Download as pdf or txt
Download as pdf or txt
You are on page 1of 19

The Asia Pacific Journal of Arts and Cultural Management is

available at: www.ojs.unisa.edu.au/index.php/ARTMJ


and is produced by the School of Management:
http://business.unisa.edu.au/management
University of South Australia: www.unisa.edu.au
Keywords participation, Survey on Time Use and Leisure Activities, logit model

Abstract This paper examines public participation in arts and cultural activities in
Japan using statistical methods. Utilising the binary regression model (logit model),
the significant demographic and socio-economic factors that determine participation in
arts and culture and their effects, are considered empirically. The model is an
estimate for cultural activities as well as for other entertainment activities. The data
used here are the microdata of the Survey on Time Use and Leisure Activities
conducted by the Statistics Bureau of the Japanese Ministry of Internal Affairs and
Communication. This is the first attempt at estimating such a model for data on
Japanese arts participation, and results, similar to Western countries, are obtained for
determinants of arts participation.

Biography Masaki Katsuura is Professor of Statistics at Meijo University, Nagoya,
Japan. He is a member of the Board of Directors of the Japan Association for Cultural
Economics (JACE) and also an associate editor of Bunka Keizaigaku (Cultural
Economics; journal of the JACE). He was a visiting researcher at the Economic and
Social Research Institute (ESRI), Cabinet Office, Government of Japan, and was also
an expert member of the Statistics Council in Japan. His research interests in cultural
economics include participation (demand), cultural statistics, and cultural industries.


Examining Arts Participation in Japan Using the
Survey on Time Use and Leisure Activities
Masaki Katsuura
Faculty of Economics, Meijo University
Asia Pacific
Journal of Arts &
Cultural
Management
343 Asia Pacific Journal of Arts and Cultural Management Vol. 5 Issue 1 August 2008 pp 343-361 University of South Australia ISSN 1449-1184
Asia Pacific Journal of Arts and Cultural Management Vol. 5 Issue 1 August 2008 pp 343-361 University of South Australia ISSN 1449-1184
Introduction

Rapid economic development in Japan after the Second World War has addressed the
material needs of the population. In recent decades, the focus of demand has shifted to
intangible aspects, especially with reference to the aging society. In this situation, the
importance of participation in various arts and cultural activities has increased, and
hence, a rearrangement of the existing social environment and infrastructure is
required.

To achieve this, it is important to understand the factors affecting participation. For
example, if parents with young children (i.e. pre-kindergarten or preschool age) attend
classical concerts or dance theatres with relatively less frequency, the need to establish
and improve day-care service facilities for children in concert halls and theatres, could
be addressed. Moreover, the relationship between participation/non-participation and
an individuals characteristics may play an important role in marketing for arts and
cultural industries.

Many studies have investigated the important factors influencing participation, in
particular among audiences of performing arts such as classical and popular music.
Baumol and Bowen (1966) were at the forefront of such research and, following their
work, a series of studies analysed which and how much individual characteristics
affected participation in a variety of arts and cultural activities from several perspectives
(see also Throsby & Withers 1979). There exist several methodologies in this research
area, one of the most useful and powerful being multivariate statistical analysis,
particularly binary regression. The representative works in applying binary regression
analysis to cultural economics are the following:

Abb-Decarroux and Grin (1992) estimated a logit model for opera, concert and
theatre attendance data in Geneva.

Lvy-Garboua and Montmarquette (1996) used a probit model for theatre
demand in France.

Prieto-Rodrguez and Fernndez-Blanco (2000) analysed the differences
between attendances at popular and classical music concerts in Spain using a
bivariate probit model.

Gray (1998b) employed a logit model for exploring factors affecting adult visits to
art museums, and Gray (2003) utilized a similar methodology for broader
activities in culture and arts using the Survey of Public Participation in the Arts
(SPPA) in the US (NEA, 2004).

Borgonovi (2004) applied ordered logistic regression to SPPA data on
professional performing arts events for analysing the frequencies of participation.

Seaman (2006) did a general survey of this field.

