Content uploaded by Antti Moisio
Author content
All content in this area was uploaded by Antti Moisio on Sep 07, 2017
Content may be subject to copyright.
Public Finance and Management ISSN 1523-9721
Volume 13, Number 3, pp. 148-166
2013
THE IMPACT OF MUNICIPAL MERGERS ON LOCAL
PUBLIC EXPENDITURES IN FINLAND
Antti Moisio
Government Institute for Economic Research, Helsinki, Finland
Roope Uusitalo
HECER, University of Helsinki, Helsinki, Finland
ABSTRACT
This article examines the effects on expenditure of municipal mergers that took place in
Finland between 1970 and 1981. We collected data on pairs of municipalities that merged
after 1970, and matched these with the similar municipal pairs that remained independent
using characteristics from the years immediately before the merger. We compared the chang-
es in per capita spending after the merger in the municipalities that merged to the municipali-
ties that remained independent over the same time period. Our results indicate that municipal
mergers did not lead to lower per capita spending. In most spending categories, the per capita
expenditure increased more in the merged municipalities than in the comparison group. Only
in the category of general administration did the per capita spending decrease; however this
decrease was far smaller than the increase in spending in other categories. For the first years
of the mergers, the spending increases may be explained by transitional costs, such as acqui-
sition of new technology, renovation of existing facilities and upward harmonization of wages
and salaries. Nevertheless, it is striking to find that even ten years after amalgamation, spend-
ing was still higher in the merged municipalities compared to similar municipalities that
chose to stay independent.
1. INTRODUCTION
As the role of local governments in public sectors has grown, so too the
question of an optimal size of local jurisdictions has become more important.
Consequently, the mergers of municipalities are salient in the political agendas
of several countries. The proponents of the mergers usually argue that a larger
municipal size will lead to economies of scale in service production as well as
lower administrative expenditures (King, 1984; Oates 1985; Fox and Gurley
2006). Other standard arguments for mergers include that they internalize pub-
lic service spillovers, improve the quality of the public services and create a
more attractive environment for business as well as for inhabitants (Slack and
Bird, 2012).
Alternatively, the opponents of municipal mergers argue that small munic-
ipalities can utilize economies of scale by purchasing the services from spe-
cialized providers, or by co-operating with other municipalities. Hence, if
municipalities are providers and not just producers of the services, the optimal
Municipal Mergers in Finland
149
size of municipalities would depend on other factors (King, 1984). Other ar-
guments against large municipal size are that small municipalities are more
flexible, have less bureaucracy and better local democracy (Oates 1972; Borge
and Rattso 1993; Dollery and Fleming 2006; Drew et al. 2012; Slack and Bird
2012; Andrews and Boyne 2012, Lago-Peñas and Martinez-Vazquez 2013).
Although the key motivation for municipal mergers is that larger munici-
palities would be more cost effective, not a lot is known about the effects of
the changes of municipal size. Much of the previous empirical research has
focused on the effect of annexation to municipal expenditures. For example,
Mehay (1981) found that population growth by annexation had a positive ef-
fect on expenditure growth rates. Liner (1992) found that annexation was posi-
tively related to expenditure growth but inversely related to per-capita
expenditure growth rates and with growth rates in public employment. Liner
(1994) found that per-capita expenditures and the number of municipal em-
ployees did not change significantly with a change in annexation laws.
Previous evidence on the effects of municipal mergers is mostly based on
time-series or cross-section analysis. For example, Nelson (1992) used time-
series data from Sweden from 1942 to 1987 to analyze the effects of municipal
mergers on expenditure growth. He found that the reduction in the number of
small rural municipalities after the 1952 reform constrained expenditure
growth, whereas the reduction in the number of larger non-rural municipalities
following the 1962 reform had a contrary effect. Nelson concluded that more
units of government serving a given population can constrain public sector
budgets, but only as long as these units exceed some threshold size. Hanes
(2001) also used Swedish data and compared expenditure growth of munici-
palities that had merged to those that stayed independent. Like Nelson, Hanes
found that the 1952 reform that reduced the number of municipalities had a
negative effect on expenditure growth of the merged municipalities. However,
the effect was apparently due to selectivity. After selection bias was controlled
for, Hanes found no statistically significant effects of reform on expenditure
growth. More recent analyses of Swedish mergers by Hinnerich (2009) and
Jordahl and Liang (2010) have focused on municipal behavior prior to the
merger. They find that municipalities accumulate debt before merger because
the taxpayers in the new municipality will share the costs. A study by Blom-
Hansen (2010) analyzed Danish amalgamation reform, also finding support for
the common pool problem. In addition, a study by Andrews and Boyne (2012)
on the performance of English county councils before voluntary restructuring
found some adverse effects on expenditure, service performance and value for
money in local service provision. A recent study by Reingewertz (2012) ana-
lyzed the effects of local government reform in Israel in 2003. Using panel
data for the years 1999–2007 and difference-in-differences methods,
Reingewertz found that amalgamations resulted in a decrease of about 9% in
Moisio & Uusitalo
150
municipal expenditures. Moreover, he found no evidence of a decrease in the
level of services provided to the residents of the amalgamated municipalities.
