Is a Local Government Free-rider on Arts

Is a Local Government Free-rider on Arts?
-Strategic Cultural Expenditures in Local GovernmentsMiki Suhara (Osaka School of International Public Policy)1
Nobuo Akai (Osaka School of International Public Policy)
Abstract
Under the circumstance of limited local resources, cultural expenditure is often set as a reduction
target. Especially in Japanese aging society, the priority level of cultural expenditure tends to be
low. The present paper examines whether or not prefectures strategically influence each other on
determining cultural spending. By estimating the reaction functions for prefectures’ cultural
expenditures, we have found that there exists free-ride behavior between prefectures’ cultural
expenditures associated with spillover effects of benefits. We have also found that the shorter the
distance from neighbor is, or the larger the neighboring prefecture’s population size is, the larger
the incentive to free-ride becomes. Further, our result have revealed that distance from neighbors
affect more on prefectures’ strategic behaviors in the provision of cultural services than the
population size of neighbors.
Key Words: Free-rider, Strategic substitute, Local cultural expenditures
JEL: H71, H72, H73, H77, Z1, H40
1
Correspondence to: Osaka School of International Public Policy, Osaka University.1-31 Machikaneyama,
Toyonaka, Osaka, 560-0043 Japan Tel:+81-(0)6-6850-5624(Dial-In), E-Mail [email protected]
1. Introduction
Following the trend toward decentralization of government, it is often argued that each
municipality needs to shape its own local identity. In recent years, to promote local development,
some municipalities regard cultural activities or heritages as cultural capital to be invested, and
launch city planning utilizing them2. On the other hand, under the circumstance of serious financial
difficulties in local municipalities, policy makers tend to set cultural budgets as reduction targets
(Edagawa,2007; Uehara,2009). The controversial issue related to “Biwako-Hall” occurred in March
2008 would be an example. This is a situation that Shiga prefectural assembly discussed further
cuts in the hall’s management cost for refinancing welfare budget in which reduction has already
been decided. Although the budget amendment act weren’t submitted finally, as Soda (2009) points
out, this issue led to the argument in the assembly representing “cultural expenditure” versus
“welfare expenditure”.
Since cultural goods and services are quasi-public goods with non-excludability, it is not possible
to prevent people from gaining access to them and enjoying the benefits. The inhabitants in
“prefecture A” can visit the cultural facilities managed by neighboring “prefecture B”, and also
participate in the cultural events held in “prefecture B”. When we give an example regarding with
the above mentioned “Biwako-Hall” issue, 54 out of 100 visitors were from other prefectures3.
Reflecting such a situation, 67% of 26,000 signatures against the budget reduction which were
submitted to the governor and the chairperson were from citizens in other prefectures4. Such
possible spillover effects due to the fact that individuals are mobile among jurisdictions may lead
each actor to take strategic behaviors in the decision of cultural expenditures. Namely, a certain
prefecture tends to have an incentive to free-ride on the supply by other prefectures. Especially in
Japan, this incentive of free-riding is expected to be high since the priority of cultural expenditures
is likely to be low because of serious local financial situations even though no one denies the
importance of it.
2
Agency for Cultural Affairs supports the municipalities which challenge local revitalization utilizing the creativity
of arts and culture. From 2007, the Agency has adopted a policy of awarding the municipalities with notable
achievements. (Agency for Cultural Affairs “Cultural Policy Now-arts and culture creative cities”
http://www.bunka.go.jp/ima/souzou_toshi/index.html)
3
From the survey in 2006. Asahi Shinbun, 2008. Biwako-hall in trouble 100milion for Turandot or welfare?
(Biwako hall ni namikaze itiokuenn Turandot ka fukushi ka), Newspapar article om March 17, 2008
4
Asahi shinbun, 2008, The shock of “Biwako” hits Opera(biwako no shogeki opera osou), Newspaper article on
April 17, 2008
1
As for the cultural policy in Japan, Uehara (2009) surveyed current financial support systems
designed by the central government. Kanetake (1986), Neki (2007), Edagawa (2007) and
Yoshimoto (2008) capture the structure and scale of cultural expenditures provided by central and
local governments through time. However, there is no study -to the best of our knowledge- which
analyzes the differences and transitions of local cultural expenditures from an econometric
perspective.
Thus, in this paper, we attempt to investigate the determinants of local cultural expenditures
focusing on the strategic interdependence theory between governments. We examine whether
prefectures’ cultural expenditures are strategically manipulated by those of neighboring prefectures,
further, whether the distance from neighbors or the size of them could be factors which influence
the strategic behaviors. To explore the issue of intergovernmental strategic behaviors resulting from
spillover effects is important from the perspective of reconsideration of role-sharing between
central and local governments on provision of cultural services.
The remainder of this paper is structured as follows. The next section provides a theoretical
framework of strategic interdependence among local governments, and shows how it can be applied
to the case of local cultural expenditures in Japan. Section 3 reviews previous studies. Section 4
presents the data and empirical model. The hypotheses and estimation results are detailed in section
5 and 6, respectively. Finally, section 7 concludes.
