This article focuses on decentralization as a governance strategy in education. The author, Gustav Karlsen, first presents theoretical aspects of the phenomenon of decentralization and contextualizes them within the education research literature. He then analyzes the dynamics of decentralization that normally take place when governing strategies designed to decentralize authority and power are implemented. In particular, he emphasizes four aspects of that dynamic interactive process. In support of his argument, he provides evidence gathered from Norway and, to some extent, from the province of British Columbia, Canada.¹ The dynamic interaction in the decentralization process, Karlsen contends, is implicit in the title of the article, "decentralized centralism."²

Decentralization has appeared on the political agenda for almost three decades and has become a worldwide trend. As a government initiative, decentralization has been evident in Western societies as well as in developing countries (Rondinelli et al. 1983). Governments have used decentralization as a strategy in various institutions within the public sector, including education where there is a prevalent belief that decentralization will bring about desired large-scale educational reforms.

My recent study of the decentralization process in education in Norway (Karlsen, 1993) revealed that arguments favoring decentralization policy were first evident in the late 1960s and became very strong by the late 1970s. The decentralization movement in education in Norway continued during the 1980s and 1990s, although the arguments and the nature of the decentralization movement itself changed over time.

Four key arguments underpin the decentralization movement in Norway. Three of these arguments were strong during the whole period, but weakened in the late 1980s and into the 1990s. The initial argument favoring decentralization policy was supported by the claim that decentralization would strengthen democracy by transferring power from central to local bodies and by bringing the decision-making process closer to the people. This belief was expressed in the platforms presented by political parties in Norway in the 1970s (Kjøl & Telhaug, 1979) as well as in the governments' official curriculum documents. The Curriculum Guidelines of 1974³ and particularly the Revised Guidelines of 1987 outlined the common basic principles and objectives of compulsory education, but at the same time gave the schools and the teachers a great deal more freedom than they had previously experienced. As early as 1969, the Educational Law for Primary and Lower Secondary Schools gave the community and the students the right to participate in decision making at the school level.

The second argument was linked to an understanding of the innovation process and school development. Over the course of the decentralization movement, for example, the traditional top-down governance and innovation strategy was gradually abandoned for a more decentralized bottom-up strategy. The favoring of the bottom-up strategy was based on the belief that true implementation would only occur if those involved at the grassroots level supported the innovation or reform (Østerud, 1991). Local initiative and local participation became important. Making political decisions at the top level did not necessarily guarantee that the reform was implemented at school level. The assumption was that people would accept a decision and be willing to abide by it if they were responsible for making that decision. As a result, the individual school was then considered a critical, important organizational unit. Decentralization was perceived as a necessary condition for school-based development.

The Curriculum Guidelines of 1987gave the local school freedom to develop local curricula and to adapt the guidelines to local needs and conditions. But even though the bottom-up strategy is still the accepted rhetoric, there has been a shift toward a more traditional top-down government strategy in the 1990s. The new curriculum guidelines in Norway, which came into effect in 1997, are mainly the result of a central initiative to bring back the top-down strategy.

The third argument favoring decentralization characterized it as a strategy for strengthening the local culture, local businesses, and the local community as a whole. The claim was that decentralization made it possible for the local school to design programs and activities better adapted to the needs of the local community. It was argued that a more flexible and locally oriented school had a positive effect on students' motivation and learning, gave them a feeling of belonging to the local community, and made them aware of the role they had to play in the community. Underpinning this argument was the belief that the local school should first of all be an agent for the local community and not for the larger society (Solstad,1994).

During the 1970s and 1980s there was a move toward a less standardized and more locally oriented curriculum. The Curriculum Guidelines of 1987 opened up the curriculum to include local knowledge and local culture. But the new Curriculum Guidelines of 1997 changed this endeavor by focusing on a national standardized curriculum and stressing a more academic and skill-oriented education.

The fourth argument has become stronger throughout the 1980s and 1990s with the growing perception that decentralization is an effective governance strategy for achieving rationalization and efficiency. Underlying this argument was the belief that local authorities and the individual schools had the competence needed to use existing funding in a more flexible and efficient way and even obtain new local resources. Decentralization from this perspective is understood in a more market-oriented way. The argument here is that schools should be more like the market system and that independence and local autonomy give schools the same opportunity as other businesses to compete in the marketplace. Decentralization is characterized as a strategy for a more privatized and commercialized school. This argument reflects neo-liberal ideology which has often been termed "New Right" and "Thatcherism" (Lawton, 1992). It is a reaction to the economic stagnation that occurred in the Western world during the late 1970s and 1980s. While public funding for education was expanding in the 1960s and early 1970s, the oil crises led to attempts to reduce public spending. From this perspective, the decentralization of authority to the local level has become an important strategy for making schools more accountable and efficient.