Therefore, Reingewertz (2012) concluded that municipal amalgamations can
internalize economies of scale.
The effects of municipality mergers can also be evaluated based upon case
studies. In fact, most merger decisions involve detailed cost calculations prior
to the merger, but only anecdotal evidence is available on post-merger out-
comes. Even if changes in per-capita spending were calculated after the mer-
ger, it is not likely that case studies could provide results that could be
generalized to a wider population of municipalities. Case studies also face the
problem of missing counterfactual outcomes. Increases in the number of tasks
of local governments leads to increases in per-capita spending, irrespective of
whether municipal merger took place.
This paper seeks to determine the effect of municipal mergers on per-
capita spending in various spending categories. We accomplish this by com-
paring changes in the per-capita spending in municipalities that merged to the
changes in comparable municipalities that remained independent over the
same time period. We collected data on all Finnish municipalities (or rather
municipal pairs) that merged during the period 1970 to 1981, and matched
these to municipal pairs that remained independent using characteristics from
the years immediately before the mergers. We then compare the changes in
average per-capita spending between the municipalities that merged and the
comparison group of independent municipalities over a period of time that
covers both the years around the merger date and the adjustment period after
the merger.
Compared to most of the previous studies our study makes the following
contributions. First, by concentrating on the changes in spending instead of the
level of spending in a cross-section, we can control for the pre-existing differ-
ences across the municipalities. Second, by comparing the changes in spend-
ing to comparable municipalities over the same time period we can control for
general changes in spending. These changes may be due to increases in the
responsibilities of the municipalities or to the business cycle or other macroe-
conomic shocks that have a similar effect on all municipalities. Further, as we
compare the changes in spending between similar municipalities we can ac-
count for the heterogeneity between municipalities and avoid comparing mu-
nicipalities that differ in characteristics affecting the spending growth.
Moreover, by using panel data for the municipalities that merged in the
1970’s, we examined changes over a long period of time so that any adjust-
ments can be accounted for.
The paper is organized as follows. In the second section we briefly de-
scribe the main features of Finnish local governments. In section three we pre-
Municipal Mergers in Finland
151
sent our estimation strategy and in section four we describe the data used in
the study. The fifth section presents the empirical results and section six con-
cludes the paper.
2. FINNISH LOCAL GOVERNMENT
Finnish local governments are responsible for a wide range of public ser-
vices. The number of tasks started to increase especially rapidly since 1970. In
1970 there were 114 separate municipal tasks; in 1990 the number of tasks
was 265 and in 2012 the number of tasks had grown to 535 statutory munici-
pal tasks (Ministry of Finance 2012). Currently, all main social welfare, health
care and education services are provided by municipalities or by joint munici-
pal authorities. Hence, the overall economic importance of the municipal sec-
tor is considerable. Local government spending as share of GDP is around
18% and municipalities employ roughly 20% of the total Finnish workforce.
In 1944 Finland was divided into 603 municipalities. The 1970s was an
extraordinary decade from the perspective of municipal mergers. Due to vol-
untary municipal amalgamations in the beginning of 1970s, the number of
municipalities dropped from 517 to 464. After that, restructuring has been
much slower. During 1980s and 1990s only twelve mergers occurred. In 2005
the government intensified the reform policy and in a few years the number of
municipalities dropped by more than a hundred municipalities so that in 2013
there were only 320 municipalities. Despite this development, the median mu-
nicipality still has less than 6000 inhabitants.
In order to cope with their increased responsibilities, the smallest munici-
palities have sought partnerships and cooperative arrangements with other
municipalities and with the private sector. The most common form of munici-
pal cooperation is the joint authority. Joint authorities are set up by two or
more municipalities, mainly for tasks that require a larger population base than
small municipalities have. Membership in a joint authority is voluntary except
for hospital services and regional councils. At present, joint municipal authori-
ties are solely financed by member municipalities, but in the 1970s the central
government funded joint authorities with state subsidies.
Finnish municipalities make spending decisions independently, and decide
on the level of local income and property taxes. The budget is approved by a
municipal council that is elected by the residents for four year terms. Opera-
tional decisions are made by a municipal board and a municipal manager; both
elected by the municipal council (see, for example, Moisio et al., 2010).