Thus the important factors, both theoretically and empirically, influencing participation,
are revealed through the above works, i.e. education level, age, arts lessons in
childhood, income, etc. Education in particular is emphasised as one of the most
important factors. Some of the important factors are theoretically explained as capital
(or from the viewpoint of investment), which was proposed by Stigler and Becker
(1977). They introduced the concept of personal cultural capital, acquired by
experience and past consumption, which is utilised in explaining behaviour in cultural
344
316 Asia Pacific Journal of Arts and Cultural Management Vol. 5 Issue 1 August 2008 pp 343-361 University of South Australia ISSN 1449-1184
activities including cultural consumption and participation. For example, participation
increases with age because cultural tastes develop over a long time, which also
explains the importance of arts lessons in childhood and education level. Although the
reason for the difference between male and female arts participation is not obvious, the
causes may be different experiences during childhood e.g. boys tend to participate
more in sports and less in art/music compared with girls. This means that gender might
not affect arts participation when all factors (including past experiences) are controlled.
(See Harris (1996) and Peterson, Hull and Kern (2000) for the effect of gender). There
may be a theoretical explanation for other factors, such as income level, working hours
etc., as for the level of general education for instance.

Previous works in this area have mainly investigated Western countries. This study
empirically examines the significant demographic and socio-economic factors
determining participation in arts and culture in Japan. DiMaggio and Ostrower (1992)
and Borgonovi (2004) indicated that race is one of the determinants of participation,
which means, even if other factors are controlled, ethnicity affects arts participation.
The homogeneity of races in Japan however provides a control for the race factor. The
purpose of this study is to examine whether, with regard to arts participation, people in
an Asian country with similar economic development as Western countries, behave in a
similar way to people in Western countries. If so, there can be confirmation of the
generality of economic theories explaining arts participation. It would be interesting to
see whether the observations made by previous studies relating to the determinants
listed above for Western countries remain valid in an Asian (i.e. non-Western) country.
Arts participation, i.e. attendance at performing arts events, visits to art museums,
playing musical instruments and other entertainment activities are included in the
analysis for comparison.

The data for this analysis use the Survey on Time Use and Leisure Activities (STULA)
conducted by the Statistics Bureau of the Japanese Ministry of Internal Affairs and
Communication. Using the STULA microdata, an estimate of binary regression models
(logit models) is undertaken, using primarily participation/non-participation as the
dependent variable, and personal and household characteristics as independent
variables for the arts and cultural activities surveyed. This research is the first attempt
at applying binary regression to public participation in such activities in Japan. There
may be reasons for not attempting such an analysis in Japan, such as constraints on
data, and this will be discussed in the next section.

The rest of the paper is organised as follows. In the next section, the STULA is outlined,
and the microdata and model used in this paper, are described in the third section. The
fourth section presents the empirical results of logit analysis. The final section makes
some brief concluding remarks.

2. Survey on Time Use and Leisure Activities (STULA)

2.1 Outline of the STULA
The data used in this paper obtained from STULA, is one of the main sources of data
on arts participation in Japan. The purpose of the STULA is to illustrate social life in
Japan from two perspectives: time allocation in a day (time use) and participation in
leisure activities during the past year (Statistics Bureau Japan, 2003). This survey has
occurred every five years, since the first survey was undertaken in 1976.

The time use survey is conducted according to international standards, but the survey
on leisure activities appears to be unique. The leisure activities surveyed are (1)
internet use, (2) studies and research, (3) sports, (4) hobbies and amusements, (5)
volunteer activities, and (6) travel and excursions, with each item classified into many
345
316 Asia Pacific Journal of Arts and Cultural Management Vol. 5 Issue 1 August 2008 pp 343-361 University of South Australia ISSN 1449-1184
activities. Arts and cultural activities, with which this paper is concerned and which
include attending classical or popular music concerts, cinemas, plays, dance theatres,
museums and playing musical instruments, are classified under hobbies and
amusements, as shown in Table 1. Respondents are asked whether or not they
participate in such activities, and if so, the frequency of such activities. Moreover, for
some activities, the purpose and the groups of people undertaking the activity
together, is also asked.

The most remarkable characteristic of this survey is its sample size. In the 2001 survey,
approximately 77,000 households were selected by a stratified two-stage sampling
method, and household members in the sample amounted to almost 200,000 people
(ten years old and above). Because of such a large sample size, many types of
aggregation are possible with a certain degree of statistical reliability, even in multiple
cross-tabulations and aggregated by prefectures and other individual and household
properties. Therefore, the results of the survey are valuable for examining factors
influencing participation in arts and culture. See Statistics Bureau Japan (2003) for the
survey items in the STULA.































