Arbejdspapir 5: Forskelle i opfattelsen af kommunale nedskæringer og velfærdskoalition (perceived municipal cuts and the welfare coalition) - fuld version

Table of contents×




The paper analyzes a poll of the perceptions of changes in provisions of municipal public service from 2007 to 08 following a large municipal reform. The poll show a considerable negative fiscal illusion: People perceive service cuts almost independently of the actual change in expenditures. Four explanations are found for the perceptions of cuts: They increases with the number of old municipalities merged by the reform. Moreover, stakeholder interest, left ideology and user interests influence the perception. The stakeholder interests are measured as “membership” of the welfare coalition, which is the fraction of the municipal popu­la­tion that receives their income from the public sector as wages or social benefits.

The present paper is an adjusted version of a paper presented at the 2009 annual meeting in the European Public Choice Society in Athens on april 4, 2009. I thank the discussant, Benny Geys, WZB Berlin, and Martin Paldam, University of Aarhus, for useful comments.  

1.   Introduction: Polled perception of budget cuts

The paper analyzes a poll from late 2008 asking a representative sample of 1104 Danes how they per­cei­ved the development from 2007 to 2008 in public service provisions. The poll covers the end of the first normal year after a large municipal reform where 275 munici­pa­li­ties were merged into 98 new ones.The reform law was approved by the parliament in June 2005, and implemented during the next 1½ years by the MoI, the Ministry of Inte­rior – a few municipalities merges even before the law was approved. The new municipalities started operating the 1st January 2007. The basis for the reform is discussed in a multi-volume white book (in Danish) from the MoI. A short English summary is available (Indenrigs- og Sundhedsministeriet 2005). The poll was made to cover the first normal year after the reform.  The paper analyzes the perceptions at two levels:

First, they are compared with the actual expenditures of the munici­pa­lities. The paper finds a considerable, but negative, fiscal illusion – that may be a measure of protest. People tend to perceive that budgets are cut, almost inde­pen­dently of the actual movements of the expenditures. Second, an attempt is made to model the perception. It depends upon the local size of the reform, and of the size of the welfare coalition in the municipalities. That is, of the share of people who receives their income from the public sector, irrespective of whether the income comes from the munici­pa­lity or other public sources.

1.1   Some institutional background: The fiscal federalism game

Danish municipalities are responsible for more than half of public expenditures, though they are partly administrating laws issued by the Parlia­ment, so they can decide over less than half of their budgets.It is difficult to assess the precise amount of decision leeway in the Danish municipal sector as there are natio­nal minimum standards even in the fields that are fully within the municipal sector, such as the primary school system. However, in most international comparisons, Danish municipalities are relatively large and powerful. There is also a regional level between the Municipalities and the State being responsible for the hospital sector, but it is weak compared to the municipal level.  Also, municipalities have some power to tax, so that income taxes can differ by 2-3% (of total incomes) between munici­pa­lities.

Thus municipalities carry considerable responsibilities. The municipal reform was driven by the concern at the central level that many municipalities were too small to carry these responsibilities. The reform was thus a typical top-down reform.

The Danish fiscal federalism game is played, with considerable dexterity, between two parts: On the side are the State represented by two ministries, MoI (the Ministry of Interior) and MoF (the Ministry of Finance), and on the other side is the LGDK, Local Government Denmark (KL in Danish). Fiscal policy is applied to the municipal sector through the frames of a joint budget constraint that is nego­tia­ted annually between the parts. Then the LGDK distributes the frames among its mem­bers by a second round of negotiations. This always causes expenditures-sliding above the frames, but efforts are made to prevent this from being known till the expenditures have been made a year later. By then a game of “respon­sibility shifting” is played.

Part of the fiscal federalism game is the propaganda “war” to rally public support: LGDK and many individual municipalities insist that the State forces them to cut, and that they despe­ra­tely need funds from the general revenue. The State keep stressing that the munici­pa­lities have more money than ever before and that they do not keep the agreements, they have signed themselves. People need a genuine interest in hard data analysis and some work to see through this screen of propa­ganda.