Local government revenues consist mainly of local income tax, property
tax, state subsidies and user fees. From the 1960s until 1993 the grant system
Moisio & Uusitalo
152
consisted mostly of matching grants. State subsidies varied depending on the
financial capacity of the municipality, and each task had a different matching
rate scale. The so called “capacity classification” divided municipalities into
10 groups based on per capita tax base, the financial condition of the munici-
pality, population density and the unemployment rate (Oulasvirta, 1997). The
higher the municipality was ranked in the classification, the less state support
(because of lower matching rate) it received. In 1993 a new grant system
based on general, non-earmarked grants was introduced. The capacity classifi-
cation was abandoned in 1996, when a separate revenue equalization formula
was introduced.
3. ESTIMATION STRATEGY
The most straightforward way to analyze merger effects is to regress per-
capita expenditures on various socio-demographic and economic variables and
an indicator as whether a merger has taken place. The approach is perfectly
valid if all other variables that affect expenditures are controlled for in the
regression, and if the functional form of the model is known. However, if the-
se assumptions do not hold, the estimates of the merger effects may be severe-
ly biased.
To avoid these problems, our estimation strategy followed standard proce-
dures in the matching literature (see, for example, Lechner 2002; Deheija and
Wahba 2002; Heckman, et. al., 1997). We defined the parameter of interest as
the change in per capita spending in the merged municipalities due to the mer-
ger. In other words, we attempted to estimate the average treatment effect on
the treated municipalities.
Let Y1 indicate the outcome in the treated and Y0 the outcome in the untreated
case. Let D be an indicator for the treatment status with D = 1 indicating that
the municipality has been merged and D = 0 otherwise. Further, let X be a set
of variables that affect both the merger decisions and the economic outcomes.
The parameter of interest is the expected treatment effect on the treated, E(Y1 |
D=1) - E(Y0 | D=1).
The fundamental problem is that data on counterfactual outcomes is miss-
ing. The average outcome in the merged municipalities E(Y1 | D=1) can be
easily estimated but their outcome, had they not been merged E(Y0 | D=1) can
never be observed. The difference between treated and untreated municipali-
ties E(Y1 | D=1) - E(Y0 | D=0) can be estimated but this is a potentially biased
estimator of the treatment effect if the treated and the untreated municipalities
differ in a systematic way.
In a randomized trial the treatment status would be independent of the
characteristics of the individual municipalities. Therefore, E(Y0 | D=1) = E(Y0
Municipal Mergers in Finland
153
| D=0), and the untreated municipalities could be used as a comparison group
yielding unbiased estimates. Non-experimental studies need to resort to vari-
ous assumptions to substitute for a randomized control group.
Both the regression and the matching methods estimated the treatment ef-
fect assuming that treatment status is independent of potential outcome in the
non-treated case, conditional on a set of control variables X. Another way of
stating this conditional independence assumption is that given the set of condi-
tioning variables, the assignment to treatment is essentially random. Condi-
tioning on the X variables removes all systematic differences between the
treated and the untreated observations.
In the matching approach each treated observation was paired with mem-
bers from the control group that have similar observed attributes and the
treatment effect was estimated simply as the difference in mean outcomes be-
tween the treatment and the matched control group. Practical difficulties arise
when the X vector is high dimensional, i.e.: the number of attributes that
should be similar in the treatment and control group is large. However, Rosen-
baum and Rubin (1983) have proven that, instead of conditioning on a large
number of X variables, it is sufficient to condition on a scalar function of X,
namely the participation probability, conditional on the attributes. If the condi-
tional independence assumption holds while conditioning on all X variables, it
also holds when conditioning only on the participation probability.
1
In this paper we used the nearest neighbor matching. We first estimated
simple logit-models explaining the municipal mergers with a set of character-
istics of the municipalities. We then selected to the comparison group the mu-
nicipalities that provided the closest match to the merged municipalities, and
evaluated the effect of municipal mergers by comparing the expenditure
growth in the merged municipalities to their nearest neighbors.
While the approach is standard in the literature, the application to munici-
pal mergers required some adjustments. An obvious complication is that a
merger involves two municipalities, and that the characteristics of both munic-
ipalities may affect the merger decisions and the economic outcomes. A relat-
ed problem is that after the merger, there is no longer separate information on
the expenditures in the previously independent municipalities, but all estimates
of expenditure growth must be based on the joint outcomes of the merged mu-
nicipalities.
1
. With a small number of characteristics matching is straightforward. One simply splits the
data into cells defined by the X variables. When there are many X variables it is more difficult
to determine along which dimensions to match units and how much weight should be given to
different variables. Propensity score matching methods are particularly useful because they
provide a natural weighting scheme that yields unbiased estimates of the treatment effect.
Moisio & Uusitalo
154
We approached the problem by re-defining the units of observation as mu-
nicipality pairs instead of individual municipalities. In principle, each munici-
pality can merge with any other municipality creating n(n-1)/2 potential
mergers. However, in practice, all mergers have occurred between municipali-
ties that share a common geographical border. We, therefore, also concentrat-
ed only on these municipalities
2
.