346
activities 1996 2001 2006
Watching sports games
*

Watching works of art
*

Watching plays, vaudevilles and dances
*

Watching movies
*

Going to classical music concerts
Going to popular music concerts
Listening to music on records, tapes or
CDs

Watching movies on video or LDs
Playing musical instruments
Traditional Japanese music
Japanese folk songs
Chorus, vocal music
Japanese dancing
Western style dancing or social dancing
Calligraphy
Japanese flower arrangement
Japanese tea ceremony
Painting or sculpture
Ceramics or handicrafts
Photographs
Making movies by video camera or 8mm
camera

Writing poems, novels, haiku, etc


Reading books as hobbies
: surveyed, blank:not surveyed
Source: Statistics Bureau, the STULA
* excluding television programs and video
Table 1 Cultural Activities Classified in Hobbies and
Amusements in the STULA (1996, 2001, 2006)
316 Asia Pacific Journal of Arts and Cultural Management Vol. 5 Issue 1 August 2008 pp 343-361 University of South Australia ISSN 1449-1184
2.2 Advantages and disadvantages of the STULA

There are many advantages in examining arts participation using the STULA. The
sample size is one of the most attractive advantages in analysing both the aggregated
published tables and the microdata. The SPPA surveys adults (eighteen years old or
above), whereas the STULA includes respondents ten years old or above, which is also
useful since the effects of attending primary and secondary school can be considered.
Moreover, the relationship between time use and participation can be analysed
because two types of survey are conducted among the same respondents (Katsuura
2005). The STULA surveys participation in many activities; not only in cultural activities
but also in sports, entertainment, and volunteer activities etc. This enables a
comparison of the determinants of cultural and other activities and their effects, and an
examination of the relationship between activities.

However, the analysis is limited to data at an aggregate level, with a few exceptions;
Katsuura (2003, 2006 ) tried to separate the effects of income and education by using a
cross-tabulated table aggregated by the microdata of the STULA, and shows the
possibility of a binary regression model using a very primitive model. This was probably
because the availability of official statistics microdata is restricted and available only
under the non-statistical purposes access system. In Japan, the use of microdata
collected by the government, is subject to the Statistics Law. This says no person is
permitted to use microdata, except via the non-statistical purposes access system.
Under this exceptional system, applicants wishing to use microdata have to be
approved by the Minister for Internal Affairs and Communication, which means the
process is very complicated and time-consuming. Even if one is approved to use
microdata, the usability is very constrained. For example, aggregation and statistical
analysis must be done according to the research plan clarified in the application form.

This is a general problem underlying the STULA and other surveys conducted by the
government. Another disadvantage of the STULA in analysing arts participation arises
from the purpose of the STULA, which is to clarify both time allocation and participation
in leisure activities including cultural activities. This means the survey, unlike SPPA,
does not focus on participation in arts. One of the most important factors in arts
participation is whether people attended arts classes in childhood or not, as suggested
in Gray (1998b) and Morrison and West (1986). This question is, of course, on the
questionnaire in the SPPA, but not in the STULA. Neither does the STULA ask about
the education level of the parents.

In addition to this lack of important questions in a cultural economic sense, the
aggregation process does not give attention to cultural activities. For example,
individual education and family income level are important factors in arts participation; it
is well known that a strong correlation between income and education levels exists.
Therefore, in examining education as a factor, it is necessary to eliminate the effect of
income from education and vice versa. The Ford Foundation (1974) attempted to
separate the two using aggregated cross-tabulation on two levels (high and low):
education and income. Gray (1998a) and Heilbrun and Gray (2001) also separated the
effects of education and income by multivariate statistical analysis and concluded that
the effect of education is much greater than that of income (OHagan, 1996). See also
Seaman (2006) for the detailed survey of the relationship and treatment of education
and income in explaining demand for cultural activities. However, cross-tabulation of
education and income level is not available in the STULA. The only possibility for
separating these two factors is to use the microdata. Figure 1 is based on cross-
tabulation by education and income level using the microdata, which provides
suggestive information on the significant effects of both education and income level on
arts participation.

