Fiscal effects of voluntary municipal mergers in Switzerland

Page created by Jeffery Bell
 
CONTINUE READING
Fiscal effects of voluntary municipal mergers in Switzerland

                             Working Paper - Do not cite or circulate without permission

                                                  Janine Studerus∗

                                                 October 23, 2016

        This paper explores the fiscal effects of voluntary municipal mergers in Switzerland, analysing 160 mergers
that occurred between 2001 and 2014. Based on a propensity score matching approach, groups of merged
municipalities are compared to groups of non-merged municipalities that are similar in terms of pre-merger
fiscal and socio-economic characteristics. The results do not provide substantial evidence for positive scale
economies. While a decrease in administration expenditures can be observed, the effect does not carry over to
total expenditure levels. At the same time, no evidence is found for improved quality of local public services.
Finally, the analysis indicates that the financial grants paid out by the cantons in order to induce municipalities
to merge might have been used to finance tax cuts, thereby ensuring the populations’ support for the merger
projects.

1        Introduction

Local government reforms have been on political agendas of several countries since the 1960s. Espe-
cially Northern European countries such as Denmark, Sweden or the United Kingdom have introduced
top-down reforms that led to substantial reductions in the number of municipalities. These reforms
were often driven by the expectation that larger local governments could provide local public services
more efficiently by exploiting size economies or internalizing spillover effects (Fox and Gurley, 2006).
        The existing evidence on fiscal effects from municipal mergers in various countries is however mixed.
Cross-sectional studies analysing the correlation of population size and municipal expenditure levels
in the United States do not show a clear picture (for example Bodkin and Conklin, 1971; Holcombe
and Williams, 2009). Studies directly analysing structural reforms in Germany, Israel and Sweden
point to effects of mergers on expenditures, debt levels and population growth. However, the study
for Germany only focuses on monocentric city regions and the reforms considered in the other studies
were top-down in the sense that some upper-level government decides on which municipalities are to be
merging (Blume and Blume, 2007; Reingewertz, 2012; Hanes and Wikström, 2008). Voluntary reforms
have for example been analysed by Blesse and Baskaran (2013) and Moisio and Uusitalo (2013) who
    ∗
    PhD student at the Institute of Public Finance and Fiscal Law of the University of St.Gallen (ja-
nine.studerus@unisg.ch). Thesis supervisor: Prof. Dr. Christoph A. Schaltegger, University of Lucerne and Institute of
Public Finance and Fiscal Law of the University of St.Gallen (christoph.schaltegger@unilu.ch).

                                                          1
found contrasting results. One reason for the variety of results might be that the effects of municipal
mergers depend on the specific institutional setting, which may differ considerably across countries.
Furthermore, diseconomies of size such as higher management and controlling costs or increased in-
formation asymmetries could lead to a less efficient local public services provision (Lüchinger and
Stutzer, 2002). Accordingly, the net effect of municipal mergers on efficiency is not a priori clear.
   Municipal mergers are currently also an important topic in Switzerland. The country is partitioned
into more than two thousand Swiss municipalities with an average population size of only around
3’500 inhabitants. Thereby half of all municipalities have less than 1’400 inhabitants. Accordingly,
the concern is often raised that the current municipal boundaries do no longer allow for an efficient
provision of public goods and services (Rühli, 2012; Dafflon, 2012). Between 2001 and 2014 more than
200 mergers have taken place, leading to a reduction in the number of municipalities by more than
500.
   So far, there is only scarce evidence on the fiscal effects of Swiss municipal mergers. Lüchinger
and Stutzer (2002) analyze the effect of four municipal mergers in the canton of Solothurn on central
administration expenditures. They do not find evidence for increased efficiency. Kuster and Liniger
(2007) analyse potential effects of seven mergers in different cantons in a case study approach. They
identify savings potential regarding staff and infrastructure. At the same time municipalities also aim
at improvements of public services, which might lead to increased per capita expenditures. Ladner
et al. (2013) ask the staff of merged municipalities in a survey to which extent they agree with different
potential effects of municipal mergers. The average agreement of the around 80 answers is highest
regarding a more professionalized administration and adjusted administrative structures. Furthermore,
public service quality is stated to have improved and the range of services supplied increased. These
effects would potentially imply raised expenditures in total. At the same time some mild agreement
consists regarding an improved financial situation and lowered indebtedness.
   The aim of this paper is to provide a systematic analysis of fiscal effects of Swiss municipal mergers.
In a first step, I want to contribute to the existing literature on potential efficiency gains by analysing
expenditure effects. Thereby the existing body of literature is extended with another specific insti-
tutional setting whereby small municipalities with substantial expenditure and tax autonomy merge
voluntarily. Assuming that per-capita costs for local public services fall with increasing population
size up to a certain threshold and then start rising again, one would expect that the efficiency gains
of mergers would be largest for small municipalities. Furthermore, since these municipalities are given
substantial autonomy to set their own tax multipliers and provide local public goods, they can be
expected to have the means to exploit potential efficiency gains. As a second contribution I explicitly
address the question whether efficiency gains may have been realized not through lower per capita
costs, but through improved quality or a higher quantity of local public services at constant per capita
cost. Assuming that major quality improvements increase a municipality’s attractiveness, the effects
of mergers on the development of population growth, immigration rates as well as rental and housing
prices are analysed.
   Somewhat more specific to the Swiss context, I also analyse the effects of mergers on municipal
tax multipliers and debt levels in order to test whether the expectations raised by cantons and mu-
nicipalities regarding improved financial situations are justified and can be sustained over the longer
run.
   For this analysis an extensive dataset of Swiss municipalities from the ten cantons Aargau, Berne,

                                                    2
Fribourg, Graubünden, Lucerne, St.Gallen, Solothurn, Ticino, Vaud and Valais, covering 160 mergers
has been constructed. The individual merging municipalities are matched with similar municipalities
from the same canton by nearest neighbour matching on the propensity score. The matched control
municipalities exhibit similar characteristics in terms of population size, income levels, per capita
expenditures, tax multipliers and debt levels prior to treatment, but have not been subject to an
actual merger. I then aggregate the fiscal variables of these actual and artificial mergers and analyse
differences before and after the respective mergers take place in a difference-in-differences analysis. The
results do not provide evidence for substantial economies of size. I do find a decrease in administration
expenditures after the mergers take place of around 40 Swiss francs per capita, which amounts to less
than one percent of average total per capita expenditures. This effect on administration expenditures
does however not translate to a statistically significant effect on total expenditures. At the same time,
I do not find any statistically or economically significant effects on the indicators for changes in the
quality of local public goods. The results further indicate that cantonal merger grants might have
been used to lower tax multipliers in order to ensure the populations’ agreement with the respective
merger projects.
        The remainder of this paper is structured as follows: The next section summarizes the relevant
literature, followed by a brief introduction to the institutional setting. In section 4 the data used and
some descriptive statistics on the pre-merger municipalities are presented. Section 5 describes the
empirical strategy and the matching procedure applied. I discuss my results in section 6 and conclude
in section 7.