1.2    Changing municipal roles and the reform    

In connection with the reform a large discussion has taken place about the changing role of local government. Was it actually changing, and if so, were the changes in a desirable direction, and how this should influence the reform. This also led to a great deal of academic literature (but little is available in English), see however the volume by Rattsø (1999).

                 In the introduction to the volume Rattsø (1999) argues that the role and rationale of the Scandinavian muni­ci­palities are changing from a historical role as providers of local public goods to new role as pro­du­cers and suppliers of “private goods which shall work for redistribution.” They still act as inde­pen­dent local authorities with local demo­cracy, but in addition they are agents of the State supply­ing private goods with a distribution externality. The changing roles must also change the fiscal federalism game: At least two legislative initiatives are steps in that direction:

First: To prevent expenditure-sliding in the municipal sector the Parliament passed in 2008 a law defining a rule for automatic reversing to the state of excess revenue from municipalities exceeding the frames. Second: The annually frame- agree­ments between the State and the LGDK have become more detailed regarding goal definition and standards of municipal measures. The municipal sector has reacted strongly against this formal clarification and emphasis on its agent role. The Left opposition in the Parliament has been especially critical towards the reforms preventing expen­di­ture-sliding. So the shift toward the role-agent has been controversial.

               Section 2 analyzes the level of fiscal illusion by comparing actual and perceived expenditure movements. Section 3 lists the theories and variables used for the modeling. The short Section 4 looks at the quality variable that should have worked but did not. Section 5 brings the regressions and finally Section 6 contains a brief summary and a discussion of the results.

2.  Actual versus perceived budget movements

The real term growth in municipal expenditure level from 2007 to 2008 differs among the municipalities. Figure 1 describes the distribution of growth rates. Change in service level is measured by compensating for changed levels of prices and wages and for changes in needs caused by changes in demographic and socioeconomic structures, and finally compensating for changes in municipal tasks imposed on the municipalities by the national parliament and government. No productivity effects are taken into consideration despite the intensified use of capital and human capital as production factors in the municipalities and despite the new possibilities of making use of economies of scale after the 2007 mergers. As a consequence our measure may underestimate the actual development in service level.

Figure 1. Change in real municipal service level from 2007 to 2008

Note: One percent categories of the population ranked after change in service level in their home municipality

No more than 36.1 pct. of the population has in reality been subject to a reduction in municipal service level from 2007 to 2008. Nevertheless an obvious situation of unclear understanding among citizens in the municipalities concerning the actual change in the welfare level of service provision in their home municipality has appeared due to conflicting information and strong partial political and institutional interests. This paper studies whether this is actually the case. Furthermore, the paper searches for a pattern of explanations in relation to why some citizens may believe that the level of welfare service provision has been reduced in their home municipality despite actual increase.

A large part of the population are actually under the impression that municipal service activities have been reduced because of budget cuts as shown in table 1. A survey executed in October 2008 finds that 49.2 pct. of the adult population is convinced that their home municipality has reduced the service level from 2007 to 2008. A large part of the population, 33.7 pct., lives in municipalities that in fact raised the service level. However no more than 4.0 pct. of the population realizes this increase in service level.

Table 1. Comparing the actual and the perceived change in service provision

Because it is not possible to explain the perceptions of municipal budget cuts by actual budget cuts in a satisfactory manner, it is necessary to search for other explanations.

Several other characteristics of the municipalities beside municipal economy may work as explanations. Additionally characteristics attached to citizens may function as explanations. At first the deviation between actual and conceived development could be considered a simple information problem. However the strong skewness of the response distribution in Table 1 does not support such an interpretation. Hence, explanations involving stakeholder interests seem should be included.

3.           Variables that may explaining fiscal illusion

The model estimated in Section 5 try to explain people’s perceptions of municipal service cuts: It uses four municipal characteristics and three hypotheses pointing to citizen characteristics. Some of the hypotheses have been ope­ra­tionalized by more than one variable to catch various aspects of the same basic idea. The variables are chosen to create a net catching each effect, and the variables within each hypothesis are therefore correlated as will be discussed. The full set of variables is shown in table 2.