Explaining the Underpinnings of the Decentralization Policy in Norway

Hans N. Weiler's (1990) explanation of the underlying motives for decentralization clarifies the process within the educational context in Norway. He explains the different motives by dividing decentralization initiatives into three categories. The first category of initiatives is characterized by the redistribution of authority from central to local levels. Normally this occurs through legislative action or allocation of resources. The second category includes initiatives where decentralization is used as a strategy to obtain more efficiency in the educational sector. The assumption is that decentralization will mobilize local resources and the use of available resources in a more efficient way. The third category focuses on initiatives where decisions concerning the curriculum and culture of learning in schools are made locally by those most affected. Decentralization of these decisions provides more room for local variance and relevance and, it is believed, more motivated students and a better culture of learning.

Weiler (1990) has a rather critical view of decentralization strategies and he finds that there is tension between decentralization efforts and the need for central control. After studying several educational systems, he concludes that decentralization seldom actually occurs, but seems instead to be absorbed into the existing centralized or semi-centralized structure of educational governance. Still, he sees decentralization as important as a way to manage conflict and to give what is called "compensatory legitimization." That is, the modern state has to compensate "the erosion of legitimacy" and decentralization rhetoric is a way to enhance its legitimacy and thereby maintain power.

Another explanation is presented by Jon Lauglo and Martin McLean. In their book The Control of Education(1985), these authors describe three rationales behind decentralization. The administrative rationale has the underlying assumption that administration in a centralized system is unnecessary, extensive, elaborate, and slow working. From this perspective, decentralization is a strategy to avoid these effects. In the political rationale, decentralization is seen as a way to reduce professional power and to maintain and even reinforce political control and power. Giving more power to parents and the local community is one way to achieve this. The ideological rationale is based on some fundamental assumptions about human beings and the society in which the individual is perceived as responsible and holistic. From this perspective, decentralization on the individual level creates conditions promoting activity and responsibility, and on the societal level conditions that sustain the rural sector and weaken urbanization.

If we apply Weiler's categories and Lauglo and McLean's rationales to my findings from Norway, there is evidence of the first of Weiler's categories, namely the redistribution of authority, and for the Lauglo and McLean's political rationale, mostly in the late 1960s and the 1970s. There is also evidence of Weiler's efficiency category and for Lauglo and McLean's administrative rationale, particularly in the late 1980s and 1990s. Decentralization as a strategy to strengthen local culture in a learning process, Weiler's third category, was evident in the 1970s and most of the 1980s, but became considerably weaker in the 1990s. At the same time, we find an understanding of decentralization as local school management, giving an increasing role to parents. Ideologically, this is not primarily an argument for local culture but rather for institutional autonomy and individual rights. In contrast to the situation in many other countries (Levin, 1997), arguments for the parents' right to choose their children's school have played little to no role in Norway.

Weiler's categories and Lauglo and McLean's rationales explain in a limited manner the arguments in which decentralization is perceived as a bottom-up strategy for innovation. They are not, however, all encompassing, nor do they consider the change in the arguments over time. In the 1970s, the most important arguments were based on efforts to establish new democratic institutions for participation, debate, and decision-making. In the 1980s and 1990s, decentralization was a way to achieve more efficiency and the democracy argument changed into a more individualistic and market-oriented one, focusing on individual rights and free choice. The political shift from left to right was blurred by the decentralization rhetoric. Still, the categories and rationales that have been developed have been important in supporting a better understanding of the phenomenon of decentralization.

As my findings from Norway suggest, it is important to consider the context in which decentralization takes place. Understanding decentralization as a governance strategy requires a historical and comparative perspective. The time and place in which decentralization occurs affects the arguments and the understanding of the rationales underpinning the process. At the same time, there is much that can be learned from the features and patterns of decentralization as a governance strategy employed in different countries. In the following, I focus on four aspects of the dynamic interaction of the decentralization process that I have called "decentralized centralism." I will use cases from Norway and from British Columbia, Canada, to illustrate my points,⁴showing decentralization as a governance strategy in the field of education.