347
316 Asia Pacific Journal of Arts and Cultural Management Vol. 5 Issue 1 August 2008 pp 343-361 University of South Australia ISSN 1449-1184





















Another disadvantage of the STULA in analysing arts participation is that the number of
cultural activities (as well as of other activities in sports or hobbies and amusements)
surveyed in 2001 is less than that in 1996 and 2006 (Table 1). In particular, listening to
music on CDs, watching movies on video, and other cultural activities such as choirs,
were eliminated in 2001, which means that the relationship between activities, such as
attending classical concerts and listening to CDs, cannot be considered. Fortunately,
these activities were reintroduced in the 2006 survey.


3. Data and model

As described in the previous section, the usability of microdata is not convenient for
researchers in Japan. However, some trial projects for providing microdata (samples of
anonymous records) have been recently undertaken to improve the availability of
microdata, and the data used in this paper are drawn from the project conducted by
Hitotsubashi University. The microdata of the STULA provided by the project differs
from the original sample records as follows:

- approximately 80% of households are extracted by simple random sampling (all
family members are included for the sample households)

- households with more than nine family members are omitted

- the ordering of households is sorted by random digits

- there are only two classifications of region: major metropolitan areas and other
areas
- people older than 80 years are classified as 80 years old.
348
Figure 1 Participation Rate for Watching Plays, etc.
by Education and Income (2001, female)
0%
10%
20%
30%
40%
50%
-100 100- 200- 300- 400- 500- 600- 700- 800- 900- 1000- 1500+
Annual Income of Households (10 thousand Yen)
Participation
Rate
Elementary, Junior high school
High School
Junior college, technological college
University, College
Source: microdata of the STULA
316 Asia Pacific Journal of Arts and Cultural Management Vol. 5 Issue 1 August 2008 pp 343-361 University of South Australia ISSN 1449-1184
The STULA conducted in 1991, 1996 and 2001 are provided as a microdata set.
However, only the result for the latest survey will be presented in the next section
because of space limitations. The size of the anonymous sample for the 2001 survey is
nearly 150,000.

From Figure 1, education seems to be a very important factor in determining
participation in arts and cultural activities in Japan, even after separating the effect of
income level. However, even if such data are available, other demographic and
socioeconomic factors have to be considered. To achieve this, a logit model is utilised
for examining the determinants of arts participation. The dependent variable is set to
one (zero) if the observation showed participated (did not participate) in an activity,
and this variable is further analysed by other explanatory variables discussed later.

Let be the probability of participation in an arts or cultural activity of
the th observation, and let be factors of participation. Then, the
logit model is specified by: (1)

An economic interpretation of the logit (or probit) model can be obtained using the
concept of utility. Lvy-Garboua and Montmarquette (1996), Prieto-Rodrguez and
Fernndez-Blanco (2000) and Gray (2003) provided a theoretical basis for the discrete
choice of a model for arts and culture participation or attendance from the perspective
of the microeconomic theory of the utility function. For details of estimation, see Greene
(2003). Gray (2003: 358) also listed the representative explanatory variables for arts
participation (children at home, age, hours worked per week, household income,
educational level, art/music lessons, sex and race) and presented their expected impact
on the rates of participation.

The model is estimated for activities mainly related to art, i.e. works of art (excluding
television programmes), watching plays, vaudevilles and dances (excluding television
programmes), watching movies (excluding television programmes and videos),
attending classical music concerts, attending popular music concerts, playing musical
instruments, and traditional Japanese music. Moreover, for comparing cultural activities
and entertainments, playing home-use video games (including use of portable game
machines) and travel are utilised as dependent variables.

The explanatory variables used in this paper are as follows: gender, age, education
(graduated and attending), income, education of the youngest child and occupation,
and with the exception of income, all variables are dichotomous. Details of these
variables are presented in Table 2.









) 1 ( P
i i
y p
i k
x x x , , ,
2 1

k
) exp( 1
) exp(
2 2 1 1 0
2 2 1 1 0
ki k i i
ki k i i
i
x x x
x x x
p



349
316 Asia Pacific Journal of Arts and Cultural Management Vol. 5 Issue 1 August 2008 pp 343-361 University of South Australia ISSN 1449-1184


































There might be other necessary explanatory variables, such as region or the number of
children. The latter are not available in the microdata provided, unfortunately. The
information of the region is as noted, expressed only by metropolitan and other areas
for reasons of confidentiality. Although the author added this dichotomous regional
variable in logit models, the signs of estimates are not appropriate in most cases. The
information about region is expected to be provided by prefectures or districts
(geographic location) in order to analyse regionality. Therefore, the regional variable is
not included in the following results.