We first created the data of all municipality pairs that shared a common
border. We then linked to these data to a large number of municipal character-
istics from the year 1970. It turned out to be useful in ordering the municipal
pairs by size so that the larger municipality always occurs first in the data. We
then used the characteristics of the larger and smaller municipalities as sepa-
rate variables explaining the mergers
3
.
Given that our matching procedure successfully identified the municipal
pairs that are similar in terms of factors that affect per capita spending, there
were several alternative estimators for the effect of mergers on municipal
spending. We used nearest neighbor matching that created the control group
from the non-merged municipalities that were closest in terms of treatment
probability to the control group.
3.1. WEIGHTING
There were a number of issues regarding the correct weighting of the mu-
nicipalities in the treatment and the control groups. First, no separate records
are kept for the municipalities that did merge, only joint spending is observed.
Calculating spending per capita after the merger automatically weights the
municipalities by their population sizes. To keep calculations consistent, the
pre-merger spending in the merged municipalities had to be calculated using
population weights. The simplest possible estimator would then be obtained
by calculating average spending per capita in treatment and in the control
groups using population weights.
However, the municipalities varied greatly in size. The largest municipali-
ty involved in the mergers (Tampere) had 160 000 inhabitants, while several
2
. A matching estimator requires that the treated and untreated observations share a common
support. This implies that all municipality pairs have a non-zero probability of merging. Since
all mergers have occurred between adjacent municipalities including municipalities that do not
share a border would also violate the common support condition.
3
. In a typical case the decision of the smaller municipality is critical for the merger. The
smaller municipality needs to compare the economic benefits of the merger to the loss of in-
dependence. The larger municipality has less to lose in the merger. Examination of the past
merger proposals revealed that all the merger proposals that were rejected were rejected be-
cause of the opposition from the smaller partner.
Municipal Mergers in Finland
155
of the merged municipalities had populations of less than one thousand. Using
population weights is then almost equivalent to ignoring the smallest munici-
palities. To some degree the problem was unavoidable. For example, the joint
per capita spending in Tampere (with a population of 160 000) and Teisko
(with a population of 2993) that merged in 1972 necessarily weights Tampere
much more heavily. However, there is no good reason to give the mergers that
involved large municipalities a higher weight in calculating the average effects
of mergers. Therefore, we calculated average per capita spending in the treat-
ment and control group using population weights within each treatment-
control pair, but calculated un-weighted averages across pairs.
3.2. TIMING ISSUES
Merging two municipalities to one is not likely to lead into immediate cost
savings. In a typical merger agreement, all municipal employees are trans-
ferred to the new organization and there is a guarantee that no one is fired
within next five years. Potential savings are likely to occur with a considerable
lag, after the tasks of the new municipality have been re-organized.
Another problem is that municipal spending may be unusual during the
years immediately before the merger. At least for the smaller of the merging
municipalities, there are strong incentives to increase spending and spread the
resulting debt with the merger partner.
To create data where the effects of mergers occurring in the different years
could be meaningfully compared, we collected for each merged pair, and its
control pair observations from the years t - 3 to t + 10, where t = 0 denotes the
year of the merger. The resulting data, therefore, followed different municipal-
ities for the different calendar years but the data was balanced in the sense that
the distribution of calendar years in the treatment and the control group was
exactly the same. This procedure was thought to fully account for general
changes in municipal spending.
4. DATA
Our main data source was the Regional Statistical Database of Statistics
Finland (ALTIKA). This database contains information on municipal finances,
and regional statistics derived from the Population Census and the Labor
Force Surveys. ALTIKA contains comprehensive information on all munici-
palities from 1975 to the present. For the years 1973-74 the data for most eco-
nomic variables were obtained from the old electronic data files in the
Moisio & Uusitalo
156
Statistics Finland archives. Data prior to this had to be collected by hand from
the annual publications of the Statistics Finland.
4
Table 1. Merged municipalities compared to other municipalities*
Merged munici-
palities
n = 82
Control
group
n=82
Other municipali-
ties
n = 375
T-test for
merger-
control
difference
Population
12,072
14,125
7,824
0.12
Taxable income
3,895.40
4,083.54
3,226.97
-0.50
Operating expenditures
818.84
827.47
795.59
-0.30
Administrative expend-
itures
48.47
46.19
44.59
0.74
Welfare expenditures
131.68
138.63
124.41
-1.23
Health care expendi-
tures
141.33
138.29
145.76
0.57
Expenditures on educa-
tion
279.67
301.47
319.81
-0.89
Share of primary pro-
duction workers
28.56
29.80
43.40
-0.28
Income tax rate
13.87
13.45
14.16
1.55
Long-term loans
403.29
394.62
379.34
-0.27
Land area, km2
325
412
659
0.21
* All variables measured in FIM per capita in 1970. The t-statistics are based on regression
models estimated separately for each covariate. As the nearest neighbor matching is done with
replacement, some municipalities are used several times as controls. This is taken into account
by using weights.