Table 2. A survey of the 13 variables used to analyze the perceived service cuts

3.1          Four variables characterizing the municipality: MRS, MSiz, MPF, MRef

MRS represents the full information non-strategic setup. Citizens believe that the home municipality has reduced the service level if the municipality actually has reduced the service level. We expect a negative coefficient expressing that larger negative values of service level increase will raise the tendency to perceive cut in service level.

MSiz represents the idea that large municipalities are more making it more difficult for the individual citizen to see through the bureaucracy and to know all local activities. Such impenetrable municipalities could possibly cause citizens to misinterpret actual changes. Thus, if large municipalities alienate citizens we expect a negative coefficient.

MPF measure the political fragmentation in the municipalities. The hypothesis states that a high degree of fragmentation is related to the perception of service-level cuts, because increased fragmentation implies strong political competition and a strong political opposition arguing that the majority coalition cut down the service level offered to citizens.

MRef represents the municipal structural reform. It measures the number of former municipalities that have been merged into the present one. The number range from 1 to 7. As a consequence of the merger the service level in the former municipalities was homo­ge­nized. If the new level became the average level, this creates losers and gainers. However, if there was economics of scale, and due to the rise in the average level in 2007, to make the reform palatable, the number of losers was smaller than the winners. If the variable gets a negative coefficient as we suspected this points to a status quo bias.

We now turn to 9 variables characterizing the citizens.

3.2. Citizens (1). The ideology and stakeholder variables: Pol1, Pol2, WC1, WC2

The Pol-variables express opposition to prevailing politics at the national and municipal level.

Pol1 measures the opposition to the Center-Right government, i.e., the respon­dent supports the Center-Left parties, from the Social Democratic party and further to the left.[1] The key politi­cal propaganda item of the opposition is precisely the Center-Right government undermines the Welfare State by incessant cuts in the welfare provisions. If the supporters of the opposition believes that, or vice versa, we should find a positive coefficient to Pol1.

Pol2 is the corresponding variable for opposition to the municipal leader, the mayor. Those opposing the mayor probably believe that the mayor perform poorly in his/her job, and as the question actually deals with the municipal provisions of goods we expect.

 The two Welfare Coalition variables, WC, model the idea (from Christoffersen and Paldam 2003)[2] that people receiving their income from the public sector form an implicit coalition for the expansion of this sector. Their stakeholder interests conse­quently cause them to be against cuts the public sphere, and hence to be suspicious of anything that may lead to such cuts. The WC variable has two parts:

WC1 is the fraction of the inhabitants that is employed in the public sector, of which about half work in the municipality where they are polled.

WC2 is the fraction of the inhabitants that receives (income compensating) public transfers. These transfers are by national rates, and thus independent of the municipal level of service provision.

3.3          Citizens (2). User interests: Ufe, UAE, Uag1-4, UInc1-8

Our data does not include detailed information about the specific use of welfare services among the individual respondents. However, it contains the usual variables characterizing the background of the respondents. Therefore formulate four hypotheses dealing with citizen characteristics which we believe can be linked to relatively strong interest in a high standard of welfare service provision in the home municipality:  

UFe covers the idea that women generally may focus more on public service than men, as the service provided often replace the serviced they provided in the traditional division of labor. Although most of that division is now gone it may still have left an asymmetric interest. Also, it appears that (Danish) women have a much smaller interest than men in the economy and hence may be more susceptible to fiscal illusion.[3]

UAE measures the fraction of citizens with an education at university level. It should be easier for such citizens to acquire (process) information, and they do have mure knowledge about the economy.  

The 4 UAg variables cover 4 age groups, which also represent groups with different user patterns in relation to the municipal service provision. Young families are in direct contact with the municipal service delivery system during day care and schools and old people use care services in own home or in residential homes for the elderly whereas middle aged people generally have weaker user contact with the municipal service delivery system. Therefore we expect to see a U-curve correlation between age and belief in budget cuts.