2. Theoretical framework
In this section, we propose the theoretical framework, based on Brueckner (2003), to discuss
clearly the relationship between the strategic behavior and the characteristics of the utility function,
by focusing on the effect of the spill-over effects on the marginal rate of substitution.
The benefits of public goods supplied by one jurisdiction may spill over to inhabitants in
neighboring jurisdictions due to the externality. Suppose that municipality i determines its public
expenditure level z i . By taking spillover effects into account, the level of z i can also be directly
influenced by the level in neighboring municipalities, z i . The utility function of representative
individual living in municipality i is determined as follow.
U  u i (c i , z i , z  i ; X i )
2
(1)
where ci
is i ’s consumption of private goods, z i is public goods supplied by other
municipalities and X i is local characteristics of i . The budget constraint when i determines z i by
maximizing equation (1) is represented as below.
ci  pi z i  Yi
(2)
where p i is price of public goods z i and Yi is total income in municipality i . Therefore, an
optimization problem of i is set up as below.
Max
ci , zi
U  U i (c i , z i , z  i ; X i )
ci  pi z i  Yi
Subject to
The marginal rate of substitution between ci and zi, defined as Ri , is derived by the first order
condition:
Ri (Yi  pi z i , z i , z i ; X i ) 
By differentiating equation (5) by z i and z i ,we obtain
R
i
zi

U zi
U ci
 pi
 
(3)
.
 pi Rcii dzi  Rzi i dzi  0
.
(4)
Therefore, the influence towards i when other municipalities increase the supply of public goods in
addition is,
dz i
1
 i
Rzi i
i
dz i Rzi  pi Rci
From Rzi  0 and Rci  0 ,
i
i
(5)
.
1
becomes positive. Equation (5) indicates that the
R  pi Rcii
i
zi
i
direction of i’s response directly depends on the sign of i ’s marginal rate of substitution, i.e. Rzi .
Municipality i’s behaviors can be classified as follows. .
(a) When R z i  0 , it yields
dz i
 0 . The behavior of municipality i is strategic substitute.
dz i
(b) When R z i  0 , it yields
dz i
 0 . The behavior of municipality i is independent.
dz i
(c) When R z i  0 , it yields
dz i
 0 . The behavior of municipality i is strategic compliment.
dz i
i
i
i
3
Thereby, if a test of null hypothesis that
dz i
 0 is rejected, we can demonstrate that spillover
dz i
effects influence Ri ( R z i  0 ) and municipality i reacts strategically.
i
As specific examples, we can image the following two cases which form different strategic
behavior toward z-i. First, when z i and z i are regarded as homogeneous within municipality i, the
utility function is written as U  U i (ci , z i  z i ; X i ) . In this case, the sign of R zi becomes the
i
same as that of Rzi , i.e. R z i  0 . This implies strategic substitute relationship between
i
i
municipalities, in other words, municipality i may free-ride on z-i (
dz i
 0 ), corresponding to the
dz i
case (a). Second, when z i is weakly separable from the consumption within municipality i, the
utility function is written as U  U i (v(ci , z i ), z i ; X i ) . In this case, R z i equals to zero. This
i
represents the behavior of municipality is independent (
dz i
 0 ), corresponding to the case (b).
dz i
This is because even though municipality i receives spillover effects from other municipalities, this
doesn’t effect on Ri .
Applying above theoretical framework to local cultural expenditures, the following logic can be
considered. Cultural goods and services are quasi-public goods with non-excludable nature and
have positive externalities. Thus, for instance, an art exhibition held in a museum in prefecture A is
open not only to inhabitants in A, but also to that in neighboring prefecture B. The benefits of
subsidization which lowers the entrance fee of local museum in A can be also enjoyed by residents
of B. Further, the cultural organizations in one prefecture may not be all that different from those
that are organized in another neighboring prefecture (Werck et al. 2008). This indicates that it is
likely to be indifferent for inhabitants to demand cultural services either “at home” or “outside
home” for habitants. Such spillover effects of cultural services with high substitutability may bring
a prefecture to free-ride on the supply by its neighboring prefectures, through influencing on its
marginal rate of substitution.
We can also expect a positive spatial interdependence just as the case of (c). Here, the behavior of
municipality i is strategic compliment represented by expenditure competition. One theoretical
approach which causes expenditure competition is “reference point effect”. This argument starts
4
from the idea that people often resort to relative performance evaluations (Werck et al.,2008).
Voters evaluate the policies of their own prefectures comparing them with the similar policies taken
by neighboring prefectures. This relative evaluation whether the policies in their prefectures are
better or worse than those in neighbors brings voters additional utility besides the direct utility from
the policies themselves. This additional (dis)utility is called “transaction (dis)utility” (Ashworth
and Heyndels,2000). Applying this idea to cultural expenditures, when cultural events are often
taken place in neighboring prefectures, absence of these events in one’s own prefecture makes
voters feel worse and generates transaction disutility. Therefore, an additional increase in cultural
spending in neighboring prefectures put pressure upon policy makers to increase it in one’s own
prefecture too, which results the situation of expenditure competition. The reference point effect
also brings incumbent politicians who wish to be re-elected an incentive to mimic neighbor’s policy.