The Dynamics of Initiating

How decentralization and decentralization efforts are initiated is a fundamental question. A logical hypothesis would be that decentralization is normally initiated from below, or as a result of pressure from below, and has a grassroots base. It follows logically then, that decentralization reforms normally have the intention of transmitting power and influence from the central level to the lower levels. Empirical studies of decentralization reforms over time, however, do not support this hypothesis. Paradoxically, one typical trait seems to be that decentralization reforms are normally initiated from the top by the authorities at the central level. My study of Norway shows that most of the educational decentralization reforms from the late 1960s on were initiated at the central national level (Karlsen, 1993). Of course, some decentralization reforms resulted from pressure and action from below, but decentralization reforms have also been met with passivity and even resistance at the local level. Given this evidence one might argue that decentralization from the national level may function to prevent a threat to central legitimacy and thereby consolidate the central power platform. In this light, it is not surprising to find that decentralization reforms have often led to new central legislation and regulations.

One recent example of decentralization reform in Norway that was initiated from the top is the new Local Governing Act (Kommuneloven) for the municipalities of 1993 that gave more local freedom and more authority to the municipalities and schools. Another is the new funding system established in 1986 (St. meld. nr. 56 1986-87). Within what has become known as the lump-sum system, the state gives a sum of money to the municipalities to cover the costs of schools as well as the costs of other sectors, such as culture and health services. In many municipalities, this lump-sum system is then employed to fund schools. The school principal, then, becomes responsible for the budget. When this funding plan was initiated, a detailed, centralized ear-marked system was replaced by central regulations for a lump-sum system that made the municipal and school levels responsible for the budget. Both of these reforms have been referred to as decentralization reforms, and both were initiated at the national level. They have been met with both support and resistance at the local level. In particular, there has been skepticism about, and resistance to the lump-sum system among the teachers and their unions who claim that the new system will lead to more disparity among schools and weaken the principle of equality of educational opportunity, a principle that has been a basic ideal in Norwegian educational policy (Solstad, 1994).

Similarly, British Columbia, Canada, has recently had a new funding system centrally imposed through the School Amendment Act of 1990. This Act instituted a lump-sum funding system with more local freedom. Bargaining rights within British Columbia provide a good illustration of the dynamics of initiating decentralization. In the late 1980s, British Columbia enacted a number of pieces of legislation with consequences for the education system (Teaching Profession Act 1987). One of these acts, The Industrial Relation Act of 1987, managed the certification of bargaining agents and the rules governing bargaining. Teachers' unions had their bargaining rights expanded to the local level; that is, they had the right to carry out collective bargaining with the local school board. Principals were now considered to be administrative officers responsible for managing schools and were excluded from membership in the teachers' unions (Ungerleider, 1996). It was a central and not a local decision to hand over to local bodies the right to bargain and the power involved in the bargaining process.

This decentralization of bargaining rights is a good example of the dynamics of initiating decentralization and of how reforms have to be assessed in a historical context. In this case, the Industrial Relations Act has to be understood in relation to the Teaching Profession Act (1987), which in the same year established a college of teachers with the power to formulate and enforce professional standards related to the training, certification, discipline, and professional practice of teaching. As well, the government removed compulsory membership in the union.⁵ In this context, the granting of bargaining rights to the local level was interpreted by the teachers' union (the British Columbia Teachers' Federation) as an attack on them, an attempt to reduce the central power of the federation and to increase the governance power.⁶ The government's initiating the decentralization of power from the central to local bodies was, in this historical context, understood as the opposite, the attempt to maintain and strengthen governmental central control.

This example illustrates a contrasting initiating dynamics where reform through decentralization was initiated from the top. However, decentralization reforms can be perceived as being precisely the opposite, and in reality they can also be a strategy for strengthening central power. The decentralization of power to the local bodies also includes making them responsible for implementing central goals. Presented here is a decentralization dynamic in which initiating is a central task, but in which implementation and accountability are local duties.

The Dynamics of Content

Another issue is related to the idea of substance in the decentralization process. Decentralization will normally involve a transmission of tasks and power from the center to the periphery. Decentralization in education has been closely connected to the balance between standardization and diversity of school content and curriculum. More specifically, the centralization(decentralization dimension has led to questions such as: What should be taught in school? To what degree should the curriculum be standardized? Should the curriculum be set at local or national/provincial level? The decision-making processes in relation to curricula in Norway and British Columbia are outlined in the following in order to illustrate the dynamics of content in decentralized centralism.

For the last one hundred years, Norway has had a system with a national curriculum for primary and lower secondary schools. During the recent period of decentralization, there has been a shift toward a more standardized content. When the Curriculum Guidelines of 1974, revised in 1987 are compared to the Curriculum Guidelines of 1939 and 1959, there is evidence of the growth over time of freedom and responsibility at the local and classroom levels. The new Curriculum Guidelines of 1997, however, are more uniform and standardized and have less diversity (Læreplanverket, 1996). The guidelines now allow fewer electives and emphasize science, basic skills, and exact knowledge.