2        Literature review

The traditional theory of fiscal federalism provides basic principles regarding how public tasks should
be allocated to different layers of government. According to the principle of subsidiarity public services
should generally be provided by the lowest level of government. This allows the government’s actions to
be aligned in the best possible way with inhabitants’ preferences (Oates, 1972). Reasons for deviating
from this principle are economies of scale and externalities in benefits or costs (Oates, 1999). In order
to avoid inefficiencies due to externalities Olson (1969) introduces the principle of fiscal equivalence.
It states that the reach of a collective good and the boundaries of the administrative entity providing
it should be aligned. If the good is a purely public good in the sense that non-purchasers cannot be
excluded from consumption, increasing economies of scale are not an issue once fiscal equivalence is in
place1 . However, if the goods provided by some entity are not fully public so that non-purchasers can
actually be excluded from consumption, it may be sensible to let the boundaries of the jurisdiction be
determined based on the respective production costs (Olson, 1969)2 .
        In line with these traditional principles the economic argument often mentioned first in debates
on municipal mergers is the one of scale economies. Thereby it is argued that average production
costs could be lowered by mergers because of the joint use of fixed assets, increased productivity
through specialization or discounts on the purchase of input goods due to larger quantities (Dollery
    1
     The argument of Olson is that even though average costs would decrease further if more users were included this
cannot be achieved since all agents getting a positive utility out of the good are already in the jurisdiction.
   2
     In the extreme, Olson’s theory calls for separate administrative entities for each type of public good. The size of
these separate entities would then be determined solely by the reach of the good provided. However, Olson acknowledges
that in case of economies of scope it might be justified to allow for some discrepancies concerning the boundaries of the
administrative entities in order to exploit such cost savings potential.

                                                           3
and Fleming, 2006)3 . Empirical studies regarding this matter however, do not show a clear picture.
Ostrom (1976) criticised forty years ago already that proponents of metropolitan government reforms
in the United States accept the assumption of economies of scale to be true without further evidence.
Lüchinger and Stutzer (2002) summarize various sources for potential dis-economies of scale in the
context of municipal mergers. First of all, an increased size of a municipality might bring along
higher costs for management, controlling and a more professionalized administration. Additionally it
might be more difficult for inhabitants to monitor and control political actors in larger municipalities.
Self-interested politicians might use this increased information asymmetry to follow their own private
goals and thereby induce larger municipal budgets. In order to clarify whether increasing or decreasing
scale economies matter more in practice, various studies with different methods and focuses have been
conducted.
       A first set of studies analyse the general relationship between public expenditures and the size of
municipalities for the United States. Bodkin and Conklin (1971) find that the correlation between
population size and municipal expenditures varies for different types of public goods. Larger mu-
nicipalities are for example associated with lower per capita expenditure levels for water supply and
public works but higher average expenditure levels regarding fire and police protection, waste removal
or general government services. The authors conclude that mergers should not be justified solely by
potential scale economies. Brueckner (1981) focusses on fire protection as a particular type of pub-
lic good. Thereby he approximates fire protection consumption by the respective community’s fire
insurance rating. Brueckner analyses 100 municipalities in the United States with more than 30’000
inhabitants. His results suggest that communities with more inhabitants are able to provide a given
fire protection level at lower per capita costs, thereby contrasting previous findings. In a more recent
study, Holcombe and Williams (2009) analyse expenditures from almost five hundred municipalities
in the United States with more than 50’000 inhabitants. They find that population density and not
population size is a major determinant of public expenditures per capita.
       More directly related to municipal mergers is another set of studies that compares groups of merged
entities with some control group after the reform has taken place. Blume and Blume (2007) compare
the financial situation of German unified monocentric city regions to regions in which the core city and
the hinterland are different administrative units. The main goal is to determine whether the effects of
mergers differ from the effects of inter-municipal cooperation. Furthermore, they compare debt levels
as well as economic growth of regions that were affected to varying extents by territorial reforms in
the 1960s and 1970s. In both cases the authors find statistically significant differences, which lets
them conclude that at least in monocentric city regions, mergers seem to be more effective than inter-
municipal cooperation. Hanes and Wikström (2008) analyse the effects of a top-down municipal reform
in 1952 in Sweden on population and income growth. One side-question is whether size and income
heterogeneity of the merging municipalities matter for the respective outcomes. The authors find that
mergers did not affect income growth rates but seem to lower the out-migration rate in the smallest
group of municipalities. Furthermore, heterogeneity in pre-merger population size and income levels
matters somewhat for the impact on population growth. Population size heterogeneity has a positive,
income heterogeneity a negative effect. In a second study Hanes and Wikström (2010) use the same
   3
    In the context of municipal mergers one does not usually refer to economies of scale in a narrow sense, since inputs
may not necessarily change proportionately as two or more municipalities consolidate. Fox and Gurley (2006) refer to the
question whether larger administrative entities are able to provide public services at lower costs than smaller ones as a
question of economies of size. However, in the literature on municipal mergers these terms are often used interchangeably.

                                                            4
setting to investigate the relevance of compulsory versus voluntary mergers for income and population
growth. The results indicate that for small municipalities the positive effect of mergers on population
growth is larger for voluntary mergers.
    A third set of studies is based on observing the data before and after the mergers take place.
Based on a difference-in-differences approach, Lüchinger and Stutzer (2002) compare four voluntarily
merged municipalities with a control group consisting of municipalities that are similar in various
respects in the Swiss canton of Solothurn. They focus on central administration expenditures because
for this type of task there is typically no inter-municipal cooperation which might already exploit
potential scale economies. The results do not indicate any effects on efficiency. In a more recent study
Reingewertz (2012) analysed the fiscal effects of a top-down municipal merger reform in Israel in 2003.
Thereby 23 municipalities were involved in 11 mergers. Also applying the difference-in-differences
methodology the author finds a statistically significant expenditure decrease of about nine percent.
From this result Reingewertz concludes that potential economies of scale have actually been exploited
in these merged municipalities. A similar study was carried out by Blesse and Baskaran (2013), albeit
concerning a reform of much larger scale in the German state of Brandenburg between 2000 and 2003.
Thereby the number of municipalities was reduced from almost 1’500 to 421. Since the reform entailed
both voluntary and compulsory mergers the authors also analyse whether the effects differ between
these two merger types. They find a decrease in expenditure levels of merged municipalities of about
70 Euros per inhabitant. Thereby the effects are weaker for voluntary mergers than for compulsory
mergers. Finally, Moisio and Uusitalo (2013) analyse 57 voluntary mergers that took place between
1970 and 1981 in Finland. Based on a matching approach the authors compare the pairs of merging
municipalities to similar pairs of non-merging municipalities before and after the merger takes place.
Thereby they find falling expenditure levels for general administration while overall expenditure levels
tend to increase in merged municipalities compared to non-merged municipalities.
    Based on the existing literature, the empirical question of whether municipal mergers improve or
worsen the financial situation of the respective administrative entities cannot be answered in general.
Economies of size might be present, but not necessarily fully exploited. Furthermore, the merger itself
might lead to opportunistic behaviour of the municipalities involved.

