Resumé

This article assumes that bureaucrats, like other actors, follow their interests. However, it is also recognized that bureaucrats in political environments face a multilevel game situation with unstable property rights, where priorities between institutional position and more pecuniary goods, e.g. budget maximization, especially in reform situations have to be made. Thus, bureaucrats realize that in order to obtain the latter, the precondition is a secured institutional position and that this position has to be obtained through a bargaining process involving transaction costs, externally due to negotiations with the political principals, and internally through union organization. Empirically, the phenomenon is examined in the case of the Danish day care system which went through a reform period during the late eighties and early nineties. It suggests that although demand for day care rose significantly during this period, day care workers (and their unions) worked for - and succeeded in - securing their institutional position.

Introduction

Public producers of welfare services face the same challenge as producers in a market: What to produce and how much? The combination of quality and quantity is assumed to be obtained by an aggregation of preferences from voters to politicians, who then install different institutional arrangements to provide the needed services. In this process, production units are often treated as passive actors who mainly follow orders from their political principal. Thus, if politicians observe a need for a specific welfare service among voters, they can, due to their hierarchical superiority, establish new service providing institutions or they can order already established institutions to change their output.

This article rephrases these questions and analyzes the relationship
between political sponsors and the service providing institutions through a
contractual perspective where ex ante and ex post strategies for the involved

---

* Jens Bejer Damgaard, Department of Political Science, University of Aarhus. Universitetsparken, DK-8000 Aarhus C, Denmark.

actors each play a significant role for the allocation of authority, costs and benefits in the Danish day care sector. Instead of politicians having full control of output from welfare institutions, it is argued that street-level bureaucrats (Lipsky 1980) within service producing units have a specific interest in not only what and how much to produce, but also in the ways in which welfare services are produced. In addition, it is argued that street-level bureaucrats have formal as well as informal ways of affecting the contracts established between political sponsors and the service providing institutions. Two conditions, however, are assumed. First, the institutional arrangement that provides the welfare services must subsidize user costs heavily. Thus, the price users have to pay - if any - must be considerably lower than the real costs of the production. Secondly, the institutional arrangement must secure the quality of the service. That is, quality is assumed to be specified according to formal or informal rules, especially concerning the ratio between users and welfare workers. Formal rules include local negotiated agreements, whereas informal rules imply that municipalities without explicit contracts regarding the workload of day care workers cannot diverge significantly from municipalities with such contracts. This ratio is important, not only regarding the productivity of the sector, but also because a lower ratio, i.e. more workers to less users, normally is associated with a higher service quality. This article argues that these two conditions provide the incentive structure for welfare workers to pursue influence in the governance structure. Street-level bureaucrats focusing on control of the governance structure obtain two advantages: They ensure the means to reach their final goal (e.g. budget maximization or demand control) and prevent or stall reform strategies that weaken their chances of reaching their presumed goals. Thus, rational bureaucrats realize that sometimes they must play two games to pursue their goals, or as formulated by Tsebelis, "... the actor is involved not only in a game in the principal arena, but also in a game about the rules of the game" (1990, 8).

Rational street-level bureaucrats should focus on the governance structure because there is a close relationship between a specific economic institutional arrangement or governance structure and user demands (Niskanen 1971, 33-35). Since this critical connection can be made, street-level bureaucrats not only shift the preference structure of users towards their production, they also ensure the users as informal alliance partners in the bargaining process with politicians (Wilson 1995, 333-334). The empirical evidence in this study stems from recent reforms in the Danish day care sector, but the tenets may be applied to other welfare states with similar institutional arrangements.

The implications of employing this approach to local service producing
institutions are that bureaucrats might seek neither budget maximization
(Niskanen 1971) nor organizational slack (Cyert & March 1963; Migué &

Bélanger 1974; Duizendstraal & Nentjes 1994) in the first case as assumed by traditional rational choice theory. Moreover, they might not seek bureau shaping as suggested by Dunleavy (Dunleavy 1991, 200-209), because these institutions, although operating in a hierarchical governance structure, are relatively flat organizations without the internal hierarchy needed for diversified maximization strategies among its members. Thus, the "leadership" in a Danish day care institution cannot be expected to have collective action problems with regard to budget maximization vis-å-vis the "bottom bureaucrats," as the absence of an internal hierarchy is evident: The leader has to be a trained pedagogue.

Instead, street-level bureaucrats must secure that they participate in the political negotiations about the input resources to the sector. Without this governance structure, street-level bureaucrats could find themselves in a more hostile environment, i.e. in a more market-based governance structure, which would place them outside control of input resources. To street-level bureaucrats and their unions, participation in the negotiations ex ante to new contracts (reforms) with the political leadership is the precondition for seeking more pecuniary goods ex post. Examples of such goods could be maximization of cost per child, securing professional values in the production, a reduction of uncertainty by participating in the planning of the sector in each municipality, etc.

The extent of this strategy's success can be examined in two ways. The first step is an investigation of the ability to force other institutional arrangements (e.g. private alternative providers) to play a minor - if any - part in the total production. The second step - if step one is successful - is to ensure that politicians make the "right decisions" regarding choice of governance structure, particularly in reform situations, to prevent outside interference in the management of the institutions and in decisions regarding the institution budgets. Both steps of the strategy have to be pursued on a national and on a local political level, because the institutional arrangement includes two different decision arenas. First, the national arena, where union monopoly and salary and working conditions are negotiated, and second, the local arena, where specific rules concerning budget and outputs are negotiated for every single day care institution.