The selection of explanatory variables was based on previous research on
determinants of municipal expenditures (Oulasvirta 1997; Moisio 2002). The-
se variables included population size, tax base, labor force characteristics and
the unemployment rate. We also collected variables that could predict the
probability of merger. These variables consisted of distance between munici-
pal administrative centers, population size, tax base, and indebtedness. In Ta-
ble 1 we report the means of the key variables measured for 1970. We
calculated these separately for the merged municipalities, for the control group
and for the rest of Finnish municipalities. We also tested whether the merged
municipalities differed from the comparison group. According to the t-
statistics reported in Table 1, there are no significant differences in any of the
characteristics included. The group means also revealed that our comparison
group resembled the treatment group much more than the municipalities that
were not matched.
4
. These databases do not contain information on the municipalities that have ceased to exist
because of a mergers. Information on those municipalities has been collected from municipal
accounts published annually by Statistics Finland.
Municipal Mergers in Finland
157
In order to examine changes in municipal expenditures over a long period
of time, we concentrated on mergers that took place during 1970-1981. This
enabled us to follow the municipalities 10 years after the merger, which
should be enough to take account the adjustment period. In total we collected
information on 57 mergers, of which 41 occurred during the 1970s. With only
a few exceptions all the mergers were voluntary. Since 1973 municipal mer-
gers have been encouraged by extra state grants.
Between 1970 and 1991 all major municipal expenditures grew continu-
ously (Figure 1). The period from 1970s to the end of 1980s was marked by
improving the welfare, health care and school systems in Finland. As a result
of this development, Finnish municipalities became responsible for a major
share of public service production. Municipalities run both the primary and the
secondary school system. Primary health care is organized around municipal
health care centers and central hospitals are managed by joint authorities of
municipalities. Municipalities are also responsible for most social services and
the social assistance.
Figure 1. The index of major municipal fixed-price expenditures 1970-
1991 (1970=100)
0
100
200
300
400
500
600
700
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
Welfare
Total operating
expenditures
General
administration
Health care
Education
As we have noted, Finnish municipalities vary a lot in population size. In
order to get an idea of the relationship between municipal size and expendi-
tures, we plotted the per-capita expenditures in the main spending categories
against population size (Figure 2). Figure 2 describes the situation in 2000. We
found that the only expenditure category that clearly decreases with increasing
population size is the general administration. For welfare services, a U-shaped
relationship between population size and expenditures was found. Per-capita
spending was lowest in municipalities with populations of about 10 000. For
the other categories the situation is less clear. In sum, there seemed to be no
Moisio & Uusitalo
158
obvious link between population size and per capita expenditures among Finn-
ish municipalities.
Figure 2.
Per-capita spending in 2000 by population size in main categories*
2000 4000 6000 8000 10000
100 1000 10000 100000
Welfare
2000 6000 10000
Per-capita spending, FIM
100 1000 10000 100000
Education
4000 6000 8000 10000
100 1000 10000 100000
Health care
2000 4000
Per-capita spending,
FIM
100 1000 10000 100000
Administration
Note: The dashed line indicates a locally weighted regression prediction of spending by log
population.
5. EMPIRICAL ANALYSIS
The 517 municipalities that existed in 1970 formed 1391 adjacent munici-
pal pairs. We first estimated the merger probabilities using data from all these
pairs. The explanatory variables used in the final model were population size,
per-capita taxable income, per-capita spending, and the fraction of primary
production workers of total employment. We included these variables sepa-
rately for both potential merger partners, so that subscript “1” denotes the
larger (by population) municipality in the pair and subscript “2” denotes the
smaller. In addition, we included the distance between the administrative cen-
ters of the municipal pairs.
The estimated coefficients are reported in Table 2. We found that even
though we concentrated on adjacent municipal pairs, the distance between the
administrative centers was a strong predictor of municipal mergers. We also
found that the larger the larger (by population) and the smaller the smaller of
the two potential merger partners was, the more likely the merger. This is also
Population
Population
Municipal Mergers in Finland
159
a natural result, as in 1970s a typical merger involved a large city that annexed
smaller surrounding municipalities.
We were surprised to find that economic variables were not statistically
significant in explaining the probability of a merger. The signs of the variables
suggest, however, that municipal mergers were more likely if the larger mu-
nicipality was economically weak and the smaller municipality was not in a
poor economic situation. This could be understood by considering mergers as
two-sided decision processes. It is not sufficient that a small municipality
seeks solutions for its economic problems from a merger, but the small munic-
ipality must also be accepted by its larger neighbor. Perhaps very poor munic-
ipalities had trouble finding partners that would be willing to share the
financial burden. The larger municipality accepts the smaller more easily if it
itself is in economic trouble. Hence, larger municipalities that have an expen-
sive public service sector were more likely to merge, perhaps hoping to cut the
costs of providing public services.
