questioning. When differences have arisen between model and practice it
has constantly been taken as an argument for the efforts to be even stronger. This is
perhaps an expected pattern of behavior in the short run. However, what is parti-
cularly interesting when studying the learning process over a longer period of time
is that responsible officials systematically have ignored experience gradually pro-
duced. Knowledge collected through evaluation has never—I emphasize never—been
compared to experience gained through earlier evaluations, and this despite the fact
that evaluations have been conducted by the same organizations over the years.
There are plenty of what a former senior official at RRV has called ‘‘cloned evalua-
tions’’ (Furubo 2003). Observations and formulations in evaluations from the 1970s
are very similar to those in evaluations from the 1990s. To a large extent the same
general successes have been reported (and emphasized), the same concrete problems
have been reported (and defused), and the same solutions have been suggested (with-
out further relating to the problems observed). And those responsible have not paid
attention to, or maybe not wanted to point out, that the successes and setbacks
reported, as well as solutions suggested, already have been made previously.
The evaluations have not been cumulative. The fact that the same problems have
been observed time after time has not led, for example, to studies designed to test the
validity of the steering model’s fundamental assumptions. Certainly there have been
studies in which the steering model’s assumptions and objectives have been discussed
and questioned. For example, over the years Swedish scholars on organizational
theory and political science have produced several papers and books criticizing the
rationalistic steering model (see, e.g., Tarschys 1978; Brunsson and Jönsson 1979;
Jacobsson 1989; Ehn and Sundström 1997). Now and then there have also been
‘‘think-tanks’’ within both RRV and Statskontoret that have criticized the model
(see, e.g., Statskontoret 1974; RRV 1976; Gorpe 1978; Brunsson 1988). But even
though many of these actors have worked very near those in charge of the model,
these critical discussions never really have influenced policymaking in this domain.
My overall conclusion regarding the second question posed in this paper is that the
learning process has been failing. Knowledge and experience regarding the con-
ditions for forming a functioning information and control system have not been
properly used by officials responsible for policy in this area. Experience has only
414 International Public Management Journal Vol. 9, No. 4, 2006
been used for single-loop learning, though there has been experience available that
could have brought about double-loop learning. In addition, even more experience
of that kind could have been collected if pilot projects and evaluations had been car-
ried out in a more professional way.
The Ministry of Finance, which was in charge of the question, took a liking to the
proposal and wrote a government bill on the basis of it (prop. 1960:126). According
to the bill, RRV and Statskontoret were to be closely aligned organizationally. Thus,
the director-general of RRV was placed on the board of Statskontoret, while the
director-general of Statskontoret was placed on the board of RRV. At the same time
the director of the Budget Bureau, located inside the Ministry of Finance, was placed
on the board of Statskontoret, a unique arrangement at the time. It was explicitly
stated that these arrangements were to facilitate cooperation and coordination
between the two agencies and the Budget Bureau.
Another central feature of the government bill was the extensive freedom of action
given to the two new agencies. The bill contained only vague directives regarding the
agencies’ working methods and internal organization. It was left to the agencies to
decide how to achieve the efficiency reforms and economy measures desired. It
was made clear, though, that RRV wasn’t to work with auditing only but also with
more general questions about steering, planning, and organizing. Many of these
questions were very close to questions assigned to Statskontoret. Further, both agen-
cies were encouraged to spread their knowledge among state agencies by informing
and educating.
When the two agencies started their work in 1961, expectations were high. But
after just a few years it became clear that the great plans were about to go into
pieces. The activities of RRV and Statskontoret had not changed to any appreci-
able extent compared to the agencies shut down just a few years earlier. In the
spring of 1963, the situation became critical; if new ideas and perspectives of a
more basic kind were not soon introduced, the agencies could be put into question
(Nybom 1980).
At this moment PB appeared like a savior. It was literally brought over from the
United States by a small group of Budget Bureau officials. Statskontoret got wind of
the model, and in 1963 Statskontoret wrote to the Minister of Finance asking for per-
mission to investigate the model together with RRV (SOU 1967:11). The request was
approved, and the government presented the directives of the Program Budget Com-
mission. But it should be noted that the request wasn’t fully met. The commission
was only allowed to examine the conditions of introducing the new steering model
at the agency level. In the request the ministries had been included. It was not until
later, in the 1970s, that the model became a tool for ministries to control agencies.