The "governance control strategy," however, has its own costs, because it involves participation in the political negotiations regarding the decisions of the welfare state in general and negotiations concerning the bureaucrats' production area in particular. A critical feature for street-level bureaucrats in public service producing institutions is therefore, externally to secure representation at the negotiating table with local and national politicians, and internally to organize employees in unions. It is through a strong unionized base that employees can influence the political bargaining process.¹

The Demand and Supply Side for Welfare Services

Welfare services in institutionalized welfare states are largely characterized by a "free" availability of these services (Esping-Andersen 1990, 18-21). Users (and citizens) pay general taxes which are used by politicians to give budgets to the producing institutions. A fair allocation of welfare services is generally put forward to justify this institutional arrangement. However, traditional rational choice literature would argue that since welfare services are mainly private goods, they might just as well be provided through the market or market-based institutional arrangements (Savas 1982, 53-75; Ostrom 1983; Mitchell & Simmons 1994, 87-101). The point in this line of argument is that users - now customers - are given an exit option in the evaluation of the services they receive (Hirschman 1970, 22-25), and that (public) service producing institutions have to respond to these evaluations. Given the notorious "waiting lists" for many publicly produced welfare services, resulting from little or no excess capacity, the exit option does not seem to work very well, as effective evaluations from users presuppose that providers experience an economic loss if users choose other providers, or as Hirschman puts it, "[n]o matter what the quality elasticity of demand, exit could fail to cause any revenue loss to the individual firms if the firm acquired new customers as it loses the old ones" (1970, 26).

In reality, the demand curve for welfare services is not completely inelastic, but since users do not pay directly to the producing institutions for the services they receive, other variables must determine parental demand. In this way, political scientists have sought and partly found determinants of the demand for specific welfare services. For day care in Denmark, a connection can surely be established between women's labor force participation rate and other socio-economic variables and 1) the demand for day care, and 2) the amount of resources spent by local governments on day care (Mouritzen 1991, 102-106; OECD 1990, 123-126; Ministry of Social Affairs 1993). Yet, in his extensive study of general and sector specific growth in expenditures in Danish municipalities from 1982-1986, Mouritzen cannot find a clear connection between the color of the political leadership (bourgeois or socialist) and the expenditure level of day care. Instead, the political leadership seems to effect variables such as parental payment and the workload for day care workers. Surprisingly, local governments with socialist majorities tend to increase the workload for day care workers, although this is due to a more stressed economic situation in these municipalities. Mouritzen's study also controls for such variables as 1) service level in the beginning of the period, 2) growth in the number of single bread winners, 3) growth in the number of children from zero to six years of age, and 4) growth in the private sector (with more people working, demand for day care should increase), and finds some support for these hypotheses

(Mouritzen 1991, 364-370). However, none of these investigations, including Mouritzen's, consider whether bureaucrats might influence the ways the local governments organize the production, and next, how the institutional set-up could affect choices made by users - not to mention the costs of the specific contractual relationship between sponsors and providers. The simple question is: Why should parents not choose to use the publicly provided and subsidized services? Danish surveys show that parents overwhelmingly prefer and are satisfied with the public service (Vedel 1992, 23-24; Bertelsen 1991, 46-49). If they already pay for a large part of the service through general taxes, there is no point in seeking the service on the private market, except in the brief periods where the service is unobtainable due to lack of supply. Besides, in these brief periods, parents tend to consult the "black market." which exists for day care services.

Public welfare service producing institutions as they are discussed here face two principals of unequal importance. First, of course, the users who benefit directly from the production and have an interest in getting high quality service. As mentioned, the price becomes less relevant because most or all fees have already been paid. Secondly, and more important, the politicians who provide the budget and expect the agreed output from the institutions. From the institution's point of view, a conflict between competing demands arises as users and politicians do not always agree on decisions concerning specific resource allocations. What, then, should day care institutions aim to satisfy: Users' or politicians' demands?

There are at least two aspects of this situation. The first is that budgets are too small to meet the quality demand specified in the rules combined with high demand created by asymmetrical payments. Public welfare producing institutions can always use larger budgets. Professionals are trained to provide the best service possible and often feel frustrated if limited budgets impede them from delivering a high quality service. The snag is that politicians cannot and will not fund all these demands.

The second aspect concerns a problem within the delivering agency which
we could call "organizational shirking." That is the agency's inability to
produce in a cost efficient way.

Agency Problems

How large should the output from a welfare institution be? How much can it produce on fixed budgets? Hospitals, schools, day care institutions and other public welfare producing institutions typically make an internalized contract with the politicians: If we receive this budget, we produce this amount of service. What politicians do not discuss with their bureaux is the value of the production. In this case: What is the value of having a large number of

children in day care institutions? This is the position Niskanen took (Niskanen 1971, 24-30). The budget-output function presupposing a specific incentive structure, where employee efforts and salaries are not directly related, represents the institutions' ability to produce their service: Only through higher budgets are they able to increase output. In this bilateral contractual relationship, politicians face the problem of asymmetrically distributed information. They are in no position to determine whether the budget-output function represents the true capabilities of the institution or the quality of the service, except through standardized measures such as the workload per day care worker. The agency problems can be specified to concern all contractual ex ante costs (Williamson 1990) and the incentive structures created by specific contractual relationships (Baker, Jensen & Murphy 1988). That is the cost of drafting and negotiating a contract under the condition of bounded rationality and (political) uncertainty (Williamson 1985, 26-29). Yet, there is an ex post side to all contracts, where potential opportunistic behavior must be taken into consideration (Williamson 1981). Opportunism defined as "self-interest seeking with guile" (Williamson 1985, 47) is the situation where the agent meets neither the principals' demands, i.e. the politicians (Christensen 1992) nor the users' demands. This leads us directly to the part concerning "organizational shirking."