If voters judge the policies of their own prefecture worse than those of neighbors, they can choose
not to vote for these politicians. To avoid such a situation, policy makers might follow the same
way of neighbors. This kind of strategic behavior is generally called “yardstick competition”
(Besley and Case,1995).
3. Previous studies
The theoretical framework of strategic interdependence among governments is argued in Besley
and Case(1995), Brueckner(2003) and Revelli(2005). In the United States and Europe, a number of
empirical studies have already been reported. For instance, Case et al. (1993) find that U.S state
government’s per capita expenditures are positively and significantly affected by its neighbor’s
expenditure levels. They find that a one dollar increase in a state’s neighbor’s spending induces an
increase of 70 cents in the state’s own spending. Strategic compliment reaction among governments
is also presented in Figlio et al. (1999), who find the evidence of spillovers in welfare expenditures
in U.S states. Redoano (2007) uses a panel dataset of European countries over the period
1970-1999 and analyses possible different strategic behaviors. She finds that there exist fiscal
interdependencies consistently with the literature on tax and yardstick competition. In Japan,
Sugahara and Kunizai (2005) estimate reaction functions for public expenditures disaggregated at
prefecture level and show that prefectures behave strategically. They find that intergovernmental
5
reactions tend to be negative with respect to the expenditures such as police or education.
As for the field of cultural expenditures, there are Lundberg (2006) and Werck et al. (2008) but
not any in Japan. These studies deal with the level of municipalities’ cultural spending in Sweden in
Belgium, respectively. Table 1 is the summary of these two previous researches. A different
approach is taken in the point how “neighbor” municipalities are defined. Lundberg(2006) defines
them based on travel time by car between municipal centers (less than 300 minutes,200 minutes
and 100 minutes) and find it statistically significant that the free-ride behavior caused by spillover
effects occurres among municipalities in Sweden. In a word, an additional increase in cultural
spending in neighboring municipality decreases that in own municipality. Moreover, it is pointed
out that the shorter the distance between municipalities, the lower the moving cost, which raise the
incentive of free-ride. On the other hand, Werck et al.(2008) assume the municipalities which share
the geographical borders as neighbors, examining if the size of neighbors can affect to the
intergovernmental strategic behaviors. As a consequence, it is found statistically significant that
there exist an expenditure competition among municipalities in Belgium. That is, an additional
increase in cultural spending in neighboring municipality induces that in own municipality. A point
to be emphasized here is that such policy interdependence occurs asymmetrically depending on the
size of neighbors. The size of neighbors do provoke expenditure competition, however in the case
that the neighbors are quite large (they used the term, “central places”), the competition doesn’t
occur. While if the neighbors are small (“non-central places”), the possibility of competition
becomes high.
These two studies come down to the opposite conclusions. Then, in which way intergovernmental
strategic reactions would be taken among Japanese prefectures? This is the first attempt to analyze
the determinants of cultural expenditures at local government level in Japan. Further, our present
paper contributes these two previous studies by considering each prefecture’s fixed characteristic.
Werck et al.(2008) doesn’t fix local characteristics due to the limitation of cross-sectional data.
Although Lundberg(2006) uses panel data, the estimation remains in including regional area
dummies, and fails to control for fixed effects completely.
6
Table1. Summary of previous studies
Authors
(published year)
Data
Lundberg J
(2006)
Sweaden
Municipalities
1981-1990
Werck et al.
(2008)
Belgium
Municipalities
2002
Obs.
Estimating
Method
2,760
spatial
SUR
304
SAR
Dependent Variable
Recreatianal and
Cultural expenditures
per capita
Weights
Strategic
Behaviors
travel time by car
(less than 100min, 200min and
300min)
1. average
Cultural expenditures 2. population
per capita
3. population*central places
4. population*non-central places
free-riding
expenditure
competition
Source: Author
4. The data and Empirical Model
4-1. the Data
In Japan, apart from preservation cost of cultural properties, local governments are the main
providers of cultural expenditures. Specifically, in 2007, almost 90% of total amount
(approximately 350 billion yen) was provided by local governments, especially by municipalities5.
Cultural expenditures without preservation cost of cultural properties in Japan are grouped into
following three categories, the costs for events and research, for managing cultural facilities, and
for constructing cultural facilities. Hereinafter, we define “the cost for events and research” as
“CER”, “the cost for managing cultural facilities” as “CMCF”, and total of them as “TOTAL”.
CER includes (1) culture and art related project costs -such as art, music or movie festivals held by
prefectures, planning and research of culture and art promotion programs, awarding, (overseas)
training of domestic artists etc, (2) subsidies for private organization of arts and culture, and (3)
subsidies for municipalities within relevant prefecture. CMCF corresponds to expenditure on
administrating prefectural museum or cultural hall, and operating events.