3    Institutional setting

Switzerland is a federal country consisting of three levels of government: the federation, 26 cantons and
2’294 municipalities as of 2016 (Swiss Federal Statistical Office, 2016). In international comparison
Swiss municipalities are small with on average around 3’500 inhabitants in 2014, ranging from below
twenty to more than 390’000. Thereby, half of all municipailities have less than 1’400 inhabitants
(Swiss Federal Statistical Office, 2015). Despite their relative smallness, Swiss municipalities generally
exhibit substantial fiscal autonomy and are responsible for a major share of public services and goods
provision. The actual assignment of tasks and competences to municipalities is however a matter of
cantonal legislation and may differ between cantons. Typical municipal tasks are basic education, social
welfare, public utilities, local police or regulation regarding buildings and streets (Steiner and Kaiser,
2013). During the last twenty years, municipal revenues and expenditures amounted to between one
quarter and one third of total public expenditures and revenues in Switzerland, excluding the social
security system (Swiss Federal Finance Administration, 2014).

                                                    5
The large majority of municipal mergers in Switzerland occur on a voluntary basis, whereby
cantons may exert more or less pressure and provide merger incentives4 . Accordingly, the extent to
which municipal mergers take place varies considerably among cantons and is correlated with differing
degrees of cantonal interference (Kaiser, 2014). Cantonal support to voluntary municipal mergers may
include a basic indicative planning5 , technical and administrative support during the merger process,
as well as direct financial grants to new municipalities. These grants can either be determined case
by case or according to a predetermined formula. The main goal is to provide direct incentives for
mergers and to reduce disparities among municipalities in order to prevent potential resistance of the
richer municipality against the merger (Dafflon, 2012).
        Figure 1 shows the municipality structure of 2014. The municipalities coloured in red are those
that resulted from municipal mergers occurring between 2001 and 2014. Table 1 provides an overview
on the number of municipalities and the average number of inhabitants of the municipalities by canton
and over time.
                      Figure 1: Merged municipalities (in red) between 2001 and 2014

                     Municipality structure at the beginning of 2014. Sources: Swiss Federal Office of Topography
                     (2015), Swiss Federal Statistical Office (2016)

4        Data and descriptive statistics

For this analysis municipality level data from ten cantons has been compiled for a rich set of variables6 .
In terms of population characteristics I collected data on population size, the shares of foreigners, in-
    4
     More recently some cantons introduced the possibility to force municipalities to participate in mergers by law.
Nevertheless, direct enforcements of municipal mergers so far remain the exception.
   5
     The idea is that cantons assess the situation of their municipalities and set some implicit target regarding the
municipality structure to be achieved in the future.
   6
     The analysis includes the ten cantons Aargau, Bern, Fribourg, Graubünden, Lucerne, Solothurn, St.Gallen, Ticino,
Valais and Vaud. The reason for the exclusion of some other cantons with merging municipalities is that some key
variables are not available in those cantons for all required years and for the relevant municpality structure. For the
canton of Valais fiscal data on the municipality level is available only after 2006. Therefore, I am not able to incorporate
mergers occurring before the year 2009 in the canton of Valais. Similarly, for the canton of Ticino I only incorporate
mergers from 2003 onwards, since some fiscal data is not available before 2000

                                                                  6
Table 1: Number of municipalities, mergers and average number of inhabitants

    Canton               2000                            2014              number of mergers   change in municipality numbers in %
             municipalites ∅ inhabitants    municipalities ∅ inhabitants      2001 - 2014                  2001 - 2014
    AG           232          2’346             213            3’029              14                          -8 %
    AI             6           2’504              6            2’642               -                            -
    AR            20           2’676             20            2’703               -                            -
    BE           400          2’359             362            2’788              27                         -10 %
    BL            86          3’024              86            3’271               -                            -
    BS             3          62’556             3            63’527               -                            -
    FR           242            977             163            1’861              43                         -33 %
    GE            45          9’085              45           10’609               -                            -
    GL            29          1’329               3           13’265               5                         -90 %
    GR           212            881             146            1’342              24                         -31 %
    JU            83            829              57            1’270              10                         -31 %
    LU           107           3’245             83            4’754              16                         -22 %

7
    NE            62          2’673              37            4’793              4                          -40 %
    NW            11          3’455             11             3’825               -                            -
    OW             7          4’631               7            5’262               -                            -
    SG            90          4’993              77            6’439              10                         -14 %
    SH            34          2’156              26            3’055              5                          -24 %
    SO           126          1’937             109            2’419              7                          -13 %
    SZ            30          4’341              30            5’092               -                            -
    TG            80           2’841             80            3’297               -                            -
    TI           245          1’266             135            2’595              31                         -45 %
    UR            20           1’762             20            1’800               -                            -
    VD           384          1’615             318            2’394              24                         -17 %
    VS           163          1’694             134            2’476              16                         -18 %
    ZG            11          9’035             11            10’917               -                            -
    ZH           171           7’086            170            8’508               1                          -1 %
    Total       2’899                          2’352                              237                        -19 %
Table 2: Number of mergers in sample

                     Canton          Number of municipalities involved        Time range of merger
                                 2     3   4 5 6 7 8 9 10+                        occurrence
                        AG      12      -   1    1   -   -    -   -     -           2002-2014
                        BE      17      3   -    -   -   -    1   -     -           2004-2014
                        FR      18     10   6    -   1   -    -   -     -           2001-2014
                        GR      10     3    1    2   1   -    -   -     1           2006-2013
                        LU       8      2   -    -   -   1    -   -     -           2005-2013
                        SG      5      3    -    -   -   -    -   -     -           2007-2013
                        SO       5      -   1    -   -   -    -   -     1           2006-2013
                        TI       3      4   -    3   1   1    -   2     -           2004-2014
                        VD      8      3    3    2   -   1    1   1     1           2002-2012
                        VS      8      3    -    -   1   -    -   -     -           2009-2014
                       Total    94     31   12   8   4   3    2   3     3           2001-2014

and outmigration, the share of young and old inhabitants and the average income of inhabitants. I
also compiled data on geographic characteristics such as area and altitude. Furthermore, I use the
municipality types as defined by the Swiss Federal Statistical Office (e.g. center, agglomeration, rural,
tourism) to account for major socio-economic differences. These variables are provided by the Swiss
Federal Statistical Office and they range from 1990 to 2014, although not all variables are available
for all years. In terms of fiscal variables, I rely on data from the respective cantonal statistical
or municipality offices. I collected data on tax multipliers, net debt and current account revenues
and expenditures. Current account expenditures and revenues are further divided into functional
subcategories7 . All data expressed in Swiss Francs is measured in terms of the price level prevailing
in December 2010. An overview on the variables used and their sources can be found in table 12 in
the appendix.
      All municipalities are divided into a treatment and a control group. In the treatment group
are all municipalities that merged exactly once between 2001 and 2014. Thereby, municipalities
that were involved into multiple mergers between 1995 and 2015 are excluded. In the control group
are those municipalities that never merged between 2001 and 2014. Furthermore, in order to have
a clear treatment and control comparison, municipalities that were already involved in a merger
between 1996 and 2000 are not considered as controls. These adjustments lead to a dataset of 1’997
municipalities, out of which 1’507 were never involved in any merger between 1996 and 2015. The
other 490 municipalities were involved in 160 mergers between 2001 and 2014. Thereof, 94 mergers
occurred among two, 31 mergers occurred among three, 12 mergers occurred among four and 23
mergers occurred among five or more municipalities (see table 2).
      Table 3 shows some descriptive statistics on the individual pre-merger municipalities. If not stated
otherwise the variables are measured three years prior to the first merger in the respective canton.
Merging municipalities appear to be fairly different from the average non-merging municipality, since
most mean differences are statistically significantly different from zero. Merging municipalities are
on average smaller both in terms of their population and area. Furthermore they tend to exhibit
lower shares of foreigners, higher shares of elderly and young inhabitants as well as lower income
levels. In terms of fiscal variables no statistically significant difference is observable in terms of per
capita expenditure levels, while tax revenues are somewhat lower for to-be-merged municipalities.
Furthermore merging municipalities seem to be more indebted and exhibit lower tax multipliers than
  7
      Functional subcategories are not available for the municipalities from the canton of Ticino.