Shirking is normally associated with individuals whose efforts in a collective working process cannot be monitored and in this situation have incentives to reduce their input (Miller 1992, 30-35). It is, in other words, assumed that workers are "lazy" and that they will only work hard given the right combination of the stick (monitoring/sanctions) and the carrot (salary). Alchian & Demsetz provide fruitful insight into the mechanisms of working relations in teams (Alchian & Demsetz 1972). The problem of shirking, they argue, is working relations where not only the monitoring but also the transfer of the monitor's information becomes increasingly costly. Team production among workers is the typical example and the solution is the firm. Contrary to the later position taken by Williamson, Alchian & Demsetz argued that the key explanation for the rise of the firm was to economize on the costs of team production. This argument was so strong that Williamson had to incorporate it in his explanation of The Economic Institutions of Capitalism (Williamson 1985, 52-54). Williamson's definition of "the primitive team" is exactly the Alchian and Demsetz notion of low human asset specificity and inseparable working relations (ibid., 242-250). Low human asset specificity means that workers only receive little job training and that the learning-by-doing process of workers is very limited. It does not mean that workers are uneducated - not at all. Doctors, engineers and lawyers all have long educations, but they need not achieve qualifications in their jobs that make them hard to replace. In fact, for many, the first job is after a long education often viewed as a springboard to something "better."

Although employees in day care institutions may view their jobs differently
- with a larger focus on job security - than the above-mentioned categories,

the same "primitive team" relations are found. Focus on job security and primitive team relations combined with the absence of an internal hierarchy in day care institutions suggests that day care workers value highly a contract with the possibility for budget or slack maximization through smaller work loads. Among day care workers top bureaucrats as described by Dunleavy are not found (1991, 191-193).

The point is that many public institutions providing basic welfare services can be described in this way. Several implications follow. First, it is extremely difficult for political principals to monitor employee behavior in these institutions. Second, institutions basically only get one budget (they do not charge for their services). Third, workers are not directly affected by the total productivity of the institution and - at least in the Danish case - workers are given strong formal and informal incentives to become union members. Fourth, negotiation costs in the sector are neglected. In order to run the sector, local administrators and representatives of the union meet on a regular basis, often weekly, to solve local contract disputes, discuss new initiatives or to plan supply and demand problems. Such negotiations are of course intensified in reform situations, where new contracts are established.

With the "conservative" attitude Dunleavy gives street-level bureaucrats, they can be expected to have a strong interest in maintaining these conditions, partly because reforms bring uncertainty, and partly because streetlevel bureaucrats indeed benefit from the distribution effects of this hierarchical governance structure (Dietrich 1994, 37^45). Thus, what the next section attempts to show is that employees and their unions (and users) have benefited tremendously from the existing institutional set-up.

Danish Day Care: A Parental Heaven?

The story of the Danish day care sector is one of apparent infinite growth in size and scope. Since 1965, when this welfare service was made universal, the number of available places has grown. Figure 1 shows the remarkable development. Except from the period 1981-1984, the sector's capacity grew in absolute numbers for all years, and the reduction in the early 1980s took place when the number of children in the relevant age group dropped significantly. Thus, the enrollment rate went up throughout the period (except in 1989) - from about 10 percent in 1969, reaching 50 percent in 1982, to approx. 65 percent in 1994, and the expansion is expected to continue. The reason is a political commitment made by the Social Democratic-led coalition government that gained office early in 1993. The government announced that by the end of 1995, all children from 1 to 5 years of age would receive an offer for a day care place.

The "day care guarantee" as it was called could not be achieved as promised, because Prime Minister Poul Nyrup Rasmussen, who announced it, had not coordinated the initiative with the producers: the local governments who run the day care sector. Although intense efforts were made locally, thousands of children were still on the "waiting list" on January 1, 1996.

National and Local Corporatism

Producing day care services locally does not mean that all the rules of the institutional arrangement are negotiated locally. Both the local governments and the unions are organized in highly hierarchical organizations: The Association of Local Authorities (ALA) and The Danish National Federation of Early Childhood Teachers and Youth Educators (BUPL)/The National Union of Nursery and Child Care Assistants (PMF). BUPL organizes skilled day care workers, whereas the members of PMF have received no or only on-the-job training. These three actors sign a collective agreement which establishes the general framework for the working conditions of the employees in day care institutions. Specifying salaries and the general

working conditions, the collective agreement is a major regulator of productivity and costs in the sector. However, the most important determinant for the costs of the sector is the number of workers per child. Most municipalities have an agreement with the local unions about the workload per day care worker. Although this gives local governments room to find local solutions, the workload seems very consistent not only from municipality to municipality, but also across time (ALA 1996, 26).

The legal basis for the local governments' operation of day care institutions lies in the Social Systems Act, originally from 1976, which makes the operation of day care services a local authority responsibility. The Social Systems Act, however, does not directly specify what kind of day care the local governments should provide. Without strictly mentioning day care institutions, the local governments appear to have a choice among different ways to provide the service, and to some extent this choice has been used. This choice, however, seems to be influenced by several considerations. Since the national corporatist structures are copied at the local level, meaning that local unions participate in the formulation and implementation of all political decisions concerning the day care sector, local politicians cannot focus exclusively on productivity without disturbances among workers and users.

Local governments can choose other institutional arrangements to provide day care services. One way is to contract with private persons to organize day care in their private homes. According to circular letters attached to the Social Systems Act, the maximum number of children allowed in each private day care is five, which means that the local government has to contract with a large number of private persons just to get a reasonable number of places. Few problems seem to follow these contracts, as standards have been developed and ex post opportunism from particularly the "day care mothers" is limited by the fact that few possibilities for shaping the simple governance structure exist. Empirically, this means that "day care mothers" are hired and fired according to the actual need of the local government. The alternations of supervised family day care contrast the development of the day care institutions. The development in the number of places of the different kinds of day care services is shown in Figure 2.

Nurseries enroll children from 0 to 2 years of age. From the age of three, the children graduate to kindergarten. Apart from the age difference, nurseries are characterized by a much larger labor input. The workload for day care workers in nurseries is half the number of children compared to the kindergartens. This general rule also applies to the age-integrated institutions. Receiving children from 0 to 6 years of age. age-integrated institutions were created to ease the children's transition from nursery to kindergarten. As Figure 2 suggests, this kind of institution has become an increasingly popular choice among local governments. In terms of productivity, however.

age-integrated institutions do not decrease the marginal amount of resources
used - precisely because of the workload rules. Children in Denmark leave
the full-time day care system at the age of 6 when they start school.

Within the hierarchical governance structure and the institutional constraints described above, the local governments decide how much should be produced and how. Furthermore, they are in charge of employing workers in their day care facilities (Damgaard 1996). For such a relatively simple service, this has created a number of problems. First, children are born throughout the year but they leave the day care sector in August to start school. Thus, there are fewer children who need day care in autumn than in spring. Firing day care workers in August only to hire them again in January is not possible according to the general agreement negotiated at the national corporatist level. This means that local governments who have implemented the "day care guarantee" face a general optimizing problem, especially if they establish more day care institutions rather than employing more "day care mothers." Flexible agreements between local unions and the local government, where day care institutions accept less children in autumn and more in the spring period, have been implemented as a solution. Yet, flexible agreements are not established free of charge. The "price" has often been that the single day care institution gets a generally smaller workload, so what

on the surface seems to be a better adjustment to demand actually hides an inadequate way of matching demand to supply. Second, parental demands shift. These shifts can be registered on both the micro and macro levels. At the micro level, parents' demands reflect not only their present work situation, but also their attitude towards leaving their children in institutions. At the macro level, parental preferences shift between supervised family day care and day care institutions, especially for the youngest children (Bertelsen 1991, 42-51). Third, just as the supply of day care services is hard to match throughout the year, the supply is hard to match over years. Planning seems to avoid the most obvious measure for how many day care places are wanted, i.e. the total number of children. Since most parents have incentives to place their children in public day care because of asymmetrical payments (see next section), the demands almost equal the total number of children. Supplying day care services, however, has never been calculated as a function of the number of children. Instead, political bargains better explain the present size and character of the day care sector. To illustrate the political content, a useful point of departure is the financial structure of the sector.

Asymmetrical Payments

Danish parents pay a fee to the local government to have their children in public day care. In the Social Systems Act, the limits of parental payments are specified. In the period 1991-1993, the percentage was gradually reduced from 35 percent to the present 30 percent. This means that the local government can charge up to 30 percent of the running cost of day care - the last 70 percent is then financed through general taxes.² It is this "discount" which has the major responsibility for shifting and thus shaping parents" preferences towards public day care. Excessive demand at the going rate and quality are created. Empirically this is expressed in "waiting lists'" at the local government offices for day care services. Parents with children on waiting lists, however, experience the lack of places as a temporary problem and are often able to find alternatives on the "black market." Without incentives to organize themselves, potential users do not press existing institutions to accept more children. Instead they largely agree with day care workers' demand to expand the sector.

Employees are not passive players in the game of shaping preferences. The unions participate not only in the collective bargaining at the national level concerning their working conditions, they also sit at the table when politicians, locally and centrally, form policies on the day care sector. Through corporative structures politicians reduce their formal hierarchical decision power to obtain easier implementation of. for instance, reforms. Occasionally, politicians have tried to bypass the influence of unions, but the results have so far turned out negatively. Strikes among day care workers are

not an unknown phenomenon, and since the institutional arrangement in Denmark places the local government as employers, they get the blame - measured as political costs - for any malfunctions (Riiskjær 1988). Parental feelings against the local government during strikes have resulted in payment boycotts under the motto: No delivery - no payment (Ketcher 1991, 317— 318). Hence, the institutional arrangement has provided an extremely strong relationship between users (and potential users) and day care workers due to their common interests in preserving a heavily subsidized service. This alliance is so strong that even discussions about alternate governance structures imply political costs that are too high.

Reforms - After All

In 1982, a conservative-led coalition government took office in Denmark. Shortly after, new ministers announced that major reforms were on the program for many sectors (Christensen 1991, 39-54). Day care institutions were not excluded from reform initiatives. The tradition of having private, but publicly subsidized schools as an alternative to public schools (Christensen 1995) seems to have been a source of inspiration for the new Minister of Social Affairs, Palle Simonsen. However, when he made a proposal according to which private day care institutions (pool schemes) could get public financial support, the unions were strong opponents. Using the argument that pool scheme institutions would develop into discount institutions, the unions started a campaign that kept the bill from being passed for six years (Kyst & Tinning 1992, 7-19). The real threat to the unions was that pool schemes were outside their control. With their formal private status, they were not intended to be a part of the municipal hierarchy and therefore not included in the collective agreement. Thus, the intention of the conservative minister was not only to reduce costs per child and to give parents more direct financial influence, but also to reduce union influence by not demanding that the pool schemes employ organized and educated personnel. Instead of initiating the proposal in 1982, Parliament only voted for experimenting with new forms of organizing day care institutions. During the next five to six years, merely 12 tryouts were tested (ibid., 8). The green light for the pool schemes came after negotiations between the still conservative-led government and the ALA who reached an agreement in 1988. During this period, the ALA changed its attitude towards the pool schemes. Originally opposed to the idea, the municipalities experienced increased financial problems, which, to some extent, were a result of the conservative-led government's policy of reducing the block grants to the local governments.

Before local governments could subsidize private pool schemes, however,
the Social Systems Act had to be changed. This brought the debate right into

Parliament where the Social Democrats and members of the Socialist Peoples Party stood firm on union arguments. Social Democrat (now minister of Labor), Jytte Andersen said, "first we want [private] institutions, where collective agreements are signed with the unions [BUPL & PMF]. . . . Third, we want to secure the pedagogical content, which mostly relies on the ratio between pedagogues and children.. .. Fifth, we want the employees [in the new pool schemes] to be trained pedagogues." (Folketingstidende 1989— 1990, column 1831).

In other words: The Social Democrats (and The Socialist Peoples Party), having strong links to the unions, were naturally opposed to any reform that threatened the unions' institutional position. The problem for the two socialist parties was that the government could muster a majority in Parliament for its proposal. On the other hand, the government knew that a radical proposal could get into trouble in the implementation phase, because local governments would need some sort of union acceptance to establish pool scheme institutions if it wanted to avoid strikes. The final government argument against radical changes was that once a Social Democratic government regained power, it could simply change the Social Systems Act. So, the conservative-led government knew that the socialist parties, and with them the unions, had to accept the pool scheme institutions if the reform was going to survive. Eventually they did. On the surface, hardly any concessions were made. But the context of policy making and reform had changed. Denmark was experiencing a "baby boom" demanding more day care institutions. In addition, public day care became a social right also for unemployed parents, and a shift in the general attitude in favor of day care institutions could also be registered (Bertelsen 1991, 51). In short, demand increased dramatically. With these prospects, organized pedagogue jobs were probably secured, and the pool schemes were accepted.

So, although the proposal passed Parliament in 1990, stalling it for eight years remains a solid union victory during which the enrollment rate in public institutions climbed steadily. BUPL explicitly accounts for the assistance of the Social Democratic Party and the Socialist Peoples Party, "although BUPL views the proposal in the Social Systems Act with great skepticism, we must also recognize that the worst parts of the proposal were abolished due to BUPL's cooperation with the Social Democratic Party and the Socialist Peoples Party." (BUPL 1993, 203).

The outcome of the proposal has been modest. Pade & Galvind counted 133 pool schemes with 3,015 children in 1993 (Pade & Galvind 1993, 22). Out of 133 institutions, only 39 could be compared directly with pure public institutions, while the rest, besides the important function of representing parental needs, are mixes of ideological/religious institutions and part-time small arrangements. The latest count of pool schemes shows that the development has not been speeded up: ALA counted 5.149 places in the

beginning of 1996 (ALA 1996, 11). Having accounted for BUPL/PMF's success in avoiding private provision of day care services (step one in the street-level bureaucrat's '"Governance Control Strategy"), we turn to look at the reforms for the public institutions.

Reforms in the Public Institutions

A revision of the Social Systems Act in 1992 made way for local governments to reform their day care institutions. Based on the idea that decentralization increased accountability and responsiveness for day care workers, the local governments had already initiated budget reforms. The 1992 reform encouraged budget reforms to go further. In addition, user representation in every day care institution through parental boards was made compulsory. To day care workers, however, the reforms could increase uncertainty if parents made use of their decision making power. Thus, the potential of having users decide activities and budgets in day care institutions increased uncertainty. It was therefore likely that day care workers would oppose radical options within the reform framework. Guidelines from the Ministry of Social Affairs specified two levels of decision making power to parental boards: Minimum and maximum. Minimum decision making power included parental influence on pedagogical activities in the institutions and an annual budget decision, whereas maximum decision making power added decisions regarding bookkeeping (including the administration of parental payments) and the opening hours of the institution to the agenda. Moreover, all parental boards have nomination rights when new employees are hired. One important institutional feature, employment relations, remains unchanged by the reforms. AH day care workers are still employed by the local governments as opposed to the private pool schemes who themselves act as employers.

Implementation of the reforms is in the hands of the local governments, who negotiate the specific solutions with the local unions. Hardly any rules are changed without prior acceptance by the two parties. Although it can be argued that such negotiations may reduce transaction costs ex post, because day care workers and the local governments are given informal incentives to stick to the contract, investments made ex ante around the bargaining table never prevent opportunistic strategies altogether (Miller & Hammond 1994). Hence, the time-consuming negotiations leading to the reformed institutional arrangements imposed certain transaction costs on both parties. These costs, however, are largely neglected in most evaluations. Since the investments made in the negotiating process result in an institutional arrangement that may change the distribution of costs and benefits among a specific set of actors, incentives to bear negotiation costs are asymmetrically divided (Knight 1992, 40-44). To politicians, radical reforms which only potentially

reduce the total costs of day care are not worth risking a political career for,
while the unions have a specific interest in bypassing reforms which change
the distribution effects to their members' disadvantage.

To pursue their goal, unions must be able to list their preferences. Assuming that the unions and their members were satisfied with the former institutional arrangements, which have been very lucrative from their point of view, the strategy in the reform process must then be to keep local governments from giving parents a high degree of influence while simultaneously securing salary-related entries on the budget in the hands of the institution leaders. Therefore, minimum parental influence combined with a budget system where the local day care institution is secured a specific amount of resources to pay employees' salaries is preferred. Put bluntly, a minimum value on the first dimension used to describe the reforms, degree of budget decentralization, and a minimum value on the second dimension, degree of parental influence, are considered highly desired by employees. Tying the salary account to a specific untouchable budget account prevents discussions concerning alternate use of these resources. Besides, two accounts for substitutes and regular employees closed for inquiries regarding alternate use leave the leader of a day care institution in charge of the largest part of the budget. Because strong informal rules between the leader (who must be a skilled day care worker) and the employees exist, uncertainty in budget decisions in the small and flat organizations are reduced. Hence, employees prefer to keep relatively tight constraints on the budget and prevent direct parental influence on especially budget decisions. This typology, shown in Table 1, adds two medium positions and one - from a union and employee perspective - highly unwanted reform model.

The employee model and the user model are defined by a combination of maximum and minimum values along the two dimensions. As argued here, employees prefer interference in budget decisions compared to giving away autonomy to parental boards. Rational day care workers can, at least partly, control budget decisions internally through informal rules between employees and users, especially in situations where tight constraints limit the power of the parental boards. These informal rules include parental desire to cooperate with day care workers who take care of their children throughout the day. Hence, we assume that rational day care workers prefer heavy constraints on parental voice options and increased budget autonomy, to the combination expressed in the user model. What has been termed "the radical model" is, obviously, the least preferred reform model to day care workers (and their unions). Since this model includes the formal rules to interfere in most activities and decisions concerning the relationship between employees and users and budget decisions, employees become more vulnerable to parental demands, which in specific situations may be quite different from the employees' demands.

Should the local government, for whatever reason, force a more radical solution through the negotiations, union representatives try to compensate their members' loss. Introducing the "day care guarantee" is one way to secure members' interests. The reason is obvious: A "day care guarantee" is a commitment local politicians make to users with benefits equally distributed between day care workers and users. Such a clear commitment to users is more valuable to employees in day care institutions, because it is more credible. It is simply more difficult for local politicians to take away a once granted service than to change a reform model. In addition, a promise made by politicians to employees is not only surrounded by uncertainty, it can also be costly to defend it:

Suppose .. . that the struggle for economic advantage were to take place within a framework
in which actors occasionally succeeded in usurping the property rights of others. . . .
Economic actors would be concerned with more than simply making efficient choices about
the use and disposition of their property. They would also be concerned with taking action to
protect their rights from usurpation - and with making current choices about their property
that recognize and adjust for the possibility that other actors might seize their rights to the
property of the future. . . . Something even more fundamental is now at work, shaping
everyone's calculations and decisions - uncertainty about the very basis of all transactions.
Welcome to politics (Moe 1990, 123-124).

Thus, if local governments press for relatively radical reforms, we should
expect to see that a "day care guarantee" follows such reforms.

If unions (day care workers) have a significant influence on reform, we expect that in no (or very few) municipality will both budgets and parental influence be pressed to maximum degrees. Further, local governments simply have to compensate union interests by introducing a "day care guarantee" in municipalities where union interests are not followed.

Results

A questionnaire was sent to the administrative personnel in all 275
municipalities in Denmark.3 Since it is the administrative personnel who has

negotiated the reforms with union representatives, they were expected to
know the content of their reform choice.

The answers allowed all participating municipalities to be placed on both dimensions: Minimum or maximum degree of parental influence (voice). In municipalities where parental boards where given more than the minimum degree of influence, as specified by the Ministry of Social Affairs, they were given the value "maximum." Thus, the operationalization is biased in favor of reform, which means that all small deviations from the ministry's list are reported.

Further, the questionnaire asked about budget constraints. This included questions about the possibility of transferring surpluses and deficits from one budget year to the next. In addition, questions were asked about the control of salary related entries on the budget. Again the operationalization favored reform so that all steps towards more decentralization are reported.

By using relatively simple dimensions to specify reform models, all
Danish municipalities could be placed in one, and only one cell, in Table 2.

With few exceptions, the local governments lean towards employees' and unions' demands in reform situations. This suggests that something, other than pure efficiency considerations guide the local politicians in the decision of which reform model should be installed. However, the choice of reform also implies that unions accept external costs of negotiation and internal costs of organizing workers, whereas local governments seem more interested in avoiding conflicts in the sector.

The point is that any hierarchical institutional arrangement producing simple welfare services where unions bargain with political principals experience ex ante negotiation or transaction costs which cannot be expected to reduce ex post transaction costs stemming from opportunistic actors, who do not follow the new rules of the game. Due to the unequal distribution of costs and benefits in favor of employees and their unions, the local government "satisfies" by accepting limited reforms which in addition also seem to please users. As Table 2 shows, the results of reform initiatives confirm that local governments, in more than 44 percent of the cases, lean more towards employees' preferences for governance structure on both dimensions. Users (and potential users) do not disagree with this choice, because their interests are represented by employees and unions. To rational

users, this also implies that they only have to pay a fraction of the transaction costs between the unions and the local governments. In return for supporting employees' demands for governance structure, users accept - within a reasonable frame - the output of the day care institutions.

41.5 percent of the municipalities reached a contract where extra decentralization elements have been attached to the "least possible reform" as imposed by the Ministry of Social Affairs. This suggests that local politicians do make reflections regarding a coherence between institutional arrangements and institutional performance. It does, on the other hand, also indicate that local politicians face a tough bargaining situation with the involved actors who are not willing to yield well-established institutional positions.

Only 14 percent of the local governments represented in this survey can be placed in the radical model. While employees were successful in following the strategy to prevent local governments from implementing relatively extreme reforms, they were not expected to accept such radical reform without getting something in return. This "something" could be a "day care guarantee" which secures user demand for day care and day care workers' jobs. Although a straight forward causal relationship between choice of reform model and possibility of having a day care guarantee is not advocated here, such results (as a proxy) could show that such contract bargaining might influence local governments' day care policies.⁴ Therefore, the questionnaire asked about the implementation of the "day care guarantee" in each municipality. In addition, day care workers who "lose the battle" of reform would have to consider the reality of their losses, since a large degree of parental influence is only damaging if parents actually use this influence, just as budget decentralization only harms day care workers if the local governments use the new rules to cut budgets.

Table 3 suggests a quid pro quo bargaining situation between the day care institutions and the local governments, where municipalities who have installed less popular reform models tend to give a "day care guarantee" to users. Less than half of the municipalities with the conservative model have a day care guarantee, whereas 85.7 percent of the municipalities implementing the radical model also guarantee a day care place to all users.

The outcome is rational to all actors involved in the game at least in the short run. In the longer run the local governments may face flexibility problems under the "day care guarantee" as mentioned earlier, especially if - or rather when - the total number of children drops.

Conclusions

This article employs a transaction cost approach to the study of the way the production of day care services are being produced in Denmark. The reason is empirical as well as theoretical. "Contracts" seem to appear in many aspects of the production of welfare services in Denmark, at least as a metaphor for the relationship between the sponsor and the service producing institutions. However, production problems are far from solved by implicitly or explicitly signing a contract.

Two basic assumptions are used in this model to show how bureaucrats control their governance structure in reform situations: 1) Users only pay a fraction of the costs, and 2) the quality of the service is laid down in a set of rules - here rules (formal or informal) about the workload for pedagogues. This situation creates excessive demand for welfare services, because users get incentives to demand more service. If rules about quality are firmly established and user payments are much lower than the real costs of production, user demands equal the supply from public welfare producing institutions. Notice that the reduction of the maximum parental payment for day care from 35 percent to 30 percent as a result of the "child package," which passed through parliament in 1992, fits well into a union strategy of securing demand. In addition, user preferences shift towards the supply from these public institutions.

In order to obtain potential gains from the institutional arrangement, day care workers have to employ two strategies to ensure self maximization in the long run. First, they must have union monopoly in the production. Internally, this means that day care workers must be well organized, a condition Danish unions have no problem satisfying. Externally, unions must take part in the negotiating process in reform situations on both the national and local levels. Although both strategies involve organization and negotiation costs, day care workers seem to receive the anticipated benefits. Secondly, unions must control the reform process. Thus, they have to keep certain issues off the agenda when politicians discuss new ways of organizing day care. As shown, the private pool schemes have only a fraction of the "market" for day care services in Denmark. These issues are what I have termed the two basic assumptions for practicing preference shaping.

Data collected from 147 out of 275 Danish municipalities show that these strategies pay off. Only 14 percent of the municipalities in this survey have implemented a reform that we would expect day care workers to dislike. The reason we would expect day care workers to dislike the radical model is the potential use of voice among parents (users). Whether users actually employ their voice option remains an empirical question open for further investigation. However, day care workers are not expected to accept relatively extreme reforms without getting something in return. A "day care guarantee," in which all families with children from 0 to 6 years of age in the municipality receive an offer for the desired service, was suggested as a "reasonable" prize. Data suggest that more extreme reform models are followed by a "day care guarantee." Hence, a quid pro quo bargaining situation between unions and local governments seems to exist.

The approach advocated here and the preliminary data presented are in a developing stage. Yet, it seems obvious that organized interests must and do seek to influence the contractual relationship between political sponsors and the welfare service providers in the early ex ante phase in order to obtain possible gains in the longer ex post phase. Likewise, it is obvious that there are transaction costs (and benefits) distributed though these complex contractual arrangements. These gains could be acquired through ex post opportunistic strategies involving maximization of slack resources, securing professional values in the production, etc. However, employing the contract approach also allows for investigation in how these possibilities are created ex ante. Thus, further work needs to be done before a firm connection between ex ante bargaining strategies and ex post benefits can be established. Such a step could be taken by analyzing how different reform models allocate (transaction) costs and benefits to involved actors in, for instance, the day care sector.

---

NOTES

1. This base seems secured for the unions, at least in Denmark (Nørgaard 1997), and Jensen, for instance, reports that 94 percent of all educated day care workers are members of a union (Jensen 1994).

2. The costs of buildings are not included in the running costs but are paid by the local authorities who normally also own them.

3. The questionnaire is part of a larger research program which also includes case studies concerning ex post opportunistic strategies carried out by day care workers in reformed institutions.

---

3. The questionnaire is part of a larger research program which also includes case studies concerning ex post opportunistic strategies carried out by day care workers in reformed institutions.

4. Also to local politicians, a day care guarantee could be attractive as a way of increasing re-election chances.

ACKNOWLEDGEMENTS

I would like to thank Asbjørn Sonne Nørgaard, Jørgen Grønnegård Christensen, Thomas Pallesen, Peter Nannestad, Lars Dahl Pedersen, Patrick Dunleavy, and other members of the ECPR workshop "Shaping Political Behavior: The Institutional Formation of Interests" (ECPR Joint Sessions of Workshops, Oslo 1996) for comments on an earlier draft.

REFERENCES

Alchian, A. A. & Demsetz, H. 1972. "Production, Information Costs, and Economic
Organization," American Economic Review 62, 777-95.

ALA (Association of Local Authorities). 1994. Økonomisk Pres og Styring - Et flerårigt
perspektiv. Økonomiske Publikationer (9). Copenhagen: Kommunernes Landsforening.

ALA (Association of Local Authorities). 1996. Dagpasning under pres. Økonomiske
publikationer (12). Copenhagen: Kommunernes Landsforening.

Baker, G. P., Jensen, M. C. & Murphy, K. J. 1988. "Compensation and Incentives: Practice vs.
Theory," The Journal of Finance XLII, 593-616.

Bertelsen, O. 1991. Offentlig børnepasning, SFI Rapport 91:4, Socialforskningsinstituttet.
Copenhagen.

BUPL. 1993. Faglig Håndbog. Copenhagen: Forlaget Børn og Unge.

Christensen, J. G. 1992. "Hierarchical and Contractual Approaches to Budgetary Reform,"'
Journal of Theoretical Politics 4, 67-91.

Christensen, J. G. 1995. "Institutional Reform and the Governance of Education," Paper.
ECPR, Department of Political Science, University of Aarhus.

Cyert, R. M. & March, J. G. 1963. A Behavioral Theory of the Finn. New Jersey. Enelewood
Cliffs: Prentice-Hall.

Damgaard, J. B. 1996. "Transaktionsomkostninger i reformsituationer i det politisk ledede
hierarki," Paper, NOPSA Conference, Helsinki 15-17 August. Aarhus Universitet.

Dietrich, M. 1994. Transation Cost Economics and Beyond. London: Routledge.

Dunleavy, P. 1991. Democracy, Bureaucracy & Public Choice. Wheatsheaf. New York.
London: Harvester.

Esping-Andersen, G. 1990. The Three Worlds of Welfare Capitalism. Cambridge: Polity Press.

Hirschman, A. O. 1970. Exit, Voice, and Loyalty. Cambridge, Mass: Harvard University Press.

Jensen, C. 1994. "Fragments for a Discussion About Quality" in Moss. P. & Pence. A., eds..
Valuing Quality in Early Childhood Services. New York: Teachers College Press. Columbia
University.

Ketcher, K. 1991. Offentlig Børnepasning i Retlig Belysning. Copenhagen: Jurist- og
Økonomforbundets Forlag.

Knight, J. 1992. Institutions and Social Conflict. Cambridge: Cambridge University Press.

Lispky, M. 1980. Street-Level Bureuacracy. Dilemmas of the Individual in Public Senices.
Rüssel Sage Foundation, New York.

Migué J-L. & Bélanger, G. 1974. "Toward a General Theory of Managerial Discretion."
Public Choice 17, 27^15.

Miller, G. J. 1992. Managerial Dilemmas. Cambridge: Cambridge University Press.

Ministry of Social Affairs. 1993. Parental Employment and Caring for Children. Copenhagen:
Socialministeriet.

Mitchell. W. C. & Simmons, R. T. 1994. Beyond Politics. Westview Press.

Moe, T. 1990. "The Politics of Structural Choice: Toward a Theory of Public Bureaucracy" in
Williamson, O. E., ed., Organization Theory. From Chester Barnard to the Present and
Beyond. New York: Oxford University Press.

Mourtizen, P. E. 1991. Den Politiske Cyklus. Aarhus: Politica.

Niskanen, W. A. 1971. Bureaucracy and Representative Government. Chicago & New York:
Aldine Atherton.

Nørgaard, A. S. 1997. "The Politics of Institutional Control: Corporatism in Danish
Occupational Safety and Health Regulation and Unemployment Insurance: 1870-1995,"
Department of Political Science, University of Aarhus.

OECD. 1990. Employment Outlook. Paris: OECD.

Ostrom, E. 1983. "A Public Choice Approach to Metropolitan Institutions: Structure,
Incentives, and Performance," The Social Science Journal 20, 79-96.

Pade, S. & Glavind, N. 1993. Rapport om puljeordninger. Copenhagen: Bureau 2000.

Savas, E. S. 1982. Privatizing the Public Sector. New Jersey: Chatham House Publishers.

Vedel-Petersen, J. 1992. Dagtilbud for børn under skolealderen, SFI-rapport 92:8,
Copenhagen: Socialforskningsinstituttet.

Williamson, O. E. 1981. "The Economics of Organization: The Transaction Cost Approach,"
American Journal of Sociology 87, 548-77.

Williamson, O. E. 1985. The Economic Institutions of Capitalism. New York. The Free Press.

Williamson, O. E. 1990. "A Comparison of Alternative Approaches to Economic
Organization," Journal of Institutional and Theoretical Economics 146, 61-71.

Wilson, J. Q. 1995. Political Organizations, Princeton, New Jersey: Princeton University
Press.