Due to the data availability, we use panel data of cultural expenditures at only prefectural level. In
this paper, we exclude the cost for preserving cultural properties. This is because large part of this
cost is subsidized by central government, it seems to be unlikely for prefectures to take strategic
behaviors in the decision to allocate budget for protecting cultural properties. Further, since we
concentrate on the ordinary cultural expenditures at prefectural level, we don’t take construction
5
While large part of municipal cultural expenditures is allocated for maintaining cultural facilities, prefectures tend
to concentrate on spending on non-infrastructural projects such as grants for private organizations or cultivation of
arts viewers (Edagawa, 2007)
7
cost of cultural facilities into account. Finally, since we define “neighbors” in terms of prefectures
which share borders in a geographic sense, we exclude the data of Hokkaido and Okinawa
prefectures from an analysis. Therefore, prefectures’ cultural expenditures to be estimated in this
paper are TOTAL, CER and CMCF over the period of 1997-2007. The total number of observation
is 495 (45 prefectures over 11 years).
As dependent variables, we use absolute amounts in natural logarithm. Although it’s often argued
that standardization of the variable is necessary to capture the difference of expenditure level on
scale, our empirical model allows including each prefecture dummy variable, which enables us to
control the possible general effects caused by scale more precisely than standardization. With
regard to above mentioned previous studies, Werck et al.(2008) couldn’t fix local characteristics
due to the limitation of
cross-sectional data. Although Lundberg(2006) uses panel data, the
estimation remains in including regional area dummies, and fails to control for fixed effects
completely. Graph 1 shows a scatter plot between year averages of residuals after regressing a
dependent variable (TOTAL) on prefecture dummies alone and year averages of population. From
this graph, it seems that no positive correlation is observed between them, implying that scale
doesn’t affect on the level of cultural expenditures anymore.
Graph 1. Relationship between population and residuals after regressing TOTAL on
-1.00e-08
-5.00e-09
Residual_m
0
5.00e-09
1.00e-08
prefecture dummies (year averages)
13
14
15
lnppl_m
16
17
Source: Author
We use following control variables based on the previous studies. As the variables related to
habitancy besides population and population density, we include the share of population above
8
65-years old and under 19-years old as measures for the aged and the young generations
respectively, and the share of male population6. As economic variables, we include prefecture
income per capita and the share of the third industry, and the unemployment rate7. The descriptive
statistics of variables are reported in Table 2. Note that all economic variables are deflated by the
gross prefecture products using 2000 as the base year.
Table 2. Descriptive Statistics
Variable
Cultural expenditure (log)
TOTAL
CER
CMCF
Population (log)
Pop. density
% over 65 year
% under 19 year
% male population
Unemployment rate
Prefecture income per capita
% the third industry
Obs.
Mean
S.D.
Min.
Max.
495
495
495
495
495
495
495
495
495
495
495
13.9
13.6
12.4
14.7
648.7
20.0
20.4
48.5
4.1
8.0
71.4
0.7
0.8
1.1
0.5
1105.7
3.2
1.5
1.0
1.0
0.1
7.3
11.4
10.3
9.8
13.7
90.2
10.8
15.9
47.0
1.7
7.7
49.7
16.4
16.1
15.6
15.8
5652.5
27.6
24.5
51.0
7.7
8.5
92.7
Source: Author
4-2. Empirical Model
As explained in section 2, prefecture i’s cultural expenditures in year t ( EXPit ) is a function of its
neighboring prefecture j’s expenditures ( NEIGH
it
). The linearly transformed estimation model is
defined as follows.
EXPit   ij NEIGH jt  X it β  uit
(6)
j i
where i represents prefectures (i=1,...,n), t year (t=1,...,T),Xit is a vector of prefectures’ social
–economic control variables and β,θ are unknown parameters to be estimated. Xit contains
population (in natural logarithm), population density, the share of inhabitants above 65-year old,
6
The source of these variables is “The Basic Resident Register” (Ministry of Internal Affairs and
Communications).
7
The source of the former two variables is Annual National Accounting” (the Cabinet Office) and that of the latter
is “Labor Force Survey” (Ministry of Internal Affairs and Communications).
9
under 19-year old and male, unemployment rate, prefecture income per capita (in natural
logarithm) and the share of the third industry.
Since there are too many parameters θij to be estimated, we cannot estimate equation (8) directly.
Instead, following liner specification is usually employed (Revelli,2005; Redoano,2007).
EXPit  NEIGHit  Xit β  t t γ  p i α  uit
(7)
NEIGH it   wijt EXP jt
(8)
j i
where p i , t t are vectors of prefecture and year dummy, uit is an error term, α, β, γ ,  are
unknown parameters to be estimated. Here, wijt are exogenously chosen weight, row-normalized so
that
w
j i
ijt
 1 , and wijt  0 if prefecture j is nor a neighbor or just j  i . In this paper, we
assume three different kinds of weight, (i.e.(1)average expenditures of neighbors which simply
shares the borders each other, (2)geographical distance between neighbors, (3)population size of
neighbors. In what context these weights should be used will be explained later in section 5, the
hypotheses. If the coefficient of interaction term NEIGH
it
is not significantly equal to zero, we
can assert the possibility of strategically interdependence relationship among prefectures on
determining the level of cultural spending.