Statskontoret had requested the Program Budget Commission to work inside
Statskontoret, an unusual way to organize a commission. The government granted
this request. The commission was to be supervised by a group of experts. The general
director of Statskontoret became the chairman and the general director of RRV the
vice-chairman of this group. The director of the Budget Bureau was also assigned.
These members were accompanied by two management professors.
The government’s approval of Statskontoret’s request to examine PB in 1963
marks the end of the critical juncture. During the time between this decision and
the decision to initiate the Rationalization Commission eight years earlier, a number
of events occurred that, according to my interpretation, triggered three different
positive feedback loops of a power-based character.
416 International Public Management Journal Vol. 9, No. 4, 2006
Empire Building
The deadlock marking the Rationalization Commission cleared the way for two
ambitious and creative officials: Ivar Löfqvist of Statskontoret and Lars Lindmark
of the Budget Bureau and later of RRV. Early on they both knew that they were
to lead the two new modern agencies. In a way, the two built their own agencies.
Löfqvist and Lindmark were already top-level officials, and if we are to believe
Dunleavy (1991), they were not primarily aiming for higher wages or for leading
big organizations. Instead, they preferred smaller agencies working with interesting
problems and populated by ‘‘the best and the brightest.’’ Their ambition was now,
when they were standing on the highest rung on the career ladder, to accomplish
something lasting—to make a mark in history.
In 1963, when it became clear that the new agencies were having great difficulties
meeting the high expectations placed on them, Löfqvist and Lindmark had strong
reasons to look around for a solution—PB. This model provided their agencies with
new ideas about how politicians and officials could and should act, and how minis-
tries and agencies, and relations between them, could be arranged for higher
efficiency. And if the ideas were useful, extensive and long-lasting activities of differ-
ent kinds would be needed for Statskontoret, RRV, and the Budget Bureau. There-
fore, it is understandable that Löfqvist, Lindmark, and their associates had great
hopes for PB, and also early on tried to take control of it. The granting of Statskon-
toret’s request gave these actors a good opportunity to organize the Program Budget
Commission so that a proposal could be formed quickly, a proposal characterized by
consensus and clarity, that could convince skeptics about the superiority of PB and
make them realize that the model had to be introduced on a full scale.
These ambitious actors quickly took a firm grip of the process. Both Statskontoret
and RRV also came to invest heavily in PB. Using PB as a base, Lindmark both
reformed methods for RRV’s performance auditing and created a new large depart-
ment for the development of methods for control, budgeting, and running infor-
mation campaigns and training programs.
These early investments had long-term effects. Statskontoret and RRV soon
became experts on the new steering model. It became natural for the central govern-
ment to assign questions about methods development, implementation, training, and
evaluation to these agencies. Also, it became natural to send people from these agen-
cies to represent Sweden in different international arenas discussing control, plan-
ning, and budgeting questions. This provided the two agencies with more
resources and expertise, which in turn made it even more natural for the central
government to give them new assignments.
The important point here is that this ‘‘self-reinforcing’’ development was driven
actively by senior officials within Statskontoret, RRV, and the Budget Bureau. This
was not only because the model provided them with status and resources. It was also
important that the two agencies were built upon the rationalistic ideas they were
asked to develop and communicate. If these ideas had been found useless, it could
have weakened the position of the two agencies and their leaders, and of the Budget
Bureau.
MANAGEMENT BY RESULTS 417
The defenders of the rationalistic steering model have used different methods to
keep the process under control. One has been to (re)organize activities. As shown
above, they have for example invested heavily in the model. They have also managed
to sidestep several state commissions that have criticized the rationalistic steering
model, sometimes by simply ignoring them and sometimes by initiating working
groups of their own that have worked in parallel with the commissions. By referring
to these working groups they have managed to take the edge off critical proposals.
A second method used has been to (re)locate persons. A number of examples can
be found where those responsible for the rationalistic steering model have picked
people they know support the model for important positions—and sometimes
removed people they regard as ‘‘troublemakers’’ or ‘‘reactionaries.’’