The 8 UInc variables cover 8 income groups, that describe the respondent’s budget restriction and consumer power. The hypothesis which will be tested is that the citizens become more critical in relation to the municipal budget situation with higher income.

3.3. Citizens (3). Service illusion: SLev

SLev tests the hypothesis that the level of expenditures expressed as total expenditures compared to total needs influences the citizen understanding of development of the welfare level in the municipal service provision. Our idea is that citizens in municipalities with high expenditures compared to needs feels a larger risk of losing welfare than citizens in municipalities characterized by a lower level of expenditures and that this will lead citizens in high expenditure municipalities to be more inclined to feel budget cuts.

4.           The service dimension that failed: Perceived service quality and actual expenditures

It is sometimes argued that the reason people perceive a fall in the public service provision is because the quality falls. The poll does not fully cover the quality dimension, which is diffi­cult to measure anyhow. However our poll dealt with service quality and asked the respon­­dents if they were satisfied with the quality of the service provide on a 5 point scale.

Table 3. The relation between perceived quality and actual expenditures

Table 3 shows the relation between the perceived quality delivered and the level of actual expenditures. As can be seen from the table no significant correlation appears and furthermore the sign is negative. This result does not support a hypothesis that changes in quality act as explanation of citizens experiencing changes in the welfare level in the municipal service provision.

5.           The regressions

The model is a simple linear model explaining perceived budget cuts by all variables from Table 2 from MRS to UInc. The model is estimated in two ways: As a probit model with 2 alternatives and as an ordered probit with 3 alternatives, see Table 4.  First all variables are included in estimates (1a) and (2a).  Then the stepwise Hendry-Mizon procedure of deletion of the most insignificant variables is pursued till only significant variable remains in estimates (1b) and (2b). In both regression formulas only 5-6 variables remains and it is almost the same. These regressions are shown in table 4.

A hypothesis is accepted as fully confirmed if we find significant parameter values in both of our model formulations – this applies to MRef, Pol1, WC1 and UAg3. A hypothesis where parameter values are only found to be significant in one of the two model formulations are seen as partially confirmed – this apples to MPF, UFe and UAg2 – whereas other hypotheses without significant parameter values are rejected. Table 5 examines the pattern of multi­colli­ne­arity.

5.1          Interpreting the main results: Table 4           

The first four variables are the municipal characteristics:

The result for MRS is striking and confirms the suspicion that followed from Table 1. It is not possible to confirm the hypothesis that the perceived service level reduction is caused by actual service level reduction in the municipality.

It is interesting to note that the size of the municipality MPop has no effect. The fragmentation variable MPF has the right sign of the one hypothesized. The more parties the more people perceive cuts. The result is strongest in the ordered probit estimation.

One of the key variable studied is the reform variable MRef. It is significant through­out and always positive, the bigger the reform is in the municipality the more the citizens perceive that the service is cut. Since the size of the municipality does not matter to people this result must be interpreted as a support for the status quo. Also, it is perhaps a sign that people disliked the top-down character of the reform.

               Turning from the municipal characteristics to the citizen characteristics the first four variables are the political ones: The first observation is the paradox that national politics matter, but not municipal. This further reinforces the impression that the perceived cuts are a protest variable – building on fiscal illusion – with little relation to facts.

               The two variable of the welfare coalition only generate one significant variable WC1, but as we shall see the other suffers from strong colinearity with Pol1. The welfare coalition is thus strong in the data. People who depends upon the public sector perceive that it is cut.

Table 4. Four regressions: For all variables and significant variables

Note: Bolded estimates are significant. a. Age reference group: UAg1, 18-30 years. b. Income reference group: No answer.  

               The UFe variable is one of the dubious ones, but it works as hypothesized.

               There is no effect of higher education.

               The 3 age variables UAg2-4 are measured relative to the group of young UAg1, and they produce considerable effect, notably the middle group. This result is contradictory to our hypothesis. An interpretation is that the middle aged citizens generally are relatively strongly engaged in politics and that this overshadows their relatively weak contact with the local welfare system as users of local public services.