This change was caused in part by an evaluation of the Norwegian educational system that was initiated by the Ministry of Education in 1988 and conducted by an OECD expert group (1988). The report questioned the decentralization process and called attention to the risk of lower-quality education in a decentralized system. The OECD report became the basis on which the central authorities initiated a new curriculum reform and a national evaluation system. The ministry started a project aimed at developing models for evaluation and accountability (Granheim & Lundgren, 1990). The new curriculum was developed over a short period of time and was initiated and controlled by the Ministry of Education. Historically, curriculum changes were normally undertaken by representative committees working over several years.

It is interesting to note that the province of British Columbia was, at the same time as Norway, involved in an intensive process of working out a revised curriculum called K-12 (Province of British Columbia,1994). The new curriculum emphasized work experience, promoted the use of information and computer technology, and demanded that all students take a second language as part of the required curriculum in grades 5 to 8. K-12 introduced a new subject called "Personal Planning" from Kindergarten to Grade 7 and "Career and Personal Planning" from grades 8 to 12. In grade 9, students, teachers, and parents were to work together to help each student develop a formal "Student Learning Plan" (SLP). It was considered important "to have students set learning goals" and "develop plans to achieve them" (p. 8). Compared with the earlier curricula, K-12 was more standardized, centralized, and skills-oriented.

Again, we can see the dynamics. The decentralization of content and power to set the curriculum at the local level is the reason for and legitimates standardization and central control. And again, we can observe the two-way process between centralization and decentralization. In particular, the new subject introduced in K-12 illustrates the dynamics of decision making in relation to content. The introduction of the subject was a central decision that had to be implemented at the provincial level. At the same time, the nature of the new subject gave the local schools and individuals more autonomy and responsibility, which was important for the implementation of central goals.

The Dynamics of Levels

As the function and responsibility in the educational system are normally located at various levels, from the center to the periphery, the authority over educational decisions is also located at various levels. The number of levels and the tasks and responsibilities associated with each of those levels, however, vary from one educational system to another. Again, examples from Norway and British Columbia illustrate the dynamics of decentralization as a governance strategy. In the following, I provide a short description of the levels of the educational systems in the two countries.

Canada is the only industrialized country that has no federal office or department of education (Levin & Young, 1994); the central-state level in education does not exist. The provisions of Section 93 in the Constitution Act⁷ give each province authority over education. For the province of British Columbia, decisions regarding education occur at three levels (Ministry of Education, 1995). The first, the central-provincial level, consists of the Ministry of Education which provides leadership for all provincial education agencies at the two lower levels. The second level is that of the school districts that are created through provincial legislation. Schools in British Columbia are organized into 75 school districts governed by elected school boards. The third and most peripheral level is that of the schools and the teachers. In the Norwegian system, there is the Ministry of Education at the central-national level, a national Education Office led by an Educational Director at the regional level, a school administration and normally a school board at the municipal level, and most peripherally, the schools and the teachers at the local level (The Royal Norwegian Ministry of Education, 1992). This four-level system is, in reality, only three levels because the Educational Directors on the regional level are governmental agents, even if they have some degree of governmental freedom.

In both Norway and Canada, educational reforms in the name of decentralization have been intended to transmit power and responsibility from the central level to the intermediate level and, further, to local level. Given this background, I will stress three points. The first point is that the decentralization of tasks and administrative responsibility does not necessarily mean a shift of power from a higher to a lower level. The lump-sum funding reforms in both countries exemplify this. On the one hand, the reform was an administrative decentralization giving the local levels more authority over the budget; on the other hand, the total amount of expenditures was still a central decision and allocation had to be done according to the central existing legal framework. In addition, there are normally tensions and conflicts connected to budget decisions. From this perspective, decentralization can be perceived as the transmission of conflicts and problems from a higher level to a lower one. Depending on the levels, decentralization may have either a conflict-reducing or a conflict-increasing effect. We can deduce from Weiler's theory (1990) that an increased interest in decentralization should be expected in periods of disagreement and conflict about educational politics (Karlsen 1993).

The second point is that decentralization of authority from central to lower levels can have a legitimizing function for the central level. The modern state is in need of legitimacy (Habermas, 1976). As Weiler (1990) claims, decentralization will give a compensatory legitimization and will thereby reestablish the basis of the state authority.