As Anselin(1988) indicates, however, there are two main econometric problems raised by the
presence of dependent variable on the right hand side of estimating model (7). First, if prefectures
do react each others’ expenditure levels simultaneously, NEIGH
it
is endogenous and correlated
with the error term u it . In this case, OLS estimators hold neither unbiassness nor consistency
anymore. Second, if neighbors’ localities are subject to correlated random shocks, there may exist
spurious causal influence between prefectures’ spending choices. That is, if we omit independent
variables that are spatially dependent, these variables fall in uit, this leads misspecification of
estimating equation (7)
Both problems can be solved by applying IV method with appropriate instrumental variables8. We
now need to consider instrumental variables which are correlated with NEIGH
8
it
but uncorrelated
As demonstrated in Kelejian and Prucha (1998), IV method yields a consistent estimation of θ even under the
presence of spatial error dependences in error term. This approach is followed by Heyndels and Vuchelen (1998),
Foucault et al. (2008) and Redoano(2007).
10
with u it . One potential source of these instrumental variables are the weighted average of neighbors’
control variables (WX), as suggested in Redoano(2007) Foucault (2008) and Werck et al.(2008). To
judge the validity of the instruments, we conduct the Kleibergen and Paap rk LM statistic for
underidentification test and Hansen-J statistic for overidentifying restriction test and pass all tests.9
(each test statistic are reported at the bottom of Table3-Table5). Finally, we correct the additional
issues such as heteroskedasticity and serial correlation in the error terms by gaining robust
estimators.
5. Hypotheses
In this section, we propose the hypotheses to test. As described in section 2, two theoretical
frameworks, free-rider and expenditure competition, are discussed. When applied to Japanese
cultural expenditure, it’s quite unlikely that the level of prefectures’ cultural spending can be the
main issue of elections. Therefore, it might be focused on the spillover effects. Now, following
three hypotheses are set up based on free ride strategic behavior based on the spillover effects.
<Hypothesis 1>
“There exists free-ride phenomenon among prefectures’ cultural expenditures”
This hypothesis implies that the sign of NEIGH
it
‘s coefficient in equation (9) is negative. If it
becomes positive, expenditure competition will be expected among prefectures. Note that precise
definition of NEIGH
it
is required depending on the hypotheses to be tested. In the case of
<hypothesis 1>, neighbors are defined as two prefectures that share a border. Thus, the weights
( wijt ) are assumed to be uniform, which is wijU  1 / n for all i,j, where n is the number of
neighbors of prefecture i .
As it is often argued that distance plays a major role whether or not to visit cultural events
(Verhoeff,1992, De Graaff et al.2007; Werck et al.), the distances from neighbors reflect the
degree of accessibility to cultural goods and services. Assume if similar cultural events take place
9
In every model, we reject the null hypothesis of the former test, implying there is no issue of weak instruments.
As for the latter test, we accept null hypothesis, indicating the instruments are exogenous to the second-stage
regression.
11
both in prefectures B and C, and the latter is closer to prefecture A than the former, then residents in
A would choose to visit events in C which is more accessible. Thus, policy makers in A would be
motivated to free-ride more on cultural expenditures in B than that in C. This leads to following
<hypothesis 2>.
<Hypothesis 2>
“The shorter the distance from neighbor is, the higher the incentive to free-ride becomes”
In this case, wijt is constructed such that,
wijD 
1 / d ij
 (1 / d
ij
)
j
where d ij is the geographical distance between capitals of prefecture i and prefecture j. Here also, i
and j are neighbors that share a border each other.
Now, we construct a weighting matrix based on neighbors’ population. Where the population size
is large, the demand population of cultural goods is large, too. Furthermore, regardless of private or
public, it is expected the agglomeration of cultural institutions including so called ‘Superstar’
museum which are well-known and serve as magnets for attracting more visitors (Blau and
Hall,1986; Frey,1998). Marginal increase in cultural expenditure in a large populated prefecture can
be more productive comparing to the same increase in small prefectures.
It is possible that cultural images and information through word of mouth spill over from large
prefectures to neighbors. Thus, the larger the neighboring prefecture is, the higher the incentive to
free-ride becomes since residents can move there to enjoy the benefits. This leads to <hypothesis
3>.
<Hypothesis 3>
“The larger the neighboring prefecture is, the higher the incentive to free-ride becomes”
In this case, wijt is constructed such that,
wijtP 
pop jt
 pop
j
12
,ji
jt
where popjt is population of j which share each border with i.
6. Empirical results
The estimation results are reported from Table3 to Table5. Each corresponds to the case using
average weighted matrix (Table3), geographical distance weighting matrix (Table4) and population
size weighting matrix (Table5), respectively.
To test <hypothesis 1>, let us take a look at Table3. The coefficient of neighbors’ expenditure is
negatively significant, indicating that prefectures’ cultural expenditures are negatively correlated
with their neighbors’ spending. Therefore, it’s demonstrated that there exists free-ride behavior
between prefectures’ cultural expenditures associated with spillover effects of benefits. The result
remains the same when we regress cultural expenditures separately for CER and CMCF (see Model
(2) and Model (3)).