                                                             8
Table 3: Descriptive statistics of the individual municipalities

                                                                     (1)              (2)             (3)
                                                                 Non-merging        Merging      Mean difference

                    Demographic Variables

                    Population                                     2280             1218                    1062***
                                                                  (5737)           (3709)                  (3.84)

                    Population growth in %                         0.61             0.41                    0.20
                                                                  (2.90)           (4.38)                  (1.13)

                    Foreigners in % (1999)                        11.71             8.70                    3.01***
                                                                  (9.13)           (7.80)                  (6.56)

                    Inhabitants above 80 in % (2000)               3.22             3.56                   -0.34***
                                                                  (1.77)           (2.52)                 (-3.29)

                    Inhabitants below 20 in % (2000)              25.12            25.63                   -0.51**
                                                                  (4.20)           (5.02)                 (-2.22)

                    Mean income in TCHF                           67.79            61.24                    6.55***
                                                                 (17.66)          (11.66)                  (7.68)

                    Fiscal Variables

                    Total expenditures per capita                  4855             5011                   -156
                                                                  (1899)           (2837)                 (-1.38)

                    Tax revenues per capita                        2702            2489                     213***
                                                                  (1012)           (891)                   (3.91)

                    Net debt per capita                            2043             2785                   -742*
                                                                  (6948)           (8462)                 (-1.94)

                    Tax multipliers                              144.42           132.59                   11.82***
                                                                 (66.69)          (54.89)                  (3.55)

                    Municipality types

                    Center mun. (small/medium/big) (2000)          0.02             0.02                   -0.00
                                                                  (0.15)           (0.15)                 (-0.34)

                    Rural mun. (2000)                              0.49             0.63                   -0.14***
                                                                  (0.50)           (0.48)                 (-5.33)

                    Suburban mun. (2000)                           0.40             0.29                    0.11***
                                                                  (0.49)           (0.45)                  (4.34)

                    Touristic mun. (2000)                          0.07             0.06                    0.01
                                                                  (0.25)           (0.23)                  (0.92)

                    High-income mun. (2000)                        0.02             0.00                    0.02***
                                                                  (0.15)           (0.06)                  (2.79)

                    Geographic variables

                    Area in hectares (1998)                        1499             1192                    306**
                                                                  (2513)           (2115)                  (2.43)

                    Inhabitants per hectare                        3.14             1.66                    1.48***
                                                                  (5.67)           (3.52)                  (5.45)

                    Observations                                   1507             490          1997
                    Column (1) and (2): Standard deviations in parentheses / Column (3): t-statistics in paren-
                    theses. Stars indicate significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01. If not
                    stated differently the variables are measured three years prior to the first merger in the re-
                    spective canton. All financial data is measured in Swiss Francs per inhabitant at the price
                    level prevailing in December 2010 based on the national consumer price index.

non-merging municipalities on average. In terms of municipality types merging municipalities are more
often rural and less often suburban municipalities, which is also mirrored in the difference between
the two groups regarding the population density variable.

5    Empirical Strategy

The key hypothesis to test in this paper is whether merged municipalities are more efficient than non-
merged municipalities in that they are able to provide their public goods and services at lower cost.
Assessing efficiency would require to measure both output and costs. The output of a municipality
is difficult to measure however. Accordingly, in a first step the focus lies on the effects of mergers on
municipalities’ expenditures. Increased per capita expenditures after a merger could thereby mean that
efficiency dropped or also that more or better output has been provided. On the other hand, lowered
per capita expenditures after a merger might point both to more efficiency or a drop in local public
services provision. As a second step I therefore try to account for this difficulty by approximating the

                                                                  9
output side by variables that may serve as indicators for a municipality’s attractiveness.

5.1      Baseline empirical strategy

When analysing the effects of municipal mergers on expenditures, a simple comparison of average
expenditure levels of merged and non-merged municipalities is very improbable to measure an actual
treatment effect but rather includes some selection bias. The reason is that since mergers are voluntary,
municipalities are more likely to merge if they anticipate particularly large gains from the treatment or
particularly bad outcomes without merging. Therefore, the outcome of an average merged municipality
group had it not merged is improbable to be the same as the outcome of an average municipality group
that actually did not merge.
      A second issue is that I do not observe the separate merging municipalities after the merger has
taken place. After the merger took place, I only observe the newly formed municipality. One way
to deal with this issue is to aggregate the data of the pre-merger municipalities and compare these
merged municipalities to non-merged municipalities. This way however, the characteristics of the
separate municipalities before the merger took place would to a large extent be disregarded, while
these characteristics might have been triggering the respective merger in the first place.
      In order to deal with these issues, I rely on propensity score matching8 . The main idea is that for
each group of merging municipalities a comparison group of municipalities is found that did not merge
but exhibits similar characteristics with respect to those variables that both influence the outcome of
interest and whether or not the merger takes place. The key assumption of this matching approach,
the so-called conditional independence assumption, states that conditional on all control variables, the
respective outcomes of merged and non-merged municipalities must not depend on the merger status
. This implies that I must be able to observe all variables that simultaneously influence expenditure
levels and whether or not municipalities merge. I argue that given the extensive dataset with a wide
range of potential control variables, including pre-treatment fiscal outcomes, I can lower selection
bias to a large extent. In order to avoid the dimensionality problem when matching on a range of
control variables, I do not directly condition on these variables, but instead estimate the respective
municipality group’s propensity to merge based on these variables, the so-called propensity score.
      Since I do not observe the data on the individual merging municipalities after the merger has taken
place, the data of non-merging municipalities is aggregated so that the fiscal outcomes of the actual
mergers can be compared to the artificial mergers before and after the mergers took place. In order to
mirror the group structure of the merging municipalities, the fiscal outcomes of each separate control
municipality is weighted by the corresponding treated municipality’s population share in the merging
group and then aggregated to the artificial merger group. Based on this adjusted sample of merged
municipalities and aggregated artificial control municipalities, I then analyse the effects of mergers on
fiscal outcomes using a difference-in-differences approach.