Table 2. Logit estimates of merger probability for mergers 1970 - 1981
Variable
Coefficient
Log(Distance)
-1.596 (0.297)**
Log(Population 1)
0.574 (0.251)*
Log(Population 2)
-1.066 (0.237)**
Taxable income/ capita 1
-0.288 (0.320)
Taxable income/ capita 2
0.139 (0.280)
Operating expenditures/capita 1
2.003 (1.242)
Operating expenditures/capita 2
-0.011 (1.360)
Primary production workers/total workforce 1
-0.005 (0.015)
Primary production workers/total workforce 2
-0.007 (0.015)
Constant
3.816 (2.969)
N
1391
Standard errors in parentheses * significant at 5%; ** significant at 1%
In the second step we created the control group from the pairs of munici-
palities that did not merge. First we ranked all municipal pairs according to the
estimated probability of a merger. We then chose for each merged pair the
closest match from the pairs that did not merge. We undertook the selection
with replacement so that the same municipality pair might be selected several
times to the control group. Also the same municipality might be in the control
Moisio & Uusitalo
160
group several times with different potential merger partners. However, we
excluded from the control group all municipality pairs where one of the mu-
nicipalities was involved in an actual merger during the period examined.
The comparison group consisted of many intuitive “twins”. Some of the
control pairs actually did merge later on in 1990s. Some of the pairs were less
intuitive. Still, in a regression analysis one would compare the merged munic-
ipalities with all the other municipalities and this implicitly involved a much
greater number of non-intuitive comparisons. Moreover, a formal test that we
reported in Table 1 found no systematic differences in the explanatory varia-
bles between the merger group and the control group.
We evaluated the effects of municipal mergers by comparing the develop-
ment in per-capita spending between the merger group and the matched con-
trols. Figures 3, 4 and 5 show the comparison between merged municipalities
and their control municipalities in different expenditure categories. The upper
part of each graph shows the development of the expenditures in treatment and
comparison groups around the year of the merger. The lower part displays the
difference in per-capita spending each year, together with the 95% confidence
bands for the difference
5
. We compare the two groups three years before and
ten years after the merger.
According to our results the mergers tended to lead to growth in per-capita
spending. Figure 3 shows that the total per capita operating expenditures were
at the same level before the merger but started to differ soon after, and after
five years the difference is statistically significant. Ten years after the merger
the difference between the two groups was 100 FIM per capita, or 6% of the
control group mean. The result is in sharp contrast with the usual pro-merger
arguments according to which mergers are justified by economic efficiency
effects that would lead to slower expenditure growth.
We calculated comparable estimates for all main expenditure categories.
Only in the case of general administration did we find that the mergers had a
negative effect on expenditures (Figure 4). The result is close to being statisti-
cally significant at 5 % level in a two-sided test. This result indicates that it is
possible to reduce administrative costs through mergers and indicates that
there might be scope to increase cost-efficiency by increasing the size of mu
nicipalities. Of course, the share of spending to general administration of the
total operating expenditures is small, so savings here will not make a differ-
5
. The confidence bands are calculated from the differences across matched pairs according to
the standard formula
ndsdtd )(
025,0
where
d
is the mean of the differences, sd(d) the
standard deviation of the differences, n the number of municipality pairs, and t0,025 the 2.5%
critical value of the t-distribution with n-1 degrees of freedom. The test takes into account the
correlation of the expenditures between the merged pairs.
Municipal Mergers in Finland
161
Figure 3. Total operating expenditures for merger municipalities vs. the
comparison group
.8 11.2 1.4 1 .6 1.8
-5 0 5 10
Mer ge d C ompa r is o n
Spending per capita, 1000 mk
-.1 0.1 .2
-5 0 5 10
Yea rs after mer g er
Merger effect
Figure 4. Per capita expenditure for general administration
.06 .08 .1 .12 .14
-5 0 5 10
Mer ge d C ompa r is o n
Spending per capita, 1000 mk
0
-.03
-5 0 5 10
Yea rs after mer g er
Merger effect
.01
1000 FIM per capita
1000 FIM per capita
Moisio & Uusitalo
162
ence if at the same time there are considerable expenditure increases in other
sectors. The estimated effect is about eight percent of the control group mean
but in absolute terms only around 10 FIM per capita ten years after the merger.
Figure 5 presents the results for expenditure in education. It appears that
per-capita expenditure in education increased in the merged municipalities
soon after the merger. This difference in per-capita spending is also statistical-
ly significant. After ten years of the merger, expenditures in merged munici-
palities are about 80 FIM per capita higher (11 %) than in the control group. A
potential reason is that political pressure made it difficult to create more effi-
cient school networks by, say, closing small schools.