We make some brief remarks on the results of some control variables. Probably because most of
local specific effects were absorbed by prefecture dummies ( p i α ), our control variables could get
less significant results10. Still, the coefficient of the rate of elder population appears negative in
model (1) (3), while it’s insignificant in model (2). In general, the elder’s opportunity costs of
consuming cultural activities are considered to be low (Schulze and Ursprung, 2000), thereby the
sign tends to appear positive. However, our result implies the possibility that for the prefecture with
high rate of elder population, cultural expenditures- especially operating cost of cultural event and
research- can be alternative with social security expenditures. In rapidly aging Japanese society,
social security cost is an urgent issue. This trend may put pressure on policy makers to give
preference to social security expenditures instead of cultural expenditures.
Next, as for the unemployment rate, the coefficients are significant with negative sign for cultural
expenditures in all models. Since demand for cultural goods is income elastic in general, the higher
unemployment rate is, the lower demand on cultural expenditures is expected to be. Moreover, just
as the case of elder population rate, policy makers might be driven to expense more on welfare than
culture.
10
Although we also estimated the model without regional dummies, F-test accepted the model which includes them.
Moreover, centered R2 of the estimation models without regional dummies remained at lower level (0.3~0.4).
13
Finally the share of male population is negative and significant in model (1) and (2). For this
reason, it’s statistically justified that female has more tendencies to appreciate arts than male
(Nagayama,1998). According to prefectures’ statistics11, the percentage of male population who
have attended in cultural activities in the past year such as listening to classic music, going to
museums, and appreciating performing arts and dance are 6%, 15%, and 9%. On the other hand,
percentages of female attendance are 12%, 22% and 19% respectively. There exists a relatively
large difference between sex.
Now, let us now turn to the test of <hypothesis 2>. From Table4, the coefficient of neighboring
expenditure also appears negative, which proves the high possibility of free-riding to cultural
spending in the closer prefecture. This implies that residents can easily move to the closer
prefectures and take opportunities to enjoy cultural goods and services provided there.
With respect to <hypothesis3>, from Table5, we can see that the incentive to free-ride becomes
larger when neighboring prefecture’s population size is larger. One possible interpretation of this
result is that spillover effects from larger prefecture not only can be direct benefits but also cultural
image or information through word of mouth.
Finally, Table6 compares the coefficients of neighbors’ expenditures of Table4 with those of
Table5. It shows that the former is always larger than the latter in all models. Therefore, distance
from neighbors can be an important factor which affects strategic behavior among prefectures.
11
Ministry of Internal Affairs and Communications “Basic Survey for Social Life (2006)” > Results based on
survey sheet A > living activities >leisure and hobby
(http://www.e-stat.go.jp/SG1/estat/List.do?bid=000001008009&cycode=0)
14
Table 3. Estimation Results (1)
Independent Variables
Dependent Variable: Cultural expenditures
Model (1)
Model (2)
Model (3)
TOTAL
Coefficient
-0.796
*
Neighbors' cultural expenditures
(0.449)
-1.140
Population
(3.017)
-0.001
Pop. Density
(0.001)
-0.180
**
% over 65 years
(0.089)
0.037
% under 19 years
(0.166)
-0.708
*
% male population
(0.413)
-0.129
*
Unemployment rate
(0.070)
-0.161
Pref. income per capita
(1.098)
-0.028
% the third industry
(0.030)
yes
Year dummy
yes
Prefecture dummy
495
Observations
0.72
Centered R2
31.43***
F-statistic
22.09***
Kleibergen and Paap rk LM statistic
0.589
Hansen J statistic (p-value)
CER
CMCF
Coefficient
Coefficient
-0.730
*
-0.865
*
(0.427)
(0.510)
1.883
-1.078
(3.666)
(5.324)
-0.005
***
0.002
(0.002)
(0.002)
-0.091
-0.357
**
(0.104)
(0.160)
0.156
-0.216
(0.210)
(0.192)
-1.089
**
0.804
(0.480)
(0.716)
-0.185
**
-0.198
*
(0.083)
(0.110)
-0.371
-0.719
(1.026)
(2.385)
-0.030
-0.034
(0.035)
(0.041)
yes
yes
yes
yes
495
495
0.69
0.68
24.98***
30.03***
23.68***
14.23**
0.202
0.554
Note: Robust standard error in brackets, *** significant at 1 % level, ** significant at 5 % level, *
significant at 10 % level
15
Table 4. Estimation Results (2)
Independent Variables
Dependent Variable: Cultural expenditures
Model (1)
Model (2)
Model (3)
TOTAL
CER
CMCF
Coefficient
Coefficient
Coefficient
-1.251
***
-0.868
***
-0.850
**
Neighbors' cultural
(0.464)
(0.331)
(0.399)
expenditures
-3.709
0.628
-2.991
Population
(3.500)
(3.677)
(4.330)
-0.001
-0.005
**
0.003
Pop. Density
(0.001)
(0.002)
(0.002)
-0.183
*
-0.084
-0.358
***
% over 65 years
(0.099)
(0.108)