5.2      Selection into mergers

In order for the basic matching assumption to be valid, the propensity score estimation must include
all covariates that simultaneously affect the outcomes of interest and whether or not a municipality
  8
      An overview on matching methods can be found in Caliendo and Kopeinig (2008)

                                                        10
group decides to merge. Two main elements are to be considered: the cantonal merger support policies
and the pre-merger municipality characteristics.
    The cantonal merger support policies directly influence the number of mergers that occur in a given
canton (Kaiser, 2014). Furthermore, these policies may generate particularly high merger-incentives
for specific types of municipalities. Some cantons provide grants to equalize disparities in debt levels
and tax multipliers. Other cantons provide higher grants to mergers of fiscally weaker municipalities.
Furthermore, certain schemes involve lump sum grants per inhabitant up to a maximum of inhabitants,
causing the merger incentives to be higher for smaller municipalities than for larger ones. These
issues are accounted for by estimating separate propensity scores for each canton and by matching
exactly on the cantons. Furthermore, I include pre-merger income levels as an approximation to
the municipalities’ fiscal strength, as well as debt levels, tax multipliers and population size in the
propensity score estimation.
    Since each pre-merger municipality votes on the merger separately, municipality-specific character-
istics can be expected to be important for the merger decision. Financially weak municipalities might
initiate or agree to a merger because they see themselves unable to fulfil basic tasks due to a low fiscal
capacity or high debt levels. The fiscal capacity is captured by the average individual income levels
per capita in a municipality. Financially stronger municipalities on the other hand would probably not
agree to a merger with financially weak municipalities if this would cause their own financial situation
to worsen drastically. It can therefore be expected that differences in tax multipliers and debt levels
among the municipalities would lower the probability of a merger9 . Accordingly, pre-merger debt
levels and tax multipliers are included in the propensity score estimation.

5.3    Matching procedure

The first step of the matching procedure is to estimate the propensity score from the dataset. The
propensity score is defined as the conditional treatment probability given pretreatment characteristics
                                                                  
                                        p(X) ≡ P r D = 1|X = E D|X                                                     (1)

    Table 4 shows the variables that have been used in the propensity score estimation as well as
the signs of the respective coefficients. The full regression results can be found in table 13 in the
appendix10 . If not stated differently in the table variable values from three years prior to the first
merger in the respective cantons are used.
    Figure 2 shows the predicted propensity scores in the treated and control groups for each canton.
As a next step for each pre-merger municipality in the treatment group its nearest neighbour in terms
of the estimated propensity score is chosen. This nearest neighbour matching is implemented with
replacement. Thereby I impose exact matching on the cantons, the main language spoken in the
municipality and whether the municipality is a center or a tourist municipality. To this end, within
each canton as well as language and municipality type subgroup all observations are ordered based on
their propensity scores. I then chose for each pre-merger municipality the control municipality with
the closest propensity score. In order to compare similar municipalities the sample is restricted to
   9
     In some cantons, the merger grant itself might be able to absorb a substantial part of these differences, however only
up to a certain extent
  10
     I use the Stata command pscore by Sascha O. Becker and Andrea Ichino (http://www.stata-
journal.com/sjpdf.html?articlenum=st0026) to estimate the propensity score and obtain balancing tests

                                                            11
Table 4: Propensity score estimation by canton

                                                                                                       AG1              BE2             FR3                             GR4             LU5             SG                             SO6           TI7           VD8     VS

               =1 if ever invovled in a merger
               Population                                                                                   -**             -               -**                             +               -           -                               -*             -**          -*     +

               Mean income in TCHF                                                                          -*             +                -*                               -              -           -                              +               -            -       -*

               Total expenditures per capita                                                               +***             -*              +                               +**            +            -*                              -              -           +       +

               Net debt per capita                                                                         +               +                +                                -             +            +*                              -**           +             -       -

               Tax multipliers                                                                              -              +                +                               +***           +            -                              +              +            +***    +*

               Touristic mun. (2000)                                                                        .               .               .                               +               .           +**                             .             +             .       -**

               Center mun. (2000)                                                                          +**             +**              +                                .              -           +**                             .              .           +        -

               Rural mun. (2000)                                                                            -               -*              -                               +               -           +*                             +               -           +        -

               Inhabitants per hectare                                                                      -              +                +                               +              +            -                               -              -*          +       +

               German speaking majority (2000)                                                                             +                -***                             -                                                                                             +*

               Population growth in %                                                                                                       +

               Constant                                                                                    +                -*              +                                -**            -           +                               -              -            -***    -

               Observations                                                                                231           363            214                                 156           97            87                             123           173           372     147
               Stars indicate significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01
               1
                 AG: Being a tourist municipality predicts failure perfectly → 1 observation dropped
               2
                 BE: Being a tourist municipality predicts failure perfectly → 17 observations dropped
               3
                 FR: Being a tourist municipality predicts failure perfectly → 1 observation dropped
               4
                 GR: Being a center municipality predicts failure perfectly → 1 observation dropped
               5
                 LU: Being a tourist municipality predicts failure perfectly → 2 observation dropped
               6
                  SO: There are no tourist municipalities in SO. Being a center municipality predicts failure perfectly → 3 ob-
               servations dropped
               7
                 TI: Being a center municipality predicts failure perfectly → 3 observation dropped
               8
                 VD: Being a tourist municipality predicts failure perfectly → 4 observations dropped
               If not stated differently the variables are measured three years prior to the first merger in the respective can-
               ton. All financial data is measured in Swiss Francs per inhabitant at the price level prevailing in December 2010
               based on the national consumer price index.

those observations in the treated groups that can be matched to a counterpart with a similar treatment
probability. To this end a caliper of 0.1 around each treated municipality is defined. If no control
municipality lies within the caliper, the treated municipality is dropped from the sample. After this
procedure 18 mergers are lost, resulting in 142 mergers in the sample. The map in figure 3 shows the
municipalities considered in the sample as treated (red) and controls (green)11 .

                                                                         Figure 2: Common support by Canton

                                                                AG                                                      BE                                                          FR                                                         GR
                                                 1

                                                                                                       1

                                                                                                                                                                    1

                                                                                                                                                                                                                               1
                              Treatment status

                                                                                    Treatment status

                                                                                                                                                 Treatment status

                                                                                                                                                                                                            Treatment status
                                                 0

                                                                                                       0

                                                                                                                                                                    0

                                                                                                                                                                                                                               0

                                                     0   .2    .4    .6    .8   1                          0     .2    .4    .6    .8   1                               0    .2    .4    .6    .8   1                              0    .2    .4    .6    .8   1
                                                           Propensity score                                        Propensity score                                            Propensity score                                           Propensity score

                                                                LU                                                      SG                                                          SO                                                          TI
                                                 1

                                                                                                       1

                                                                                                                                                                    1

                                                                                                                                                                                                                               1
                              Treatment status

                                                                                    Treatment status

                                                                                                                                                 Treatment status

                                                                                                                                                                                                            Treatment status
                                                 0

                                                                                                       0

                                                                                                                                                                    0

                                                                                                                                                                                                                               0

                                                     0   .2    .4    .6    .8   1                          0     .2    .4    .6    .8   1                               0    .2    .4    .6    .8   1                              0    .2    .4    .6    .8   1
                                                           Propensity score                                        Propensity score                                            Propensity score                                           Propensity score