For the sake of brevity we do not report in detail the results for the remain-
ing expenditure categories. However, they can be briefly summarized as fol-
lows. Expenditures on welfare were not affected by the mergers. The changes
in welfare expenditures are very similar in the treatment and comparison
groups and the difference is far from being statistically significant. As for the
health care expenditures, we found that ten years after the mergers, per capita
expenditures in the municipalities involved in the mergers was around 40 FIM
(12 %) higher than in the comparison group. This result is in the borderline of
being statistically significant at the end of the period.
Figure 5. Per capita expenditure for education
.3 .4 .5 .6 .7 .8
-5 0 5 10
Mer ge d C ompa r is o n
Spending per capita, 1000 mk
0
-.05 .1 .1 5
-5 0 5 10
Yea rs after mer g er
Merger effet
In sum, the increase in total per capita operating expenditures for the
merged municipalities originated mainly from increased expenditures on
.05
1000 FIM per capita
Municipal Mergers in Finland
163
health care and education. The decrease in expenditure for general administra-
tion is way too small to compensate for the increases in other spending catego-
ries.
In addition to comparing the changes in per-capita spending, we also com-
pared the changes in taxable income per capita, the income tax rate and long
term loans per capita. The aim was both to figure out what were the source of
funds that the merged municipalities used to finance additional expenditures
and to examine the validity of claims that increasing municipal size creates
financially stronger units. If there was a positive effect on tax revenues and if
municipalities exhausted higher tax revenue to increase the quality of services,
our findings might be due to effects on the revenue side of local government
budgets.
To examine these effects we used a similar setup as with the spending es-
timates. We found no effects on either the taxable income per capita or on the
tax rates. Hence, we found no support for the claim that larger municipalities
would provide a more attractive environment for firms which could eventually
increase their tax base. However, we did find some evidence that the merged
municipalities accumulated more debt after the merger than the comparison
group. These effects were not very precisely estimated and the differences
between merged municipalities and the matched controls were therefore not
quite statistically significant. However, the point estimates were quantitatively
large enough so that they could provide funding for the share of spending that
the municipality had to finance from its own resources.
In fact, the matching grant system may largely explain why we found no
decrease in per-capita spending. Under the matching grant system the munici-
palities had little incentives to improve efficiency. To examine the issue more
closely we performed a similar analysis for the mergers that took place after
the matching grant system had been abolished in 1993. This turned out to be
difficult because of the very small number of mergers. There were no mergers
between 1982 and 1989, and only nine between 1989 and 1997. Due to small
sample sizes the estimates were very imprecise. Between 1997 and 2004 mer-
ger activity increased and between 2005 and 2009 59 merges took place.
However, the analysis will have to wait until enough data of post-merger out-
comes becomes available.
6. CONCLUSION
In this study we analyzed the effects on expenditure of Finnish municipal
mergers. Our empirical results are based on a comparison of spending changes
in the merged municipalities and the municipalities that had similar character-
istics but remained independent. The main benefit of the method was that by
Moisio & Uusitalo
164
using the matched control group and by comparing changes over time we were
able to isolate the effects of the mergers from the other factors that affected the
expenditures.
According to our results, increasing the municipal size by merging two
small municipalities did not reduce local government expenditures. Per-capita
spending on general administration decreased but this had little effect on total
expenditures. At the same time, spending increased in key expenditure catego-
ries such as education and health care.
A partial explanation for our results may be that state grants to municipali-
ties were almost totally based on matching grants during the period. The mu-
nicipalities did not have strong incentives to constrain their spending after the
mergers, when the state always covered a share of their expenses. The munici-
palities seem to have used the extra grants, as well as the savings made in ad-
ministration, in new expenditures. However, the sum of extra grants and
savings in the administrative costs was less than one fifth of the increase we
found in total operating expenditures. The rest of spending increases were ap-
parently financed by accumulating more debt. This was possible because in
Finland the municipal borrowing was not tightly constrained by fiscal rules.
An important caveat in our study was that we could not measure changes
in the availability or quality of services. Therefore, we cannot be sure to what
extent our results can be explained by improvements in services, such as the
opening of new schools or heath care centers. Similarly, we do not know if the
municipalities improved the quality of their services after the mergers - for
example by reducing the class sizes or hiring more nurses and doctors.
However, equally likely is that our results can be explained by permanent
problems in streamlining the municipal organizations after the merger. Often,
the municipal employees have considerable political power in the merger pro-
cess, and bargaining between interest groups is required in order to carry out
the process. All this may create frictions in the municipal organization that can
cause organizational slack. In any case, our results suggest that mergers do not
provide an easy solution to the economic problems of the public sector at the
local level.