Finally, and perhaps most importantly, those running the effort have managed to
keep a firm grip of the process by controlling the flow of information. Statskontoret,
RRV, and the Budget Bureau have been able to decide when to examine the steering
model, how to examine it and by whom, and how to interpret results from these
examinations. Very often they have chosen to examine the model themselves. And
the conclusion drawn, as noted, has to a large extent been that the effort has been
successful. At the same time they have repeatedly made the point that the reform
hasn’t yet been fully implemented and that more resources would be needed to finish
the work. They have also argued that these resources should be given to them,
because they have worked with these kinds of questions for a long time and therefore
possess unique expertise. What they haven’t noted is that they have made similar
promises about near breakthroughs before. Instead, earlier experience and expressed
hopes have been suppressed. Over and over the rationalistic steering model has been
presented as a new and fresh idea, and this has increased the engagement and the
hope of success among both politicians and officials.
If earlier experience has been suppressed, contemporary experience has been
avoided. As mentioned earlier, pilot projects and evaluations have been handled
quite poorly. Problems have not been thoroughly discussed, and solutions have
mainly involved fine-tuning—largely putting new labels on old techniques. Officials
have also managed to sell—and defend—the model using general statements about
the importance of formulating objectives, following up results, and delegating deci-
sions from the ministerial to the agency level. By pointing at the simple and obvious,
and avoiding a discussion about the more far-reaching and specific assumptions and
objectives characterizing the model, they have succeeded in putting the burden of
proof on critics. They have constantly rushed on, hoping that refinements and a
changed manner of speaking will make problems disappear. This has made the steer-
ing model a moving target.
Groupthink
Decisions made during the critical juncture did not only give central actors the
incentives and possibilities to defend the rationalistic steering model but also to work
closely together. From the beginning Statskontoret and RRV were directed to work
418 International Public Management Journal Vol. 9, No. 4, 2006
on very similar tasks. The idea was to facilitate cooperation between the two agencies
and the Budget Bureau; it was not to produce checks and balances between them.
These initial arrangements had important long-term effects. The three organiza-
tions that dominated the Program Budget Commission stayed in control through
the whole course of events. In every given moment the process has also been domi-
nated by only a few individuals, among whom interactions have been intense and
informal.
In the light of these facts, it is plausible that theories about groupthink can be use-
ful for understanding the course of events. The process has been marked by several
conditions often pointed out as crucial for groupthink to emerge. Participants have
been handling quite complex questions where clear answers are hard to find. At the
same time, they have been working in rather attractive environments offering posi-
tions with high status, relatively good wages, and stimulating work. Furthermore,
there have been a number of strong-willed and charismatic people in leading posi-
tions within these groups.
In this environment, criticism against the rationalistic steering model has had dif-
ficulties taking root. Early on the actors got committed to one way of thinking about
steering and controlling state activities: the opinion arose that there were only two
ways of controlling state activities: through an old-fashioned steering model, based
on detailed rules, or through a modern rationalistic steering model, based on objec-
tives and results. This ‘‘dualistic’’ thought has been very strong over the years (and
still is), and it has almost automatically made the critics reform enemies and reac-
tionaries in the eyes of group members.
The study also shows that the group members have kept to themselves and
refrained from establishing proper procedures for scrutiny and for information
search that could have challenged dominant ways of thinking. The environments
have been secluded. At times the ‘‘in-group’’ thought has been so prominent that
members have been subject to euphoria and ignorance. Absolutely sure of the legit-
imacy of their own mission, group members haven’t considered it necessary to have
their own ideas tested. By talking to each other, the certainty that dominant ways of
thinking about problems and solutions within the group are superior has only
increased. It should also be noted that those who have left the group soon have
stopped being euphoric and instead turned critical. Typical for groupthink is that
the feeling of hubris and arrogance is temporary and that people tend to sober up
as soon as they leave the group.
Professional Expertise
Finally, the study shows that one particular profession came to dominate the
course of events as a consequence of different decisions made during the formative
phase. Here, the way the Program Budget Commission and a subsequent ad hoc
commission—the Budget Commission—were put together was of great importance.
They were highly dominated by people specializing in accounting and budget tech-
niques. The commissions introduced a large number of economic concepts, mainly
MANAGEMENT BY RESULTS 419
efficiency. This idea wasn’t unknown in Swedish public administration, but now it
was used in a much wider sense than before and as a value on par with other tra-
ditional values such as democracy, transparency, and the rule of law (Amnå 1981).