From a national perspective, Canada's decentralized dimension has given the provincial levels a foothold in their struggle to gain federal respect for provincial and regional differences. But this does not necessary mean that Canada has a highly decentralized school system. Focusing on provincial control over schooling may make federal actions less visible. The federal government is, in spite of the Constitution Act, involved in education (Hodgson, 1987); it runs educational programs, is responsible for educational cooperation with other countries, and through federal departments is responsible for educating personnel. There is also cooperation between provinces at the federal level in education. This is a very interesting observation in a system that so strongly advocates that education is a totally decentralized, provincial responsibility.

In relation to the curriculum reform in British Columbia (K-12), efforts have been taken to establish a national standard. In a protocol issued in 1993, ministers from Western Canada⁸ declared that "the ministers of education share many common educational goals and acknowledge the task of ensuring greater harmonization of ways to achieve them" (Western Canadian Protocol,1993, p. 2). This harmonization-seeking centralization movement is reinforced on the national level by the federal government's work to set national standards and indicators for education, and to develop national assessments (Ministère de l'éducation 1994). The Council of Ministers of Education, early in the 1990s, started to work seriously to develop criteria for standards and evaluations that will have a clear impact on education nation-wide (Kelsey, 1992).

The Canadian example can be used to illustrate the dynamics among levels as part of decentralized centralism. Centralization processes, such as harmonization and setting a national standard, would probably not be possible without a strong ideology and statement of provincial responsibility for education in a decentralized system sensitive to local needs and conditions.

The third point is that the same phenomenon can be perceived in exactly opposite ways, depending on one's level in the system. In Norway, for example, the municipal level has the responsibility for running the compulsory school system according to central laws and regulations. The municipal level will clearly be perceived as a center in relation to the school level, but at the same time as periphery in relation to the state level. Therefore, the transfer of authority from the state level to the municipal level from a state perspective will be viewed as decentralization (devolution). From the schools' perspective, these actions may be perceived as centralization. This impression of being controlled may be reinforced because the school level is so close to and dependent on the municipal level. The decentralization/centralization dynamics among levels is connected to the roles and functions to which people are assigned. The same reform can be perceived to be contrary, and even be contrary, depending on the level.

The Dynamics of Simultaneity

The analyses presented of the decentralization process in relation to initiating, content, and levels, show that the model of decentralization and centralization as waves following and replacing each other is far from reality. But we can probably observe the rhetoric of waves, depending on what sounds good in public opinion. The cases referred to from Norway and British Columbia indicate that the process usually goes in two directions at the same time.

In addition, the government strategy in the late 1980s and 1990s came to fit very well into the new Management by Objectives (MbO) model introduced by the renowned if not legendary Peter F. Drucker (1954, 1964, 1977). Drucker based the development of this model on notions from his early works published in the 1950s and 1960s.⁹ MbO was originally developed as a strategy for more cost-efficient production and leadership in industry and business, but in the 1980s it was introduced into the public sector, including education. The MbO model is linear and stresses the importance of planning both on macro- and micro-levels. It is a goal and outcome model in which decentralization is fitted into a centralized strategy. Setting central goals and standards for outcomes are tasks for the central level and therefore centralization, while the means and the responsibility for implementation are local duties and therefore decentralization. Both centralization and decentralization are involved at the same time in the MbO model.

Our two empirical cases, Norway and British Columbia, show that the MbO model has been strongly emphasized, in particular in the importance attached to setting national/provincial goals. In the 1980s, official documents in Norway described the MbO model as a cure-all for every governing and management problem. The government decided that the MbO model should be introduced throughout public administration (FAD 1986; St. meld nr. 37 1990-91).

The same has been the case in British Columbia. One example of this is the government's enacting of new legislation that directly and indirectly affected the educational system. The new Independent School Act (1989), which received royal assent in 1989, was an enabling act.¹⁰ Decentralization and centralization forces occurred at the same time as contradictory phenomena. Kelsey et al. stated in a study of British Columbia, for example, "the decentralization in the Act (School Act 1989) has in effect been countered by a good deal of regulatory re-centralization" (1995, p. 42). This governance strategy adapted principles from the MbO model. Common central goals in a broader program called "Year 2000" were combined with a decentralized strategy that included local freedom to implement the goals within a central framework.

The authorities' enthusiasm for the MbO model is understandable. The model maintained central power and control and a kind of hidden discipline at the same time. The model gave local "freedom" and probably encouraged more acceptance for working under pressure because of the local involvement in the planning and implementation. And most importantly, the dynamics of simultaneity as an important element in decentralized centralism was an adequate strategy to handle the persistent dilemmas in the two countries' educational policies (Levin & Young, 1994; Karlsen 1993).