(0.140)
0.127
0.204
-0.162
% under 19 years
(0.188)
(0.209)
(0.189)
-0.561
-1.039
**
0.899
% male population
(0.438)
(0.496)
(0.688)
-0.118
*
-0.158
*
-0.191
*
Unemployment rate
(0.065)
(0.086)
(0.105)
-1.409
-0.968
-1.278
Pref. income per capita
(1.329)
(1.146)
(2.039)
-0.039
-0.049
-0.039
% the third industry
(0.035)
(0.034)
(0.037)
yes
yes
yes
Year dummy
yes
yes
yes
Prefecture dummy
495
495
495
Observations
0.65
0.67
0.68
Centered R2
23.99***
20.47***
14.91***
F-statistic
19.65***
29.268***
18.25***
Kleibergen and Paap rk LM statistic
0.927
0.270
0.565
Hansen J statistic (p-value)
Note: Robust standard error in brackets, *** significant at 1 % level, ** significant at 5 % level, *
significant at 10 % level
16
Table 5. Estimation Results (3)
Independent Variables
Dependent Variable: Cultural expenditures
Model (1)
Model (2)
Model (3)
TOTAL
CER
CMCF
Coefficient
Coefficient
Coefficient
-0.544
***
-0.515
**
-0.497
**
Neighbors' cultural expenditures
(0.203)
(0.230)
(0.234)
-0.679
2.511
-1.726
Population
(2.868)
(3.451)
(4.157)
-0.001
-0.005
**
0.003
Pop. Density
(0.001)
(0.002)
(0.002)
-0.148
*
-0.068
-0.308
**
% over 65 years
(0.084)
(0.099)
(0.131)
0.015
0.146
-0.270
% under 19 years
(0.144)
(0.189)
(0.172)
-0.630
-0.986
**
0.603
% male population
(0.419)
(0.485)
(0.619)
-0.112
*
-0.149
**
-0.188
*
Unemployment rate
(0.067)
(0.074)
(0.102)
-0.123
-0.504
0.339
Pref. income per capita
(0.963)
(1.010)
(1.494)
-0.026
-0.029
-0.025
% the third industry
(0.025)
(0.029)
(0.033)
yes
yes
yes
Year dummy
yes
yes
yes
Prefecture dummy
495
495
495
Observations
0.73
0.70
0.70
Centered R2
34.5***
27.00***
15.91***
F-statistic
28.27***
35.60***
49.19***
Kleibergen and Paap rk LM statistic
0.879
0.605
0.503
Hansen J statistic (p-value)
Note: Robust standard error in brackets, *** significant at 1 % level, ** significant at 5 % level, *
significant at 10 % level
Table 6. Comparing the effects of distance with population size
Wights
population
distance
TOTAL
-0.544
***
-1.251
***
CER
-0.515
-0.868
**
***
CMCF
-0.497
**
-0.850
**
Note: Robust standard error in brackets, *** significant at 1 % level, ** significant at 5 % level, *
significant at 10 % level
17
7. Concluding remarks
Under the circumstance of limited local resources, cultural budget is often set as a reduction target.
Especially in Japanese aging society, the priority level of cultural budget tends to be low. The main
purpose of this paper is to examine the prefectures’ strategic behaviors on determining cultural
expenditures associated with spillover effects of benefits. By estimating the reaction functions of
prefectures’ cultural expenditures, we have found the significantly negative relation among them.
This result indicates that a marginal increase in cultural expenditure in neighbors deters cultural
spending in own prefecture. Specifically, since the cultural facilities or events within prefecture A
are open both to inhabitants in A and in its neighboring prefecture B, prefecture B has an incentive
to free-ride on the cultural expenditure provided by A and won’t increase its own spending.
We have also found that the shorter the distance from neighbor is, the larger the incentive to
free-ride becomes. Moreover, our estimation results have revealed that the incentive to free-ride
becomes larger when neighboring prefecture’s population size is larger. We have finally found that
distance from neighbors affect more on prefectures’ strategic behaviors in the provision of cultural
goods and services than the population size of neighbors.
Taking such intergovernmental strategic actions into account, the mechanism of internalizing the
externality of cultural goods and services-such as promoting wide-range cooperation of local
governments- will be required in order to supply them efficiently and effectively.
References
Agency for Cultural Affairs, 2010. Cultural Policy in Japan 2010 edition (in Japanese: heisei 22
nendo wagakuni no bunka gyosei).
Akira, Neki,. 1998. Legal system of cultural policy and cultural budget in Japan (in Japanese:
bunka housei to bunka yosan –housei no taikei to yosan no kouzou-), In: Akira, Neki., (Eds)
Development of Cultural Policy –Promotion of Arts and Culture, and Protection of Cultural
Property- (in Japanese: bunkaseisaku no tenkai –geijyutu bunka no sinkou to bunkazai no
hogo-). Broadcasting University.
Akitoshi, E., Support for local cultural activities –the roles of central and local governments- (in
Japanese:tiiki bunka katudou no sien – kuni to tihoukoukyoudantai no yakuwari-) In: Akira,
18
Neki., (Eds) Development of Cultural Policy –Promotion of Arts and Culture, and Protection of
Cultural Property- (in Japanese: bunkaseisaku no tenkai –geijyutu bunka no sinkou to bunkazai
no hogo-). Broadcasting University.
Anselin, L., (1998). Spatial Econometrics: Methods and Models, Kluwer Academic Publishers.
Ashworth, J., and Heyndels, B. (2000). Reference point effects in local taxation: It all depends on
how you look at it. National Tax Association Papers and Proceedings, pp.335-341.