The study clearly demonstrates how the massive introduction of economic theory
developed into a gradually strengthened position for the economics profession.
From the start the steering model was dressed in an economic-theoretical linguistic
garment not very easy for outsiders to understand (Andrén 1976). Reports, hand-
books, and brochures throughout the years have featured complicated models and
diagrams. And officials and politicians have repeatedly complained that the steering
model was too complicated, time-consuming, and something mostly professional
experts cared about.
In order to meet the demands of the steering model, agencies have been forced to
set up special units led by professional experts. Often the head of these units have
been recruited from RRV or Statskontoret. This has generated a growing pro-
fessional network, reinforced by frequent informal contacts, and by recurrent infor-
mation and training campaigns led by RRV and Statskontoret. These professional
experts came gradually to enjoy increased status. During the 1970s they quickly
climbed from the lower levels to become senior officials. All this had the effect that
this economics profession took an even stronger grip on the process, which in turn
increased demand for these experts among agencies.
The fact that the process has been dominated by a particular profession can help
us understand the course of events. That many problems haven’t been discussed in
greater detail, and that the relations between problems and proposed solutions
haven’t been clarified, can be interpreted in the light of theories about professions:
The professional experts in charge haven’t considered it necessary to explain their
ways of thinking and acting in more detail to ‘‘outsiders,’’ and especially not to poli-
ticians. Why should they? The ‘‘outsiders’’ wouldn’t understand the complex
material anyway.
Those in charge of the effort have been more disposed to turn to their professional
colleagues. This has certainly been the case when they have run into problems. This
can help us understand why the same solutions have been proposed time after time,
and that the experts in charge have devoted more time to refining and adjusting the
model than trying to understand the problems that have appeared recurrently—the
paradigm held by the profession simply hasn’t offered any alternative solutions.
Surely, many of the changes made can have been perceived as profound by the pro-
fessional experts, but to an outsider it looks like the wheel has been reinvented time
after time.
A Counterfactual Argument
I have now, in some detail, argued that the period between 1956 and 1963 was
marked by criticality by showing that a number of events occurring during these
eight years triggered three different, power-based positive feedback loops that since
then have reinforced the rationalistic steering model. Here, I will argue that these
events were also contingent.
420 International Public Management Journal Vol. 9, No. 4, 2006
If the events occurring between 1956 and 1963 were not contingent, we must be
able to explain them by the same forces that later supported the rationalistic steering
model. That means that it should be possible to explain them by: 1) senior officials
trying to build themselves ‘‘empires’’; 2) a tight secluded group of officials stuck on
certain ideas about steering and control; and 3) a strong, specialized economics pro-
fession. This is not a very probable explanation, at least not during the first years of
this period. In 1956, when the Rationalization Commission was set up, no such tight
group and no such profession were in control of Swedish PMP. And even if Ivar
Löfqvist and Lars Lindmark already then had quite strong positions, it is not very
likely that they had the power or skill to purposefully put together the commission
in order to create the deadlock that followed.
I argue that the deadlock was unintended. When populating the commission with
people from the agencies to be reviewed, the government was simply trying to use the
knowledge and experience existing within these organizations. It was not trying to
produce a deadlock—that was an unforeseen consequence. So was, therefore, the
space for maneuver given to Löfqvist and Lindmark. One can certainly argue that
the high level of ambition combined with the wide bailiwick and the extensive free-
dom of action given to the new agencies should have called for more reflection. For
example, it was hardly unimaginable that it might be problematical to intertwine the
two new, highly progressive and advanced (as they were believed to be), agencies in
the name of cooperation and coordination.
Perhaps the risks would have been given more attention if the Rationalization
Commission had been put together in another way. One feasible alternative was
to make the commission a parliamentary one, complemented by experts. At the time
most important commissions were parliamentary. Such an alternative composition
would probably have increased the chances of not ending up in a deadlock and of
getting different arrangements more properly discussed.
In this alternative scenario it is also quite possible that the Program Budget Com-
mission, which was set up in 1963, would have been arranged differently. The
arrangement selected—to organize it inside Statskontoret—was an unusual one,
which made it possible for the group of people ‘‘first out of the gate’’ to see to it that
the commission was put together and worked in a way that ensured a smooth ride.
Again a parliamentary commission would have been a most feasible alternative.
With a parliamentary commission—composed of politicians, officials from different
agencies, and experts of different kinds — perhaps the proposal would not have had
such a strong economic bias. And maybe empirical studies would have been ascribed
more value; after all, criticism against PB was not lacking.
Here, we can also ask if it’s possible to begin the counterfactual argument in the
early 1960s. Was the decision to organize the Program Budget Commission accord-
ing to Statskontoret’s wishes contingent even in a situation with the course of events
prior to that decision? This is more uncertain. In my view, forces had—just because
of earlier events—begun influencing the process in a way that made an alternative
decision less likely. As discussed above, at the time there was a group of committed
people acting in favor of the decision actually made. However, it’s quite hard to
determine the strength of this group, since there also was a group inside the Ministry
MANAGEMENT BY RESULTS 421
of Finance highly unsure of the superiority of PB. The Minister belonged to that
group. And the fact that he didn’t fully grant the request from Statskontoret—he
only gave permission to examine the possibility of introducing the model on an
agency level—indicates that there perhaps was a possibility to arrange the Program
Budget Commission in an alterative way. However, to redirect the course of events
more permanently, I think such an alternative decision would have had to be com-
plemented by a clarification of, and a separation between, the roles of RRV and
Statskontoret and by stressing the need for each to check the other, preferably by
making RRV a pure auditing agency. That was much harder to do in 1963 than
in 1960, when the decision was made to set up the two agencies. Perhaps we can con-
clude by saying that the decision made in 1963 was marked by some contingency, but
it was rapidly fading.
I want to underscore that I am not suggesting that decisions made during the critical
juncture explain why Sweden adopted PB. In any case, Sweden would almost certainly
have tried PB at some point. After all, Sweden wasn’t the only country to try the
model. The Swedish government also had a strong belief in central planning and social
engineering, which goes quite well with models such as PB. Furthermore, Sweden was
quite America-oriented at the time, and the idea was seen to come from the U.S. How-
ever, I do believe that the rationalistic steering model in general would have evolved in
a different way if other decisions had been made during the critical juncture. If critics
had been able to get their voices heard, and if those leading the effort had had stronger
incentives to develop the model on the basis of experience about how it worked in
practice, the model probably would not have gotten such a strong foothold.
CONCLUSIONS
In this paper I have discussed the origin and development of management by
results, which today is meant be applied as a general steering model in Swedish
government. The study shows that this public management policy (PMP) reform
hasn’t primarily been a story about rational adaptations to world-wide changes in
the economy and technology, or imitation due to government’s disposition to follow
rules and act appropriately. The course of events is better understood by forces of
inertia inherent within Swedish public administration and the way the PMP sector
has been organized and regulated. Different initial decisions made in the early
1960s, when the rationalistic steering model was introduced under the name of pro-
gram budgeting (PB), can be interpreted as a critical juncture creating positive feed-
back loops, which were to have a stabilizing effect on the model. This doesn’t mean
that historical institutionalism can explain all PMP reforms. It merely shows that pol-
icy reforms sometimes get locked in by forces making institutions ‘‘sticky.’’ Scholars
trying to develop a process understanding of PMP need to take this into account.
Thus, historical institutionalism should be seen as a complementary perspective to
be used along with other ones.
The paper illustrates how an empirical study can be done using a historical insti-
tutional perspective. For example, in assessing a critical juncture I have stressed the
422 International Public Management Journal Vol. 9, No. 4, 2006
importance of applying some energy trying to demonstrate that the juncture has been
marked by criticality, contingency, and sensibility. The study also shows that it can
be difficult to separate the constituting part of the path dependency (the critical junc-
ture) from the reproducing part (positive feedback). In this case three different posi-
tive feedback loops have supported the institution. These forces were triggered at
different times during the critical juncture. The ‘‘empire force’’ was triggered first
due to the way the Rationalization Commission was put together. It created certain
conditions necessary for the other two forces to be triggered. Here, the process dur-
ing the critical juncture bears traces of a reactive path dependency. However, these
events were marked by enough contingency to stretch out the critical juncture to
1963.
Further, the study shows that a critical juncture can be quite long—eight years in
this case. Here, the concept ‘‘critical juncture’’ (or ‘‘formative moment,’’ another
expression sometimes used) isn’t that apt. It can foster an idea that the researcher
has to identify some instantaneous event. Perhaps we should talk about critical or
formative phases.
Regarding the three positive feedback loops, they have all been power-based, but
they were interpreted from different power perspectives. Hence, regarding motives,
intentions, and rationalities, the actions underpinning developments have been quite
varied. This might give rise to the question of how these three forces relate to each
other. Can the three perspectives really be combined when explaining the same
course of events? Is it possible that participants have supported the rationalistic
steering model by acting both strategically and with their own wishes and desires
in mind, as assumed in the ‘‘empire force,’’ and non-strategically and with the pub-
lic’s wishes and desires in mind, as assumed with the other two forces? I think it is
possible. The study shows how this problem can be handled within the historical
institutional perspective. Firstly, important actors have been studied on different
levels. This has made it possible to interpret the actions of these actors at different
levels from different power perspectives. For example, the ‘‘empire force’’ may be
more applicable to senior officials than to lower ones, while the opposite applies
for the other two forces. Secondly, the problem is handled by making use of history.
All three forces haven’t been equally prominent throughout. Their importance has
varied over time. The ‘‘empire force’’ was most important in the beginning, while
the ‘‘professional force’’ became more important later on.
This way of reasoning might puzzle the reader. I have earlier talked about self-
reinforcing path dependencies. But here I am saying that one force actually has
decreased and another varied in importance. How can that be? Here, the term
‘‘self-reinforcing’’ can be misleading. It has a strong deterministic connotation.
However, the kinds of path dependencies studied here do not proceed automati-
cally. After all, we are dealing with social processes, and the study shows that it
takes active actors for the course to maintain its direction. For example, the
‘‘empire force’’ has been exercised by actors actively reorganizing activities and
persons, and controlling the flow of information. In the same way it has taken
active actors for the other two forces to proceed. And it is really this level of activi-
ties that has decreased (in the case of the ‘‘empire force’’) and varied (in the case of
MANAGEMENT BY RESULTS 423
the ‘‘groupthink force’’), not the government’s disposition to assign missions and
direct funds to these actors, making it possible for them to uphold the institution
of study.
NOTES
1. This paper is part of a larger study, which is only available in Swedish: Stat På villovä-
gar: Resultatstyrningens framväxt i ett historisk-institutionellt perspektiv (Sundström 2003).
(The Wayward State: The development of Management by Results in a historical institutional
perspective). At a number of points in this paper I will draw conclusions from empirical
material reported in the larger study but not in this paper. I would like to thank Rune
Premfors and two anonymous reviewers for giving useful comments on this article. A special
thanks to Steve Kelman, who has given useful comments on the paper and made language
corrections.
2. The anthology, Structuring Politics. Historical Institutionalism in Comparative Analysis
from 1992, by Steinmo, Thelen and Longstreth, is usually considered a starting point.
3. Most of the interviewees were identified through documents, i.e., commission reports,
evaluations, government bills, and handbooks. Some interviewees found through documents
have also suggested other interviewees.
4. For some comparisons between the three perspectives, see Hall and Taylor 1996 or
Peters 1999.
5. Another one is reactive or evolutionary path dependency. Unlike self reinforcing path
dependencies, where an early institutional pattern is reproduced and stabilized over time,
reactive path dependencies imply that early contingent events starts a chain of reactions giving
the development determining properties. The development is reactive in the sense that each
occurrence in the course of events partly is a reaction to the previous occurrence; each occur-
rence in the course of events is dependent on the previous occurrence and the cause of the fol-
lowing one (Mahoney 2000). Katharine Thelen discusses what I would like to point out as a
third kind of path dependency (Thelen 2003). Perhaps it can be called incorporative path
dependency. Like self reinforcing path dependencies, Thelen is interested in political institu-
tions that persist over long stretches of time. But here institutions are not underpinned by sta-
bilizing and self reinforcing forces triggered early on in the process and which then stay
constant through the whole course of events. Instead, the forces supporting the institutions
change over time. Some institutions seem to survive because new actors are incorporated in
the process while it proceeds. These actors may succeed or act side by side with the actors
who previously supported the institution, and they have other motives, other possibilities,
and other ways of exercising their powers. Also, they may add new properties to the institution
in order to bring it in line with changing social, political, and economic conditions—and to
make it more useful to them. This can also include a redirection of the institution to a new
set of goals. Thus, much change can be going on beneath the surface of apparently stable for-
mal institutional arrangements. But even if the institution is ‘‘conquered’’ by new actors, the
institution as such will not be abandoned or destroyed. Thelen refers to this as processes of
‘‘layering’’ (additional institutional arrangements) and ‘‘conversion’’ (a redirection of the insti-
tution to a new set of goals).