                                                                VD                                                      VS
                                                 1

                                                                                                       1
                              Treatment status

                                                                                    Treatment status
                                                 0

                                                                                                       0

                                                     0   .2    .4    .6    .8   1                          0     .2    .4    .6    .8   1
                                                           Propensity score                                        Propensity score

       In order to assess the quality of the matches, the standardized mean differences between treated
and control municipalities in the total sample are compared to the standardized mean differences in
  11
    This map is based on the municipality structure from the year 1998, so the merging municipalities are still drawn as
separate entities

                                                                                                                                        12
Figure 3: Mergers (red) and controls (green)

                     Municipality structure at the beginning of 1998. Sources: Swiss Federal Office of Topography
                     (2015), Swiss Federal Statistical Office (2016)

the matched sample (table 5)12 . It is clearly visible that the standardized differences are much smaller
for most variables in the matched sample compared to the total sample.
       I then aggregate the data for the treated municipalities and their counterparts to the respective
merger group. For each pair of treated and control municipality group the variable time is set equal
to zero in the merger year. So time = 0 in the graphs below indicates the results in the first year
after the merger. The results are displayed for up to ten years (time = 9). For the interpretation of
the results one needs to be aware that the panel is not balanced with respect to the time variable.
Since a large part of mergers occurred between 2009 and 2014, there are not too many observations
for the later years (see figure 4 showing the number of observations over time for the variable total
expenditures). Furthermore, since most early mergers occurred in the canton of Fribourg, the results
shown below for the late years are very much dominated by mergers from the canton of Fribourg.

5.4      Difference-in-differences analysis

Based on the aggregated values of the matched sample, I want to analyse whether expenditure patterns,
debt levels and tax multipliers are different for merged municipalities after the merger has taken place,
compared to non-merged municipalities. The first step consists of a very basic difference-in-differences
panel regression as in equation 2. Thereby the fiscal outcome variables (Yit ) are regressed on a constant
(α), a dummy for the merged municipalities (mergeri ), a dummy indicating whether a point in time
lies after the merger (af tert ) and an interaction term of these two dummies (mergeri ∗ af tert ). β3 ,
  12
    The standardized difference is defined as follows, with X̄ being the respective variable group means and V the
respective variance

                                                                  |X̄T − X̄C |
                                            SD = 100 ∗ r          h                i
                                                             0.5 ∗ VT (X) + VC (X)

                                                                 13
Table 5: Standardized mean differences

                                                                             Total Sample                                  Matched sample

                                                                   Treated       Controls          SD            Treated            Controls      SD

Demographic Variables
Population                                                         1217.88         2280.07        21.99              945.98         1124.72       8.10
Population growth in %                                                0.41            0.61         5.26                0.49               0.36    3.18
Foreigners in % (1999)                                                8.70           11.71        35.47                8.34               8.97    8.46
Inhabitants above 80 in % (2000)                                      3.56            3.22        15.56                3.57               3.66    3.41
Inhabitants below 20 in % (2000)                                     25.63           25.12        11.00               25.76            25.76      0.00
Mean income in TCHF                                                  61.24           67.79        43.78               61.14            60.67      4.16

Languages
German speaking majority (2000)                                       0.42            0.59        33.11                0.41               0.41    0.00
French speaking majority (2000)                                       0.39            0.31        15.80                0.41               0.41    0.00
Italian speaking majority (2000)                                      0.14            0.08        17.95                0.13               0.13    0.00
Romansh speaking majority (2000)                                      0.05            0.01        19.78                0.05               0.05    0.00

Fiscal Variables
Total expenditures per capita                                      5011.36         4855.41         6.46             4838.62         4785.35       2.22
Net debt per capita                                                2785.14         2042.95         9.59             2895.12         1964.87       8.66
Tax multipliers                                                     132.59          144.42        19.36              132.80          133.78       1.72

Municipality Types
Center mun. (small/medium/big) (2000)                                 0.02            0.02         1.72                0.01               0.01    0.00
Rural mun. (2000)                                                     0.63            0.49        27.93                0.65               0.66    0.98
Suburban mun. (2000)                                                  0.29            0.40        22.99                0.29               0.29    1.03
Touristic mun. (2000)                                                 0.06            0.07         4.88                0.04               0.04    0.00
High-income mun. (2000)                                               0.00            0.02        16.92                0.00               0.00    0.00

Geographic Variables
Area in ha (1998)                                                  1192.20         1498.65        13.19             1144.35         1497.31      17.16
Inhabitants per hectare                                               1.66            3.14        31.36                1.49               1.45    1.28

Observations                                                           490            1507                                430              430
The standardized difference is defined as follows, with X̄ being the respective variable group means and V the
respective variance

                                                                                |X̄T − X̄C |
                                                            SD = 100 ∗ r        h               i
                                                                           0.5 ∗ VT (X) + VC (X)

. If not stated differently the variables are measured three years prior to the first merger in the respective
canton. All financial data is measured in Swiss Francs per inhabitant at the price level prevailing in December
2010 based on the national consumer price index.

       Figure 4: Number of mergers observed over time and by canton

                                                                             Total expenditures

                                          150

                                          125
             Number of mergers observed

                                          100

                                          75

                                          50

                                          25

                                           0
                                                −10 −9 −8 −7 −6 −5 −4 −3 −2 −1         0     1    2     3   4   5     6    7    8     9

                                                       FR             BE                   VD                   LU                    SO
                                                       TI             GR                   AG                   SG                    VS

                                                                              14
Table 6: Effects on total expenditures per capita

                                                                     (1)        (2)

                                     Merged                         233.8      233.8
                                                                   (307.2)    (307.5)

                                     Merged * after                 85.0
                                                                   (156.8)

                                     Merged * 1 year after                     494.0***
                                                                              (180.8)

                                     Merged * 2-3 years after                  19.6
                                                                              (139.3)

                                     Merged * 4-5 years after                  -28.3
                                                                              (242.4)

                                     Merged * 6-7 years after                  249.4
                                                                              (375.2)

                                     Merged * 8-10 years after                 -205.6
                                                                              (354.6)

                                     Constant                      4909.0*** 4909.0***
                                                                   (159.6)   (159.7)

                                     Observations                   4286       4286
                                     Standard errors in parentheses. Stars indicate
                                     significance levels: * p < 0.10, ** p < 0.05, ***
                                     p < 0.01. After-dummies omitted.

the coefficient on this interaction term is the coefficient of interest.

                    Yit = α + β1 ∗ mergeri + β2 ∗ af tert + β3 ∗ mergeri ∗ af tert + ε                (2)

    In a second step the timing of effects is analysed in more detail. There might be short-term fiscal
effects that run out after some period of time. On the other hand it would also be possible that some
effects only turn up after some adjustment period. The after-dummy is therefore split into several
time intervals such as ”one year after the merger”, ”two to three years after the merger” and so on.