REFERENCES
Andrews, R. and G.A. Boyne (2012) “Structural Change and Public Service
Performance: The Impact of the Reorganization Process in English Local
Government”, Public Administration, 90(2): 297–312.
Becker, S. and A. Ichino (2002) “Estimation of average treatment effects
based on propensity score”, Stata Journal, 2(4): 358 – 377.
Municipal Mergers in Finland
165
Blom-Hansen, J. (2010): Municipal Amalgamations and Common Pool Prob-
lems: The Danish Local Government Reform in 2007. Scandinavian Political
Studies 33(1): 51–73.
Borge, L.-E. and J. Rattsø (1993) “Dynamic responses to changing demand: A
model of the reallocation process in small and large municipalities in Nor-
way”, Applied Economics, 25(5): 589-598.
Dehejia, R. and S. Wahba (2002) “Propensity score-matching methods for
nonexperimental causal studies”, The Review of Economics and Statistics
84(1): 151-161.
Dollery, B. E. and E. Fleming (2006), “A Conceptual Note on Scale Econo-
mies, Size Economies and Scope Economies in Australian Local Govern-
ment”, Urban Policy and Research, 24(2): 271–82.
Fox W. F. and T. Gurley (2006), Will Consolidation Improve Sub-National
Governments? Policy Research Working Paper 3913, World Bank.
Hanes, N., (2003) Spillovers, amalgamations and tactical redistributions, Ph
D. thesis, Umeå Economic Studies No. 604, Umeå Universitet.
Heckman, J., Ichimura, H. and P. Todd (1997) “Matching as an econometric
evaluation estimator: Evidence from evaluating a job training programme”,
The Review of Economic Studies 64(4): 605-654.
Hinnerich, B. (2009): Do Merging Local Governments Free Ride on Their
Counterparts when Facing Boundary Reform? Journal of Public Economics
93: 721-728.
Jordahl, H. and Lian, C-Y. (2010): Merged Municipalities, Higher Debt: On
Free-riding and the Common Pool Problem in Politics, Public Choice 143:
157-172.
King, D. (1984) Fiscal Tiers: The economics of multi-level government,
George Allen & Unwin.
Drew, J., Kortt, M. and Dollery, B. E. (2013) “Economies of Scale and Local
Government Expenditure: Evidence from Australia”, Administration and Soci-
ety online volume December 2012.
Lago-Peñas, S. and J. Martinez-Vazquez (eds.) (2013) The Challenge Of Local
Government Size: Theoretical Perspectives, International Experience and Pol-
icy Reform, Edgar Elgar.
Moisio & Uusitalo
166
Lechner, M. (2002) “Some practical issues in the evaluation of heterogeneous
labour market programmes by matching methods”, Journal of the Royal Sta-
tistical Society Series A165(1): 59–82.
Liner, G. H. (1992) “Annexation Impacts on Municipal Efficiency”, The Re-
view of Regional Studies 22:75-87.
Liner, G. H. (1994), “Institutional Constraints, Annexation and Municipal Ef-
ficiency in the 1960’s”, Public Choice 79: 305-323.
Mehay, S. (1981) “The expenditure effects of municipal annexation”, Public
Choice 36:53-62.
Ministry of Finance (2013), Kuntien tehtävien kartoitus, Ministry of Finance
publications 2/2013, Helsinki, Finland.
Moisio, A. (2002) Essays on Finnish municipal finance and intergovernmental
grants, Doctoral thesis, Government Institute for Economic Research (VATT)
Research reports 93. Helsinki.
Moisio A., H. Loikkanen and L. Oulasvirta (2010) Public services at the local
level - The Finnish way, VATT policy Reports 2, Government Institute for
Economic Research, Helsinki 2010.
Nelson, M. A. (1992) “Municipal amalgamation and the growth of the local
public sector”, Journal of Regional Science 32:39-53.
Oates, W. (1972) Fiscal Federalism, Harcourt Brace Jovanovich, Inc.
Oates, W. (1985) “Searching for Leviathan: An empirical study”, The Ameri-
can Economic Review 75(4):748-747.
Oulasvirta, L. (2007), Real and perceived effects of changing the grant system
from specific to general grants, Public choice 91(3): 397-416.
Reingewertz, Y. (2012), Do municipal amalgamations work? Evidence from
municipalities in Israel, Journal of Urban Economics 72: 240–251.
Rosenbaum, P. and D. Rubin (1983), “The central role of the propensity score
in observational studies for causal effects”, Biometrika 70(1): 41-55.
Slack, E. and Bird, R. (2012), “Merging municipalities: is bigger better?”, in
Moisio, A. (ed.) Rethinking local government: Essays on municipal reform,
VATT Publications 61:83-123, Government Institute for Economic Research,
Helsinki 2012.