6. Following Peter Hall, social learning is here defined as ‘‘deliberative attempts to adjust
the goals or techniques of policy in response to past experience and new information’’ (Hall
1993, 278).
424 International Public Management Journal Vol. 9, No. 4, 2006
7. At present there are nine ministries containing all together about 4,600 officials, of
whom around 120 are political appointees. Over 1,500 officials are working in the Ministry
of Foreign Affairs and about 60 in the Prime Minister’s Office.
8. According to the agency Statskontoret, there were 547 agencies in December 2005
(Statskontoret 2005).
9. This restriction has two components: First, decision should, as a main rule, be taken by
collective vote in the full Cabinet (at least five ministers need to be present). This means that an
individual minister is prohibited from issuing orders for agencies under his or her purview.
Secondly, in cases where an agency in the capacity of a public authority, and on the basis
of public law, decides on either rights or obligations for an individual citizen the agency is
to be guided only by the law. In such cases no one, not even the Cabinet as a collective or
the parliament, can decide how the agency should decide.
10. Over all, budgeting means giving each agency its appropriation for administrative
costs in ‘‘a bag’’ and letting them decide how to distribute the money among costs for prem-
ises, staff, education, etc.
REFERENCES
Amnå, E. 1981. Planhushållning i den offentliga sektorn? Budgetdialogen mellan regering och
förvaltning under efterkrigstiden. Norstedts.
Andersson, C. 2004. Tudelad trots allt–dualismens överlevnad I den svenska staten 1718–1987.
Stockholm University: Department of Political Science.
Andrén, N. 1976. ‘‘Program Budgeting: The Swedish Experiment.’’ in A. F. Leeman, ed., The
Management of Change. Haag: Institute of Social Studies, Martinus Nijhoff.
Argote, L. 1999. Organizational Learning: Creating, Retaining, and Transferring Knowledge.
Kluwer Academic.
Argyris, C. 1990. Overcoming Organizational Defence. New York: Allyn & Bacon.
Barzelay, M. 2001. The New Public Management. Improving Research and Political Dialogue.
Berkeley: University of California Press.
Barzelay, M. 2003. ‘‘Introduction to the Process Dynamics of Public Management Policymak-
ing.’’ International Public Management Journal 6(3): 251–282.
Bennett, A. and A. L. George, 1997. Process Tracing in Case Study Research. Paper presented
at MacArthur Foundation Workshop on Case Study Methods, Belfer Center for Science
and International Affairs (BCSIA), Harvard University, 17–19 October 1997. (www.geor-
getown.edu/bennett/PROTCG.htm)
de Bruijn, H. 2002. Managing Performance in the Public Sector. New York: Routledge.
Brunsson, K. 1988. Ny budgethandbok? RRV, memo, diary no 1988: 462.
Brunsson, N. and S. Jönsson, 1979. Beslut och handling. Om politikers inflytande på politiken.
Stockholm: Liber Förlag.
Christensen, T. and P. Lægreid, 2006. Autonomy and Regulation. Coping with Agencies in the
Modern State. Cheltenham: Edward Elgar Publishing Ltd.
Downs, A. 1967. Inside Bureaucracy. Boston: Little, Brown.
Dunleavy, P. 1991. Democracy, Bureaucracy and Public Choice. Economic Explanations in
Political Science. Hemel Hempstead: Harvester Wheatsheaf.
Ehn, P. and G. Sundström, 1997. ‘‘Samspelet mellan regeringen och statsförvaltningen,’’ in
Det svåra samspelet, SOU 1997:15.
Feldman, M. S. and B. T. Pentland, 2003. ‘‘Reconceptualizing Organizational Routines as a
Source of Flexibility and Change.’’ Administrative Science Quarterly 48: 94–118.
MANAGEMENT BY RESULTS 425