6     Results

6.1   Expenditure Effects

Table 6 reports the regression results for total expenditures. Column 1 displays the baseline results,
while column 2 differentiates with respect to the years after the mergers have taken place. The overall
effect on total expenditures per capita is statistically insignificant. Looking at time patterns it can
be seen that one year after the merger there seems to be a statistically significant increase in total
expenditures per capita. This may be due to the accounting for the cantonal merger grants. For the
later years there are no statistically significant results. Therefore, one can not conclude that municipal
mergers have had any effect on total expenditure levels per capita.
    Figure 5 provides a graphical representation of the group means of treated and controls before and
after the merger has taken place, in order to visualize the regression results. Right after the merger
there seems to be an increase in total expenditures for the group of merging municipalities, which
might be due to the cantonal merger grants. At later points in time (year six onwards) the difference
in expenditure levels between the two groups diminishes. It needs however be kept in mind that for
those late years, there are only about 50 municipalities in the sample.
    As a next step, the effects for the different expenditure types are analysed separately. Thereby
the focus lies on net expenditures. The difference between gross expenditures and net expenditures
is that category-specific revenues are subtracted from the gross expenditure levels. The reason is

                                                             15
Figure 5: Group-means of total expenditures per capita over time

                                                                          Total Expenditures
                                                      6000

                                                      5500

                            Expenditures per capita
                                                      5000

                                                      4500

                                                      4000
                                                             −10   −5             0                  5    10

                                                                        Years after the merger

                                                                        non−merged               merged

that if municipalities provide certain tasks together, gross expenditures might be counted twice in
the non-merged or pre-merger municipality groups. If one municipality provides some tasks for other
municipalities and is then compensated financially these payments are captured in category-specific
revenues. The costs for these services are then counted twice if gross expenditure levels before the
merger takes place are aggregated. The net expenditures represent the amount that needs to be
financed by general tax revenues. On the other hand however one can no longer distinguish between
cost savings and a shift from tax financing to higher charges for specific services.
   Table 7 shows the overall results for the various expenditure categories. It needs to be kept in mind
that no data is available for the canton of Ticino on different expenditure categories. Therefore these
results do not include municipalities from Ticino. For almost all expenditure categories the overall
coefficients on the interaction terms are negative but statistically insignificant. Only the category of
administration expenditures seems to be affected negatively by municipal mergers in a statistically
significant way.
   The time pattern of these effects is summarized in table 8. It can be seen that there is a highly
statistically significant negative effect of the mergers on administration expenditures two to three
years after the merger takes place. Administration expenditures decrease on average by about 40
Swiss francs per inhabitant which amounts to about 10% of average administration expenditures
per capita and to less than 1% of average total expenditures per capita. The respective coefficients
are similar for the later years, although just marginally (4-5 years after the merger) or no longer
statistically significant (more than 6 years after the merger), potentially due to the decreasing number
of observations. Another statistically significant effect can be made out for finance expenditures in the
year right after the merger occurred. This is potentially due to a one-off increase in finance revenues
when cantonal merger grants are received. A graphical representation of these results can be found in
figures 9 and 10 in the appendix.

6.2   Effects on quality of public services

The results of unchanged per capita expenditures after municipal mergers found so far could mean
that efficiency did not change or also that more or better output has been provided at the same cost.
As a second step I therefore try to approximate the output side by some variables that may serve as
indicators for a municipality’s attractiveness.

                                                                          16
Table 7: Baseline effects on functional categories of net expenditure

                                          Admin        Education      Safety       Culture         Health/Social        Economics      Finance

                    After the merger        -1.9          71.9***       2.5            21.3**         112.4***            33.1          -96.9***
                                           (12.1)        (24.4)        (5.2)          (10.8)          (15.9)             (36.0)        (37.2)

                    Merged                  12.2          -6.8          4.5            19.4            -7.0               34.2          -50.4
                                           (27.9)        (61.0)        (9.1)          (12.2)          (30.1)             (63.5)        (75.1)

                    Merged * after          -33.4**       -18.1         -9.2           -15.5           15.5               -35.9         -52.3
                                           (14.8)        (35.3)        (7.1)          (12.6)          (20.9)             (42.5)        (60.5)

                    Constant               419.2***      873.1***      72.2***        79.4***         529.5***           339.8***      212.3***
                                           (15.2)        (41.8)        (6.4)          (8.4)           (23.6)             (46.5)        (44.9)

                    Observations           3864           3864         3864           3864             3862               3860          3838
                    Standard errors in parentheses. Stars indicate significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01.
                    Municipalities from the canton of Ticino are not included since data on expenditure categories is missing.

                             Table 8: Effects on functional categories of net expenditures

                                               Admin       Education      Safety         Culture       Health/Social       Economics      Finance

               Merged                           12.2          -6.8          4.5            19.4               -7.0            34.2          -50.4
                                               (27.9)        (61.1)        (9.1)          (12.2)             (30.1)          (63.6)        (75.1)

               Merged * 1 year after            -12.0         -14.9         -8.4            -6.2              -5.6             7.1         -206.4***
                                               (14.1)        (18.9)        (5.7)           (8.3)             (23.3)          (55.2)        (71.2)

               Merged * 2-3 years after         -44.3***      -24.0         -9.0           -9.0               27.7            -35.8         2.0
                                               (13.3)        (26.3)        (8.3)          (10.0)             (20.6)          (45.8)        (68.9)

               Merged * 4-5 years after         -39.4*        -12.9         -13.0          -14.5              23.9            -51.2         4.5
                                               (23.1)        (46.0)        (10.4)         (16.3)             (27.3)          (65.6)        (81.1)

               Merged * 6-7 years after         -10.0         -18.9         -6.0           -25.8              14.2             1.3          -78.5
                                               (31.5)        (59.7)        (10.5)         (21.6)             (38.4)          (104.5)       (105.4)

               Merged * 8-10 years after        -50.4         -16.9         -8.4           -28.6              5.1             -95.6         -41.4
                                               (30.9)        (71.4)        (11.1)         (22.6)             (38.0)          (74.2)        (146.4)

               Constant                        419.2***      873.1***      72.2***         79.4***           529.5***        339.8***      212.3***
                                               (15.2)        (41.8)        (6.5)           (8.5)             (23.6)          (46.5)        (44.9)

               Observations                     3864          3864         3864            3864              3862             3860          3838
              Standard errors in parentheses. Stars indicate significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01. After-dummies
              are omitted. Municipalities from the canton of Ticino are not included since data on expenditure categories is missing.

      Assuming that increased quality on local public goods provision should be reflected in an inflow of
residents, the effects of municipal mergers on the population growth rate, the net immigration rate13
as well as on the median rental and housing prices in the municipalities is analysed. The results in
tables 9 and 10 do not show any effects of municipal mergers on these outcome variables, which is also
clearly visible in figures 6 and 7.