Besley, T., and Case, A. (1995). Incumbent behavior. Vote seeking, tax setting and yardstick
competition. American Economic Review, 85(1), pp.25-45.
Blau, J. R., and Hall, R. H. (1986). The supply of performing arts in metropolitan areas. Urban
Affairs Quarterly, 22(1), pp.42-65.
Brueckner, J. (2003). Strategic interaction among governments: An overview of empirical studies.
International Regional Science Review, 26(2), pp.175-188.
De Graaf, T., Boter, J., and Rouwendal J. (2007). On spatial differences in attractiveness of Dutch
museums. Free University Amsterdam, mimeo.
Figlio, D., Kolpin, V., and Reid, E. (1999). Do States play welfare games? Journal of Urban
Economics, 46, pp.437-454.
Foucault, M., Madies, T., and Paty, A. (2008). Public spending interactions and local politics.
Empirical evidence from French municipalities. Public Choice, 10, pp.57-80.
Frey, B.S., (1998). Superstar Museums: An Economic Analysis. Journal of cultural economics. 22,
pp.113-125.
Getzner, M. (2002). Determinants of public cultural expenditures: An exploratory time series
analysis for Austria. Journal of Cultural Economics, 26, pp.287-306.
Geys, B. (2006). Looking across borders: A test of spatial interdependence using local government
efficiency ratings, Journal of Urban Economics, 60(3), pp.443-462.
Hajime, K., 1996. Financial issue of cultural policy in Japan (in Japanese: nihon ni okeru bunka
seisaku no zaisei mondai), Disputes of Economics Research and Study, Kyoto University,
Vol.11, pp.19-32.
Heyndels, B., and Vuchelen, J. (1998). Tax mimicking in Belgian municipalities. National Tax
Journal, 51, pp.89-101.
Hirota, S., and Minoru, K., 2005. Empirical analysis of financial expenditures competition, Papers
19
presented at the ninth annual meeting on Japanese Public Choice.
Kelejian, H., and Prucha, I. (1998). A generalized spatial two-stage least squares procedure for
estimating a spatial autoregressive model with autoregressive disturbances. Journal of Real
Estate Finance and Economics, Vol.17, pp.99-121.
Lundberg, J. (2006). Spatial interaction model of spillovers from locally provided public services.
Regional Studies, 60, pp.531-644.
Mitsuhiro, Y., 2008. Reconsidering cultural policy –its expanding role and required paradigm shift(in Japanese: saikou bunka seisaku –kakudai suru yakuwari to motomerareru paradigm shift-).
Nissei Report, Vol.51 pp.37-116.
Redoano, M. (2007). Fiscal interactions among European countries: Does the EU matter? CSGR
working paper series, No.222/07.
Revelli, F. (2005). On spatial public finance empirics. International Tax and Public Finance, 12(4),
pp.475-492.
Sadanori, N., 1998. Arts and Culture, and Gender Statistics, (in Japanse: geijyutu bunka to Gender
Statistics) Journal of Cultural Economics, Vol.1, No.2, pp.1-6.
Schulze, G. G., and Ursprung, H. W. (2000). La Donne e mobile-or is she? Voter preferences and
public support for the performing arts. Public Choice, 102, pp.129-147.
Syuji, S., 2009. Art as a Shared Property of a Community:A Study on the Cultural Policy of Local
Governments (in Japanese: “kyoudouzai” tosite no a-to no arikata wo kangaeru – jititai bunka
seisaku no arikata ni tuite no kousatu-), Publication of faculty of management of Atomi
Gakuen Jyosi university, Vol.8, pp.75-90.
Verhoeff, R. (1992). Explaining differnces in the geographical reach of performances. Journal of
Cultural Economics, 16(2), pp.73-82.
Withers, G. (1979). Private demand for public subsidies: An econometric study of cultural support
in Australia. Journal of Cultural Economics, 3(1), pp.53-61.
Werck, K., Heyndels, B., and Geys, B. (2008). The impact of ‘central places’ on spatial spending
patterns: evidence from Flemish local government cultural expenditures. Journal of Cultural
Economics, 32, pp.35-58
20
Yukiko, U., 2009, Financial support system for arts and cultural activities ( in Japanese:
Geijyutubunka katudou heno zaisei sienn no arikata) National Diet Library ISSUE BRIEF
NUMBER 628.
Newspaper articles and WebPages
Agency for Cultural Affairs “Cultural Policy Now-arts and culture creative cities”
(http://www.bunka.go.jp/ima/souzou_toshi/index.html) (access date: 2010/05/28.)
Asahi Newspaper, 2008. Biwako-hall, In trouble, 100milion for Turandot or welfare? (Biwako hall
ni namikaze itiokuenn Turandot ka fukushi ka), Newspapar article om March 17, 2008
Asahi Newspaper, 2008, The shock of “Biwako” hits Opera(biwako no shogeki opera osou),
Newspaper article on April 17, 2008
Ministry of Internal Affairs and Communications “Basic Survey for Social Life (2006)”
(http://www.e-stat.go.jp/SG1/estat/List.do?bid=000001008009&cycode=0)
2010/05/28.)
21
(access
date: