University of Helsinki

1. Introduction

The objective of this research is to contribute to the study of the structural causal determination and systematic evaluation of the production and distribution of public goods. The focus is on the national central government policy strategies and their success in realizing the objectives of the legislation pending on the production and distribution of the major merit goods, i. e. public services.

The study will first outline a general typological framework for the characterization of the central government policy strategies, the control ideologies they imply, and the desired and undesired consequences they lead to. Secondly, the development of the Finnish system of the production and distribution of public services is described briefly, and the interpretation of the development is geared to the changes of the central government policies and control strategies. A more detailed case study evaluating the success of the change of central government policy strategy is carried out in the field of social welfare services. An attempt is also made to develop a more general evaluative design that could be used in general to judge the success of central government policies.

The notion of 'invariance' in an evaluative research design

We can first briefly outline the meaning of the idea of 'invariance' in relation to the two principal research perspectives in the study of the production and distribution of public services, i. e. 'structural causal' and 'systematic evaluative'. We understand that the notion of 'invariance', having already a well-known and established meaning in social research, can be conveniently applied and more fully exploited also in the area of collective goods. We believe as well that introducing

---

* The authors are indebted to Ilkka Heiskanen for his constructive suggestions in structuring this paper.

* The authors are indebted to Ilkka Heiskanen for his constructive suggestions in structuring this paper.

this notion should give more theoretical coherence to the policy-analytical explanations of the politico-administrative mechanisms or structures regulating the distributory processes within the public sector: a refinement of our explanatory apparatus follows in the subsequent sections.

In sociological terms, following Johan Galtung and Erik Allardt, 'invariance' is defined as a generalization held to be valid also for situations occurring in the future.¹ In this vein we understand invariance as a consistent relationship between given variables, recurring in a defined situation and self-perpetuating. In the field of collective goods, for instance, a well-established causal invariance is the relationship that tells us the hopeless determination of the local output on the disposable local resources.

Both of the research perspectives we mentioned, structural causal and systematic evaluative, aim at finding invariances. The main difference between them is that the evaluative approach starts where the structural causal leaves off. This means that in causal analysis the researcher is satisfied if he has been able to locate and define invariant relationships between variables, e. g. factors which determine policy outcomes. These relationships are interpreted as structural consistencies of decisive importance for the result or outcome of a given public policy. The systematic evaluation approach incorporates the idea that the revealed invariances, because they are generated by politico-administrative structures or structural arrangements, are changeable. Therefore invariances are seen as reflecting the ends of given politico-administrative arrangements and, consequently, a given policy change in these arrangements should result as a breakdown of a corresponding invariance.

The researcher employing a systematic evaluation approach in the above sense looks for invariances to find out their connections to relevant government policies. The focus is on the evaluation of the government policies or administrative arrangements to establish a given invariance - or what is often more important - to break an established invariance which has become undesirable or which is not in accordance with the desired course of action. An 'old' invariance can be replaced with a 'new' desirable one and, accordingly, the policy change is supposed to result in a breakdown or modification of a given undesirable invariance. Therefore the effectiveness of the policy strategies or politico-administrative structures to introduce a given policy can be evaluated empirically according, for instance, to their ability to generate desired invariance changes.

Throughout this study we try to entertain the idea that a given policy strategy change can be matched against a corresponding invariance. Our study is not concerned much with the general problems of evaluative inquiry, e. g. what the basic categories of the evaluation of public policies are.² On that topic we take a rather pragmatic view: the evaluation of the policy strategy change is made with reference to the aims proclaimed by the decision-makers. The administrative apparatus by means of which the desired policy change is carried out to affect the given undesired invariance is explained later, and the evaluation of the success of the policy strategy change is made accordingly, which means that the observed results are simply related to the aims.

2. Observation of Needs and Willingness to Redistribute: A Typology of Policy Strategies and the Control Ideologies they Imply

The starting point in the following discussion is that we understand an invariance, irrespective of its desirability, as being generated by the politico-administratve structures which carry out governmental policies in a defined situation and at a particular point in time. A policy is here loosely defined as a legislatively stated course of action, authoritatively formulated and substantive in content, e. g. a piece of legislation designating the aims and the tasks of the local authorities in a given field of public services. Politico-administrative structure is a catch-all notion in this context. It is essential to recognize, however, that we define it as a permanent structural arrangement within the public sector with the purpose of executing government policies. It can be conceived as a stable linkage system mediating between the two levels of government, central and local. We call it 'policy strategy' to emphasize its close interconnectedness with a substantive policy. The crucial argument that follows is that it is precisely the policy strategy which has the capacity to establish or break the invariances, and the given policy only shows the substantive target, target population, and the sectorial-specific ends to strive at within the common and more 'universal' framework of the existing politico-administrative structures or policy strategy.

Johan Galtung asks in his seminal essay on 'invariances' how one finds third variable(s) with which it is possible to transcend a given 'invariance', and he finds three basically different situations to start with: (1) the variable is already part of the invariance, (2) the invariance is empirically imperfect, and (3) the invariance is empirically perfect.³ Galtung's idea is that a new reality can be created by changing the critical variable or by giving a new value to it. In our special field, invariance-breaking presumes establishing a new invariance which replaces the old one, and this implies the need for changing the government policy strategy. For the moment we shall reserve judgement on the invariance topic as very straightforward, with no references to the constraints counteracting the desires of the decision-makers to transpose or establish invariances with the aid of a given policy strategy change. Such a discussion is, of course, of vital importance when interpreting the possible failures of government policies and policy strategies, and therefore we leave the discussion on the structural constraints to follow in the empirical evaluation section of this study.

In Finland, the development of policy strategy in the field of public services has aimed at the realization of 'egalitarian' values, a manifestation being the determined pursuit of sectorial policies of levelling out the regional differences in the production and distribution of public services. Consequently, the most stubborn invariance to be transposed by means of the policy strategy is, of course, that between the local output and local disposable resources. This is what we might call the 'master invariance', and as long as it prevails it restricts passage of the 'master policy', regionally equal distribution of and access to public services, or any single policy part thereof. This is also, in a nutshell, what we try to evaluate in the empirical sections, i. e. the success of the policy strategy change

to break the 'master invariance' in a particular field of social welfare services.
What is, then, the policy strategy of the Finnish system of the production and

distribution of public services? How could we describe it in general terms of theoretical interest and at the same time preserve substantiveness and empirical applicability of such a construct? In the case of Finland we argue that the basic linkage system determinants and the character of a particular policy strategy for affecting the 'invariances' via the central-local relationships can be reduced to and deduced from the two following considerations:⁴

(1) To what extent do the public decision-makers observe the needs to be

satisfied and how do they structure them in their policies? And

(2) To what extent do the public decision-makers desire to change the existing

system of the distribution of benefits and costs?

The problem of needs in public decision-making cannot be treated here in any detail. The first consideration largely determines the role and the status of the public sector in a society by defining the tasks of the public economies and establishing the division of labor between the public and the private. When applied to defining the character of the prevailing system of central-local linkages we understand the meaning of the first consideration in a much narrower sense. To us it indicates the perceived need of the central government to control and guide the local performance in the production and distribution of public goods.

It is conceivable, of course, that the degree of control is dependent on the rankordering of the priorities of the needs meant to be satisfied collectively - by means of collective resources - and these considerations respectively will be incorporated in public policies and control strategies. All this reasoning of the public decision-makers is finally reflected in the legislation on public goods, and in particular, in the varying stringency of the central government's desire to control the local performance. In our typology of the policy strategy, the abovedescribed control dimension will be the first determinant.

The second consideration leads us to the arena of resource mobilization and allocation of costs, which means decision-making of the extent and thoroughness to change the existing system to distribute costs and benefits in the society in the first place, and, secondly, to define the 'loosers' and the 'winners' in such redistributory decisions. We leave the question of the individual level to move over to the level of the central-local linkage system.

Within the public sector the problem is identical; if the central government aims at changing the balance of benefit distribution among the population of the individual municipalities, it has to change and regulate, accordingly, the prevalent system of the distribution of costs. The mechanism to affect the distribution of costs within the Finnish public sector is the widely extended central aid system, in which the 'winners' and 'loosers' of the moment are delineated according to a complex grading formula.

A very important specification of the Finnish central aid system should, how-

ever, be made in this context; there is no equivalent in Finland to the block grant system such as is the case, e. g., in Great Britain. Central aid, even if graded according to specific criteria, is allocated only to carefully designated spending categories. Despite this fact, we consider the present Finnish graded central aid system as redistributive. The other option, which was used almost without exception until 1969, is what we call merely the 'distributory' or 'compensatory' mode of distribution of costs within the public sector; the aim was to compensate the local government performance, regardless of the wealth of a particular municipality and according to a fixed provision, let us say 40 % of the acceptable expenses.

With a cross tabulation of these two dimensions, degree of central control and
the mode of the distribution of costs between the central and local levels of
government, we have a typology depicting the basic types of policy strategies:5

The increase of supply and demand of public goods has, of course, conditioned as well as caused the central government to change more rapidly the policy strategies to meet collective needs. The trend of development has very clearly been from the 'liberal' type of strategy in the 'egalitarian' direction, mostly from type 1 to type 4.⁶ This means that the central level of government has required the local governments to follow more closely the nationally uniform standards in the production of public goods. Simultaneously, this has also led to the adoption of the redistributive principle in the sphere of the distribution of costs within the public sector.

It is fairly easy to cite reasons for the observed development in the policy strategies. The type 1 policy strategy presumes excessive freedom of the local government in relation to the central level, indicating the liberalistic idea of safeguarding local autonomy. This has been the major determinant leading the financially weak local governments to float around rather uncontrolled, depending upon the influence of 'market mechanisms'. In fact, the easy yielding of local governments to the market forces under the liberal policy strategy of the central government resulted in a detrimental colinearity: the service output of the local governments tended to follow the regional differentiation of wealth, labour, and capital.

This observation becomes significant when we emphasize that the Finnish local

governments with their sovereign budgetary and tax-levying rights are strongly and directly tied to the supply of local resources. This is also to insist that the 'master invariance', the determination of local output largely on the basis of local resources, became established as an undesirable consequence of the overemphasis of local autonomy on one hand and of scarcity of national resources impeding massive compensatory efforts on the other. Thus the later change of the policy strategy became necessary to comply with the growing demands for even regional development.

It is generally believed that the 'egalitarian' type of policy strategy will disengage the local governments from the magic circle maintained by the skewed distribution of resources. This is based, of course, on a very simple conception of the possibilities of central guidance, and it implies that local level decisionmaking is mainly seen as only reconciling needs and resources with nationally uniform legislative guidelines.

For the evaluative case study we shall specify the linkage arrangements in more detail later. The fact remains that the above described broad administrative context is a prerequisite to understanding the functioning of the local political processes in Finland.

3. From Liberal to Egalitarian Policy Strategy: Some Macrolevel Considerations

Before turning to analyze our specific case of the total services we will first in more concrete terms briefly point to those macro processes which have preceded and contributed to the change of the policy strategy from liberal to egalitarian in Finland in the 1960'5. Thereafter we will substantiate the discussion by showing some major financial effects which can be seen as parallel developments revealed by the SNA (System of National Accounts) figures.

As our typology implies, the change of the policy strategy from liberal to egalitarian in the central-local linkage system means two separate things. First, there should exist an ever-increasing interest by the central government for determining the amount of need for public services at the local level so as to have a sufficient information base for the central level allocations. Secondly, there should also be an increased central government orientation to redistribution, i. e. to levelling out the differences in the actual need satisfaction between municipalities, the Units being subjected to the central government measures.

The first one of these inevitable changes can be expected to stand out as an increased central government interest in surveying the state of the local governments in general, and as an increased desire to secure a firm foothold in the regional and local planning processes in particular. The second of these changes should show up in the level of legislation, i. e. as an ever-increasing amount of legislative norms clearly defining the objects to be carried out with the assistance of graded central government allocations (whatever the grading in each particular case might be). And thirdly, both of these macro-level changes should also be

reflected in an increasing amount of expenditure on the public services shouldered
by the central government, and in an increasing share of the community output
which is taken by the public economies.

We think that all these macro-level changes must be validated before one can indeed demonstrate that the change of the policy strategy in the sense we have suggested has actually taken place in Finland in 1960'5. There is, in fact, ample evidence in Finnish studies which substantiates the validity of the first argument.⁷ Perhaps the most important proof, however, is the central government's monitoring system of grading the municipalities for the allocation of central finance. This is a sort of a 'real time' system, in which the level of performance in public services, the overall financial development, changes in supply of local resources, in the price of income tax unit, etc. among the municipalities, are monitored annually and the revision in the position of each particular municipality with regard to its location in the grading scale is made accordingly.⁸ The second macrolevel change, the increase of legislation aiming at the levelling out of the differences in need satisfaction, is as evident in Finland as anywhere else in Europe, so we do not feel guilty for having violated the principles of scientific inquiry by disregarding in this paper the validation of this argument.⁹

The third macro-level of observation where the changes of the two previously mentioned developments should be found is the level of public expenditure. Here we pause for a moment to take a closer look at the development of the Finnish public sector.¹⁰ We pass over the vast material of a historical and of a more recent but politically transient nature, and we shall try only to elucidate the few factors which are related to our limited theme.

Figures 1 and 2112¹¹ are constructed on the basis of the SNA statistics with the spending of the two levels of government illustrated separately. This procedure provides the reader with a clear graphical presentation of the magnitudes of government spending in Finland, and also a profile of the differences between the central and local levels of government. The actual spending figures are related to the gross domestic product at market prices, which has the virtue of depicting the relative shares the government spending takes out of the total community output.

The financial macro-effects that the recent policy changes in Finland have brought forth can be seen by visual inspection of the figures. The change of the policy strategy is indeed a clear parallel to the overall growth of the Finnish public sector. We will first point to two features which relate to our earlier discussion: (1) in Figure 1 the share of the central government transfers to the municipalities and municipal associations shows emphatically the increasing interest/responsibility the central level of government has encountered in assisting the local level to produce and distribute public services. (2) The production and distribution of public services has in turn become a responsibility of the local governments, as shown in Figure 2 in the 'consumption expenditures'. In Finland this is a matter of necessity because the central government lacks an areal service network. The spending of the local governments is rather exhaustively described by accounting for the growth of the public services - mostly health,

education, and social services - which take up the lion's share of municipa
consumption expenditures. Practically all of the central finance is via the comple:
grading system allocated to the municipal consumption expenditures.12

The development of the central-local linkage system from liberal to egalitariai has, of course, strengthened the cooperation between the two levels of govern ment, and consequently led to a pronounced financial control due to the peculia rity of the central aid system. This has resulted in an uneasy compromise witl

the earlier predominant ideal of local autonomy; the withdrawal has taken place in face of a potentially intolerable tax burden. Within the public sector, development in recent years points clearly to a key trend: the concentration of power at the central level of government.¹³

4. How to Test the Success of the Policy Strategy Change: An Evaluative Design

As we stated at the beginning of this essay the final aim will be the testing of the policy strategy change in the field of public services to show how a politicoadministrative arrangement in a specific case is able to direct the local government performance in the desired direction, and, especially, how the policy strategy change copes with the undesired invariance, i. e. the determination of the local output on the resources. Before going into the details of our case study, we might diagrammatically simplify the starting point for two reasons: firstly, to clarify the concepts used in the evaluation; and secondly, to characterize in simple terms the hierarchical Finnish system of central-local linkages. The following diagram is adopted from the systems analytical study by Laszlo, Levine and Milsum, and it is applicable to our research design.¹⁴

The diagram emphasizes the necessity to conceive of the linkage system as a hierarchical interplay of the two levels of government, the lower level or local government being clearly subordinated to the central level of the government. We must, however, recall that the closer subordination is the case only in the field of public services. We should also pay attention to the fact that the diagram is essentially a description of a control system and also therefore acceptable to us.

The upper part of the diagram depicts in our case an individual local government as a 'dynamic system', which has the tendency of seeking a balanced internal condition, a tolerable correspondence between the inputs and the outputs. This sort of local dynamics, which is in our opinion a correct preliminary assumption, has led the financially weak local governments to yield to market forces, to decrease or increase their output in accordance with the supply of local resources (tax revenues). A highly undesirable consequence in terms of the growing demands on equal regional development has indeed resulted: the level of services of the weak municipalities lags far behind the wealthy ones. This means the establishment of what we have called the 'master invariance'; the principal aim of the 'control strategy', described in the lower part of the diagram, should be the transposing of such an invariance to guarantee a sound correspondence between the local performance ('actual system output vector') and the nationally desired standard performance ('desired system output vector').

The notion of the 'control strategy' in the diagram is identical with the notion of the 'policy strategy' we have used in our earlier discussion, and it incorporates the means of the central government to steer the local performance in the desired direction. This is why it is called the 'controlled input vector'. In the Finnish system it aims at establishing a homeostatic situation with a minimum of variance between municipalities in public services. As we have pointed out in a previous context, the means of control and guidance are (1) the specific national legislation, i. e. the specificity of legislative aims, and (2) the system of state aid,

i. e. the method of the distribution of the costs. By cross-tabulating these two dimensions, as was done on page 97, we arrived at the basic types of policy strategies - 'liberal' and 'egalitarian' - which of necessity are approximations of real life, but for certain are indicative of the mode of the control ideology they imply.

The crucial tool for the empirical evaluation of the success of the policy strategy is the notion of 'system error' shown in the lower part of the diagram. It is located between the 'control strategy' and the 'evaluation of performance', which should, for instance, show planners how to correct the control strategy if there are discrepancies between the 'actual system output vector' and the 'desired system output vector', i. e. if the performance of the municipalities does not coincide with the desire of the central government plans. In order to test the success of the policy strategy we should define what we mean by the system errors in each particular case, then try to operationalize them and finally to find them empirically.

It is a matter of choice between alternative models of explanation that determines the definition of the 'system error'. One might, for instance, wish to discover whether the local performance in a particular field satisfies the needs of the target population, and the system error will be defined accordingly. In our case the system error is used to evaluate a given policy strategy, and it is the actual discrepancy observed:

(1) As a difference in the aim of what is being achieved compared to what

is being desired, and

(2) As a difference in the methods of how they are applied as compared to the

rules designating their application (application of state aid system).

Both of the variables mentioned above are part of the policy strategy, and we believe that the success can be judged reliably by tracing the system errors. We omit the discussion on the tolerable limits of the magnitudes of the system errors, however important the topic might be. If, for instance, 'zero system errors' are desired, it can be done, of course, at the expense of infinite control effort.¹⁵ In our study we have avoided the debate deliberately.

5. The Case of Finnish Social Services: An Empirical Illustration of the Evaluation of the Realization of an Egalitarian Policy Strategy

Testing the'functioning of the egalitarian policy strategy (Type 4) against the aims of the legislator

As we pointed out in the previous chapter the basic task in our evaluative design is to find out the system errors (1) both in the realization of the aims of a particular public service and (2) in the realization of the method of recompensation.¹⁶ Our case is a special field among the social services - the municipal home care services

- which means arranging help and care directly in the home of a citizen or a family who needs it. The final aim of the legislator requires the municipalities to provide in rural municipalities one home care assistant per 1500 citizens and in urban municipalities one assistant per 2500 citizens.

This performance cannot, however, be enforced by law, and therefore the legislator's only possibility is to come half-way to share the production costs, which at the municipal level show up as a variable level of tax burden accrued by the resources allocated for the given services. Therefore the concrete aim is, in fact, the equalization of the cost basis for producing the required amount of home care assistant vacancies among the municipalities. The system error for each municipality is the distance which remains to the ideal situation of equal cost basis after the central government financial measures have been carried out.

The system error in the method of recompensating the costs is found by regressing the actual allocated state aid (variable Y) on the criterion variables which according to the specific legislation should determine the amount of recompensation. The criteria which in a pronounced manner accentuate the local resource factor are condensed into a measure which classifies the municipalities on a ten point scale (variable Xi). Point zero indicates the 'winners', i. e. those who get the most compensation, with the compensation decreasing gradually as one moves closer to ten. The other variable in the regression model is the real net allocation of the municipality for the service performance (variable X₂). Because these two criteria, variables Xi and X₂, jointly (not independently) determine the state aid allocation, the regression model should of necessity be a multiplicative one. In this case the result turned out to be the following:

The system error is the residual Y, the deviation of the observed amount of the
state aid allocation from the predicted allocation as given in the above multiplicative
regression equation.

The basic set-up for the evaluation of the egalitarian policy strategy is brought about by first standardizing the system errors, tricotomizing them by setting the class boundaries at a distance of one standard deviation, and then crosstabulating the system error variables. The result is shown in Table I. The system error SE_R -minus indicates that the actual recompensation is less than predicted, zero means that actual results approximate what is expected, and SE R -plus indicates that the municipality receives more compensation than what is expected according to the application criteria. The system error SE_A -minus shows that the actual costs of producing the required performance are less than the average, zero means that the actual cost approximates the average, and SE_A -plus indicates that the costs are above the average. Thus the SE_A -distribution shows the deviation from the desired aim of the legislator with regard to equal cost burdens, and correspondingly, the SE_R -distribution depicts the discrepancies in the application of the method of recompensation.

Table I indicates the general set-up of how the Finnish municipalities behave under the central government policy strategy which we characterized as egalitarian. The 'ideal' setting is clearly cell 5, in which both of the system error types approach zero. The dispersion is, however, remarkable around the ideal situation, which includes only 29 % of the muncipalities. It should be noted that cell 8, for instance, indicates that receiving compensation in an average way tends to place almost a fifth of the municipalities in a disadvantageous position with regard to the proclaimed aim of the legislator. Correspondingly, getting more compensation than expected by the model tends to move a number of the municipalities to a favourable position with regard to others (cell 3). These cases are clear implications of the character of the prevalent control system of the central government, and we shall return to the interpretations a bit later.

In addition, the correlation coefficient -.38 between the two system error variables shows that the deviations in the method of recompensation tend to produce divergence in the realization of the aim. However, the regional variation of the cost factor has decreased considerably by virtue of recompensation: Without the recompensation the coefficient of variability (.100) would have been 56, and after the recompensation it was 25. This testifies, of course, that the system basically works as desired, but not in a very predictable manner as the scatter in Table I shows us.

We shall not report here a mass of data utilized to find the connections of
the different aggregates of the municipalities (cells 1-9) to numerous background

variables. The search, however, showed emphatically that the least deviant cases represented averages in most of the background variables. The conclusion which can be drawn is that the egalitarian control strategy reacts inadequately in non-normal situations. This is understandable if one knows the mechanisms of the control strategy; the method of recompensation, especially the grading of the municipalities from 0 to 10, is based on the 'universal' criteria emphasizing the resource factor. The local peculiarities are disregarded, which make some handicapped others advantaged with respect to recompensation, but not always in a manner which is nationally desired.

In this context we might also focus the perspective to show that the egalitarian
strategy cannot satisfactorily remove the municipalities from resource determination
as indicated by the correlations below (for the explanations, see p. 107):

The system errors in the realization of the aim of the legislator (SE_A ) associate clearly with the resources of the municipalities in an undesired manner; the more resources the less costly the local performance tends to be. This means that the redistributive compensation leaves the 'master invariance' intact.

Systems errors in the recompensation (SE_R ) correlate as clearly with the high unit costs and with the service level (services per capita), which in turn suggests that in the application of the recompensation method such special criteria are in fact taken into account. One should raise the question whether such extra compensation indicates that the system is open to manipulation, and further, is that possibility a privilege of the few only? A reasonable remedy should be the increase of the built-in elasticity of the recompensation method, e. g. at least a partial allocation of central government monies as block grants to be used at purely municipal discretion.

Changing the policy strategy from liberal to egalitarian

The above discussion of the system errors emphasized the divergencies which could be found between the actual and desired state of affairs. It does not, however, provide evidence of the possible improvements which, by virtue of the egalitarian policy strategy, might have come about at the level of the local service network. Therefore, we will shortly discuss the determinants which seem to affect the service network in situations where the administering of a social welfare service in the central-local linkage system is disengaged from one type of control strategy and moved under another control system, i. e. from liberal to egalitarian, according to our typology (page 97) from type 1 to type 4. We might recall that the liberal policy

strategy is the one with loosely defined national objectives and with no recompensation
based on grading the municipalities (see page 99). The substantive object of
our study remains the municipal home care services.

In the following regression model the variable to be explained is the increase of the home care services during the first three years after implementation (1971-1973) when the egalitarian policy strategy was in effect. The explanatory factors are as follows:

(1) The previous activity: the change of service level (home care assistant
vacancies per capita) from 1966 to 1970, which is the last five-year period
of the liberal policy strategy

(2) The service level in 1970: the level attained when moving into the egali

tarian control system

(3) The unit costs: allocated resources/the number of vacancies

(4) The effectiveness of the use of the services: the provided services/one vacancy
(5) The need of the services: the relative number of large families
(6) The wealth of the municipality: total tax revenue per capita (see page 106)

The regression analysis clearly shows that the most essential effect on the improvement of service level is attributable to the previous activity of the municipality. This observation simply means that moving to another control system has not dramatically changed the behaviour of the local governments; instead, the previous line of action has continued. This discovery is reinforced with the evidence that the already existing high service level at the time of moving to the egalitarian control system has curbed, or at least restrained, the desire/need of the municipalities to improve their service level (as the behaviour of variable 2 clearly shows us). It should also be noted that the high level of unit costs tends to work in the same direction. These observations raise important questions about the efficiency of the control systems to change reality, to transpose invariances just by giving new values to the variables of the control strategy.

As an interesting alternative interpretation of the function of the egalitarian
control strategy we should emphasize the behaviour of variable 4, the effective

ness of use of the services. In systems analytical terms it is a 'withinput'. and clearly exogenous to the control system in its entire anonymity.¹⁷ The behaviour of the variable shows that the demands to improve the service level are also generated within the processes of carrying out a given service activity. This sort of 'withinput' does not necessarily have anything to do with objective needs, but rather, it is closely linked with the propensity of the citizen to utilize the existing service structure; it is as if the customers are socialized to make extensive use of the well-behaved service structure, thus exercising distinct pressure towards further elaboration of the services. Demands to improve the service structure come out of its own operative efficiency. This shows that not merely the supply of the services but also the variable propensity of the people to consume what is offered is conducive to the development of the service networks in general. In this matter the control strategy - be it liberal, egalitarian or whatever - does not make any difference.

As a final comment we should also add that the meaning of the need-variable (No. 5) remains obscure. Statistically it is the least powerful explanatory factor, which means, of course, that the need consideration cannot easily be caught or accounted for by the prevalent policy strategy. This is fairly odd, but it reflects all in all what an extremely unspecified and difficult task it is for decisionmakers to try to cope with need considerations in general, and in the field of welfare services in particular.

As a balance, we have reason to point out that the change of the control strategy from liberal to egalitarian has not altered the existing service structure to any considerable extent. The course of action seems to continue as before the change. The egalitarian method of the redistribution of costs works as desired, i. e. equalizes to a certain extent the production costs, but nevertheless does not affect those factors which determine the service level. This means that despite the 'master invariance', the determination of local output on the resources is affected by reducing the relative cost factor; the persistence of the established local peculiarities remain intact, and the actual local performance goes on as before. The answers cannot be found by solely empirical observation of the municipalities; therefore we would like to focus the perspective at the end of our discussion on the theme of the structural constraints.

6. Conclusions: On the Structural Constraints in the Application of a Policy Strategy

In conducting this study we have had varying objectives, both theoretical and substantive. By introducting the notion of 'invariance' into the policy analytical tradition of our discipline we have made an effort to provide some new insight which could approach the fairly 'classical' themes of the governmental control strategies from an evaluative perspective. As an empirical illustration a case of the field of social welfare services was subjected to closer scrutiny for testing the research apparatus, and, of course, for demonstrating the capacity of a given type of the control strategy to comply with the aims of the legislator.

The substantive findings clearly suggested that the egalitarian control or policy strategy did not remove the municipalities from the determination of the 'master invariance', as we called the dependence of the local output on the resources. We emphasized, however, that the 'master invariance' was affected, but that the local peculiarities determining the service level remained intact. With respect to the aims of the legislator the findings do not confirm the judgment that policy strategy has been successful in attempting to transpose an undesired invariance. Therefore we will conclude with some broader perspectives on the interpretations of the control strategies in general, and of our Finnish case study in particular.

The integration of the controlled unit with the control system

The meaning of the first category can be condensed into a broad statement maintaining the argument that the subjects of a control strategy will have tendentious strivings to become integrated with the control strategy proper.¹⁸ This means that they assimilate the norms determining the behaviour of the steering mechanism and develop strict adherence to the administrative mechanics. By and large the integration process is desirable, because it guarantees the efficiency of the functioning of the system in routine matters; we might call this a socialization stage.

One should, however, raise the question of the undesirable consequences which are generated by the integration process. We would refer only to a couple of such undesirable effects brought forth by the Finnish experience.₁₉ First, the integration is apt to lead to avoidance of control in general. Secondly, it is apt to lead to conspicuous exploitation of the control strategy in two different senses: (a) in potential speculation to gain the most favourable status for the central government recompensation, and (b) in excessive and conspicuous use of central finance when made available. We believe that these sorts of chances are never distributed normally among the controlled units; quite the opposite is true. From the perspective of the democracy of government control these questions deserve serious consideration.

What does the integration or 'oversocialization' with the control strategy mean in our case? To base judgement on Table I does not prove, of course, that the deviance from the aims of the legislator, seen clearly in the general set-up, results in any specific sense from the factor we just mentioned. We argue, however, that the egalitarian control strategy is open to manipulation which is akin to such considerations, and these themes, in more concrete terms, are occasionally also openly debated among the technocrats in Finland. Whatever the frequency of the 'exploitation' and avoidance of the control strategy might be, it does pose indeed an important problem for further and more extensive research.

The local units as dynamic systems

In our evaluative design the municipalities under the central guidance were
depicted as dynamic systems seeking a tolerable correspondence between the in

puts and outputs. The control strategy of the central government was conceived as a mechanism to keep the local units in conformance with the nationally desired line of action. We cannot have evidence out of the limited material which is afforded by our case study for judging the strength of the internal dynamics which is generated by the local units under control. We can, however, following the reasoning by Laszlo, Levine and Milsum, put forward a fascinating question of the extent to which the forces generated within the dynamic system tend to nullify the forces which attempt to produce changes.²⁰

In an evaluative study the change-resisting forces have, of course, several interpretative meanings. In the most desirable sense those forces tend to cushion an attempt to change by some initial resistance. This sort of natural intertia prevents a control effort from showing up immediately; but if there are too many inertial forces on the local level, the control effort might never produce the desired outcome. This is the case in point if the inertia is exclusively borne out of a weak resource base, and the control system in this case should help by financial recompensation.

It should be emphasized, however, that there are inertial factors of great variety which produce resistance irrespective of the resource factor. One could find inertia in the local administration due to the lack of skill or abundance of skill as well, i. e. 'management inertia', or in the level attitudes of the citizens or politicians. Therefore, the deviances we found in our empirical case study should almost have been expected because the control strategy in Finland is basically a strategy of pure financial guidance. Even if we cannot specify on the basis of the data the effects of no-resource based inertia, the results of the regression analysis showing the persistence of the previously pursued policy of the local governments are clearly implicative of the existence of such factors.

On the basis of our previous discussion one might conclude that we have overemphasized the change-resisting forces which are generated by the subjects under the control strategy. This is not quite the case because we have aimed at pointing toward only some possible avenues for the interpretation of the results of an evaluative research design. In this vein it is sensible to specify that there are also resistance decreasing factors incorporated in the modern central-local linkage arrangements, which, at least in the long run, might turn out to be detrimental to participant democracy.

Those processes which indeed lower the inertial level to an undesirable extent are related to the local input processes. In Finland, where the system of production and distribution of public services is more and more committed to the nationally uniform determination of needs and services, this means accepting at least two undesirable features. First, realizing the typical situation of heavy input overload at the municipal level, those inputs which are not enforced by the central government or otherwise strongly supported will most easily become blocked up and they will likely be the ones which have only specific local significance. And the normal local input processes are thus constrained. Secondly, the restricted transfer of local needs and demands also gradually deprives part of the meaning and political significance of various local 'gate-keeper' groups -

both political and parapolitical - maintaining the responsiveness of the local political system to its environment. If this community is violated the local political system becomes, of course, less responsive to local needs and more responsive to the control effort of the central government.

From the perspective of an evaluative research design this means that an effort to introduce a desired policy change via central-local linkages should also be related to the functioning of local input processes. One might, as well, raise the issue of what all this means for maintaining autonomy and democracy in local input processes.

---

NOTES

1. Johan Galtung, 'Science as Invariance-Seeking and Invariance-Breaking Activity', Sociologisk Forskning 1-2 (1974), p. 66; Erik Allardt, 'Om samhallsforskning I: Den traditionella synens tvångstrojor', Nya Argus No. 17-18 (1975), pp. 223-226. The concept of 'invariance' is clearly defined in the essay by Galtung, op.cit. We have, however, taken the liberty to depart from it to fit the logic to an entirely different field of research.

2. See, e.g., Ilkka Heiskanen, 'On Democracy, Equality, Efficiency and Justice as Criteria for Developing Local Government and Designing Linkages between the Central and Local Levels of Government', Research Reports of the Institute of Political Science, University of Helsinki, DETA 17, 1975.

3. Johan Galtung, op.cit., p. 80.

4. The problematics of the central-local linkages in the production and distribution of public services in Finland are discussed extensively in Tuomo Martikainen and Risto Yrjdnen, 'Kollektiivihyodykkeiden tuotanto ja jakelu poliittis-hallinnollisesta nakokulmasta' (The Production and Distribution of Public Goods from the Politico-Administrative Perspective), Research Reports of the Institute of Political Science, University of Helsinki, DETA 10, 1974. The following discussion on the policy strategy is found there in more detailed and concrete form.

5. Martikainen and Yrjonen, op.cit., p. 77.

6. See, ibid., pp. 78-83, where we discuss the developments of the policy strategy choices in Finland in more detail. We should emphasize that the dominant policy strategy, especially in education and health before the adoption of the redistributive mode of the central aid system, was clearly Type 2. There are, of course, cases which also indicate the control ideology of the 'egalitarian autonomous' type, e.g., in the field of cultural services, but these cases are indeed quite rare at the moment.

7. See, e.g., Ilkka Heiskanen, op.cit.; Markku Pekonen, 'Central Finance and Central Planning as Alternative Means of Central Guidance of Local Self-Governing Units in Finland', Research Reports of the Institute of Political Science, University of Helsinki, DETA 16, 1975; Krister Ståhlberg, 'Teori och praxis i kommunal planering', Meddelanden från Stiftelsen for Åbo Akademi Forskningsinstitut, No. 4 (Åbo: Åbo Akademi, 1975).

8. Pentti Ronkko, 'Kuntien valtionaputarpeen mittaaminen' (Measurement of Need for Central Aid to the Municipalities), Research Reports of the Institute of Local Government, University of Tampere, 1/1972.

9. For the validation, see Raino Aaltonen (educational services); Hannu Ekholm (health services); Ritva Mitchell (cultural services); and Risto Yrjonen (social welfare services). Publications on these topics are in preparation for the DETA-series.

10. The development of the Finnish public sector from 1809 to 1974 is found in Tuomo Martikainen, 'Julkisen sektorin kasvu Suomessa' (The Growth of the Public Sector in Finland), Research Reports of the Institute of Political Science, University of Helsinki, DETA 13, 1975.

11. Ibid., pp. 114-115. 12. Ibid., pp. 108-135.

---

13. Ibid., pp. 117-119, 39-40.

14. C. A. Laszlo, M. D. Levine, and J. H. Milsum, 'A General Systems Framework for Social Systems', Behavioral Science, Vol. 19, No. 2 (1974), p. 84; see also Martikainen and Yrjonen, op.cit., pp. 7-13.

15. See Laszlo, Levine and Milsum, op.cit., p. 85.

16. The data is reported in full in Martikainen and Yrjonen, op.cit., pp. 85-155. The interpretation of the data, however, is made here from a different perspective. There are no direct references made to the original source.

17. For the application of systems analytical concepts hi political science, see Dag Anckar, 'David Eastons politiska teori: en intern och extern modellkritik', Acta Academiae Aboensis, Series A, Vol. 50, No. 2 (Åbo: Åbo Akademi, 1974); for the concept of withinput', see especially pp. 47-52, op.cit., and Krister Ståhlberg, 'Några anteckningar kring David Eastons syn på det politiska systemet', Statsvetenskaplig tidskrift, No. 4, 1974, pp. 244-254.

18. For further elaboration, see Ilkka Heiskanen, 'Strategisten rakenteiden analyysi poliittishallinnollisten kokonaisjarjestelmien vaikutusten erittelyssa ja evaluoinnissa' (On the Analysis and Evaluation of the Political and Administrative Systems via Partial Analysis of Their Strategic Structures), forthcoming in the DETA-series.

19. See, e.g., Aulis Poyhonen, 'Kuntien kantokykyluokitusjarjestelma ja kunnan mahdollisuus vaikuttaa kunnan kantokykyluokan maaraytymiseen', Suomen Kunnallislehti, 7 (1974), pp. 43-47.

20. Laszlo, Levine and Milsum, op.cit., p. 81.