6.3      Further fiscal effects

As a further step the effects of mergers on net debt levels and municipality tax multipliers are con-
sidered (table 11). The estimates for the effects of mergers on net debt levels are mostly negative but
 13
      The net immigration rate is defined as immigration minus outmigration as a share of the existing population

                                                Table 9: Effects on quality outcomes

                                                      Rental prices     House prices           Net immigration        Population growth

                        After the merger                   9.8***          457.2***                   0.4***                  0.5***
                                                          (2.1)            (86.7)                    (0.1)                   (0.1)

                        Merged municipalities              0.6             -10.6                      -0.1                    -0.1
                                                          (3.5)           (115.7)                    (0.1)                   (0.1)

                        Merged * after                     -3.8            -88.9                      0.2                     0.2
                                                          (2.8)           (109.6)                    (0.1)                   (0.2)

                        Constant                         167.1***          4129.5***                  0.7***                  0.8***
                                                         (2.6)             (84.9)                    (0.1)                   (0.1)

                        Observations                      2942                 2942                  4412                    4406
                        Standard errors in parentheses. Stars indicate significance levels: * p < 0.10, ** p < 0.05, *** p
                        < 0.01. Datasource rental and house prices: Wüest & Partner

                                                                               17
Table 10: Effects on quality outcomes

                                                        Rental prices   House prices               Net immigration   Population growth

Merged municipalities                                        0.6            -10.6                        -0.1               -0.1
                                                            (3.5)          (115.9)                      (0.1)              (0.1)

Merged * 1 year after                                        -2.4            -65.6                       0.2                0.2
                                                            (2.0)           (79.6)                      (0.3)              (0.2)

Merged * 2-3 years after                                     -2.2            -5.0                        0.3                0.2
                                                            (2.4)          (101.5)                      (0.2)              (0.2)

Merged * 4-5 years after                                     -3.6           -115.7                       0.3                0.4
                                                            (3.5)          (155.6)                      (0.2)              (0.2)

Merged * 6-7 years after                                     -6.3           -167.8                       0.1                -0.0
                                                            (4.2)          (174.8)                      (0.2)              (0.3)

Merged * 8-10 years after                                    -5.4           -123.3                       0.3                0.2
                                                            (4.8)          (171.6)                      (0.3)              (0.3)

Constant                                                   167.1***        4129.5***                     0.7***             0.8***
                                                           (2.6)           (85.0)                       (0.1)              (0.1)

Observations                                                2942             2942                       4412               4406
Standard errors in parentheses. Stars indicate significance levels: * p < 0.10, ** p < 0.05, *** p <
0.01. Datasource rental and house prices: Wüest & Partner. After-dummies omitted

                                        Figure 6: Group means on house and rental prices

                                                                             Housing Prices
        Median selling price per m2

                                      4800
                                      4600
                                      4400
                                      4200
                                      4000
                                      3800

                                              −10            −5                      0                          5            10

                                                                          Years after the merger

                                                                        non−merged                    merged

                                                                             Rental Prices
        Median rental price per m2

                                      190

                                      180

                                      170

                                      160

                                      150
                                             −10             −5                    0                            5            10

                                                                         Years after the merger

                                                                        non−merged                   merged

    Figure 7: Group means on inmigration and population growth

                                                                            Net immigration
        Net immigration rate in %

                                       2

                                      1.5

                                       1

                                       .5

                                       0
                                            −10             −5                     0                            5            10

                                                                         Years after the merger

                                                                        non−merged                   merged

                                                                           Population growth
        Population growth in %

                                      2.5
                                       2
                                      1.5
                                       1
                                       .5
                                       0
                                            −10             −5                     0                            5            10

                                                                         Years after the merger

                                                                        non−merged                   merged

                                                                            18
Table 11: Effects on debt levels and tax multipliers

                                                             Net Debt           Net debt           Tax Multiplier    Tax Multiplier

                     After the merger                             -945.3**                             -12.4***
                                                                 (366.1)                               (1.5)

                     Merged                                       310.5           310.5                  -1.5             -1.5
                                                                 (590.6)         (591.1)                (5.3)            (5.3)

                     Merged * after                               -172.9                                 -3.7*
                                                                 (446.3)                                (2.1)

                     Merged * 1 year after                                        -108.2                                  -3.7***
                                                                                 (464.9)                                 (1.3)

                     Merged * 2-3 years after                                     -341.8                                  -3.7***
                                                                                 (453.9)                                 (1.3)

                     Merged * 4-5 years after                                     -486.8                                  -4.0*
                                                                                 (677.6)                                 (2.3)

                     Merged * 6-7 years after                                     -103.9                                  -3.2
                                                                                 (657.7)                                 (3.6)

                     Merged * 8-10 years after                                    418.6                                   -3.5
                                                                                 (753.6)                                 (5.5)

                     Constant                                    1777.8***       1777.8***             124.9***          124.9***
                                                                 (425.4)         (425.8)               (3.8)             (3.8)

                     Observations                                 4266             4266                 4726             4726
                     Standard errors in parentheses. Stars indicate significance levels: * p < 0.10, ** p <
                     0.05, *** p < 0.01. After-dummies are omitted.

not statistically significant. The estimates for the tax multipliers are negative but only marginally
statistically significant in the baseline specification in the third column. However, the results show a
clearly statistically significant negative effect of mergers on municipal tax multipliers for up to three
years after the merger. The coefficients for the later years are still negative but only marginally or
no longer statistically significant. The reason for these results might be that merged municipalities
typically adjust their tax multiplier to the lowest tax multiplier in the group in order to ensure the
populations’ agreement with the merger project.
               Figure 8: Effects on net debt per capita and municipal tax multipliers

                                                                                     Net debt
                                            3000
                      Net debt per capita

                                            2500
                                            2000
                                            1500
                                            1000
                                             500

                                                    −10     −5                            0                      5                10

                                                                              Years after the merger

                                                                             non−merged                merged

                                                                                 Tax multipliers
                                            140
                      Tax multipliers

                                            130

                                            120

                                            110

                                            100
                                                   −10     −5                           0                        5                10

                                                                              Years after the merger

                                                                             non−merged                merged

7    Conclusion

This paper explores the fiscal effects of municipal mergers in Switzerland. Applying a matching
approach, 160 mergers between 2001 and 2014 are analysed, considering the separate year effects for
up to ten years after the merger. Furthermore, both total expenditure and the separate spending

                                                                                19
categories are considered. The results do not yield substantial evidence for scale economies. While a
decrease in administration expenditure levels can be observed for some years, I do not find statistically
significant negative effects on overall expenditure levels. Furthermore, this study does not provide
conclusive evidence that the quality of local public services has been improved in a substantial way
either. At the same time, the results indicate that the financial grants paid out by the cantons
might have been used to finance tax cuts, thereby ensuring that a merger is accepted by the involved
municipalities’ inhabitants.
   One potential reason for these results might be that efficiency gains through increased size are
already exploited through inter-municipal cooperation prior to the merger. If voluntary mergers mainly
occur between municipalities that already engage in extensive inter-municipal cooperation, as seems
to be the case for Switzerland, this could be an explanation for previous findings, according to which
expenditure effects of mergers are more pronounced for compulsory mergers than voluntary mergers.
Compulsory mergers might also take place among municipalities that do not yet cooperate and where
efficiency gains have not yet been exploited through pre-existing inter-municipal cooperation.
   The analysis leads to the conclusion that voluntary municipal mergers in Switzerland should not
primarily be promoted for financial reasons. Other reasons, such as alleviating the difficulty to recruit
staff or simplifications from a cantonal viewpoint might nevertheless prevail.

                                                   20
You can also read