               Finally the eight UInc1-8 variables measuring the income of the respondents, fail to show a clear pattern.         

               The same is the case with SLev, illusion linked to the absolute service level.

5.2          Multicollinarity and explanatory power: Table 5

Table 5 use the probit model (1) for a set of experiment to reveal the multicollinarity of the variables. However, before these results are discussed the reader should note the last two rows of the table that show the goodness of fit for the model in form of its ability to predict. Crudely, put it predict 60% right and 40% wrong. So it is 50% more right than wrong.

Table 5. Prediction when combining the explanations using the probit model

As we can see in the table the explanatory power of the model does not change significantly when changing combinations of the variables WC1 and Pol1. The percentage of right predictions is almost equal when using different combinations of these variables. Our seemingly very different variables must hence contain strong multicollinearity.

The UFe variable, sex, is not coming up with a significant contribution in the probit model to explain the belief that the service level has been reduced. On the other hand the EFe variable works well in the ordered probit model. The prediction technique is unfortunately not suitable for use in connection with an ordered probit approach. A simple correlation matrix calculated on the ordered probit model shows a close correlation between UFe and WC1 (public employed) as well as a close correlation between UFe and Pol1 (ideology). These results underline the existence of strong multicollinearity between the variables for welfare coalition - political affiliations and gender.

6.           Discussion

In itself it is remarkable that only 4 pct. of the citizens in Danish municipalities assume that their home municipality has expanded the service level from 2007 to 2008 in spite of the fact that one third of the citizens actually should experience an increased real expenditure level if aligned with reality. It is especially thought-provoking that the understanding of development in service level found among the citizens is not at all correlated with actual development in service level. Misinformation seems to be striking and to be contributing to a pressure on the municipal economy. These results contain at least two important challenges.

The political and administrative management in Danish municipalities is challenged because they have not succeeded in giving their citizens a true understanding of how the local service level is developing. Further municipal information efforts are crucial. Without satisfactory information it seems as political fragmentation in local politics lead to improbable strong attention on budget cutting rather than on enhancing the popular involvement in a constructive way. Lack of information may also explain why it is difficult for citizens in newly amalgamated municipalities to realize the actual development of the performance in their municipality. The amalgamation processes taking place since the municipal structural reform of 2007 do not seem to be finished in the municipalities. This fact might also underline the importance of devote more effort to integrating different ways of designing and organizing service provision when following up after the amalgamation itself.

The idea of local democracy and the democratic institutions as such are challenged by the welfare coalition being so strongly guided by stakeholder interests. If political decisions are taken contrary to the interests of the welfare coalition political costs seem to be considerably high. As shown in Christoffersen and Paldam (2003) the result can be that the welfare coalition can block municipal efforts of raising cost efficiency in the municipal service provision. Optimality claims a separation of stakeholder interests and citizen/user interests. Unfortunately no simple solution exists in order to realize such a separation. 


Christoffersen, H., Paldam, M.,  2003. Markets and municipalities: A study of the behavior of the Danish munici­pa­lities. Public Choice 114, 79-102

Indenrigs- og Sundhedsministeriet (Ministry of the Interior and Health), 2005. The local government reform – In brief.

Nannestad, P., Paldam., M., 2000. What do the voters know about the economy? A study of Danish data, 1990-1993. Electoral Studies 19, 363-92

Rattsø, J., ed. Fiscal federalism and State-Local Finance. Edward Elgar, Cheltenham

Rattsø, J., Sørensen, R., 1998. Local governments integrated in a welfare state: a review of Norwegian local government performance. Available in: Rattsø (Ed.): Fiscal Federalism and State-Local Finance. Edward Elgar.


    Hent analyse

    Hent Filstørrelse: 462,8 KB

    Del denne side


    Henrik Christoffersen

    Tidligere forskningschef

    +45 41 22 04 76

    Hent analyse

    Hent Filstørrelse: 462,8 KB

    Del denne side


    Henrik Christoffersen

    Tidligere forskningschef

    +45 41 22 04 76