350
Table 2 List of Explanatory Variables in Logit Model
Category Variable name
Gender GENDER
AGE10 10-19 years=1 otherwise=0
AGE20 20-29 years=1 otherwise=0
AGE30 30-39 years=1 otherwise=0
AGE40 40-49 years=1 otherwise=0
AGE50 50-59 years=1 otherwise=0
AGE60 60-69 years=1 otherwise=0
EDUG_JUN elementary or junior high school =1 otherwise=0
EDUG_HIGH high school =1 otherwise=0
EDUG_COL junior college or technological college=1 otherwise=0
EDUG_UNI university, college or graduate school =1 otherwise=0
EDUA_ELE elementary school =1 otherwise=0
EDUA_JUN junior high school school =1 otherwise=0
EDUA_HIGH high school =1 otherwise=0
EDUA_COL junior college or technological college=1 otherwise=0
EDUA_UNI university, college or graduate school =1 otherwise=0
Income INCOME
EYC_BEF before kindergarten and others=1 otherwise=0
EYC_KIND daycare and kindergarten=1 otherwise=0
ETC_ELE elementary school=1 otherwise=0
EYC_JUN junior high school=1 otherwise=0
EYC_HIGH high school=1 otherwise=0
EYC_UNI junior college, university and etc.=1 otherwise=0
OCU_MANAGER managers and officials=1 otherwise=0
OCU_CLERK clertial and related=1 otherwise=0
OCU_SALES sales=1 otherwise=0
OCU_SERVICE service=1 otherwise=0
OCU_PROTECT protective service workers=1 otherwise=0
OCU_AGRI agricultural, forestry and fishery=1 otherwise=0
OCU_TRANS transport and communication=1 otherwise=0
OCU_MANU manufacturing=1 otherwise=0
OCU_TEACH teacher=1 otherwise=0
OCU_ARTIST artist=1 otherwise=0
OCU_PROF professional and technical work=1 otherwise=0
Description
male=1, female=0
log of annual income of household
Note: The benchmarks in the model (all related dummy variables take zero) are as follows: female for the
gender variable, more than 70 years old for the age variable, and never attended school for education
variable, no child for education of own youngest child variable, not working for the occupation variable.
Income is a quantitative variable using the midpoint in the income classes.
Age
Education of
own youngest
child
Occupation
Education,
attending
Education,
finally
graduated
316 Asia Pacific Journal of Arts and Cultural Management Vol. 5 Issue 1 August 2008 pp 343-361 University of South Australia ISSN 1449-1184
The expected signs of the coefficient of the explanatory variables are as follows (see
Gray 2003). The sign of the coefficient of gender is not always obvious if other things
are equal, especially if the experience of art during childhood is given as the capital.
However, as stated in the previous section, we cannot use the information on the
experience of attendance in arts lessons during childhood in the STULA, so the
expected sign could be negative because of greater accumulated capital for females.
The effect of age is complicated because, while human capital theory insists upon a
positive linear effect of age, the life-cycle effect suggests a nonlinear relationship
between arts participation rates and age. This is the reason for assigning dichotomous
variables for each age class. Relating to ages, cohort is one of the factors affecting arts
participation. However, as the logit model is estimated using cross-section data, this
factor is omitted. In terms of education, the higher the level of education, the greater the
coefficients of graduated education. Although McCarthy and Jinnett (2001: 19)
indicated that the reason for this is not clear, Nagayama (1998: 260) suggested that the
reason may be having enough free time and opportunities to feel and experience arts
during an emotional period through broad friendship. Of course, if education is
regarded as capital, the human capital theory can be directly applied. Even though,
according to this, the coefficients of attending school are similar, we could not
determine the expected sign of the coefficients because of the situation in Japanese
education and society. In elementary school, pupils have more opportunity for attending
live performances than older school students for a number of reasons, including more
art/music appreciation courses at the elementary level. This might also be caused by
competitive entrance examinations and low birth rates (the less number of children, the
more the parents have the opportunity to take their children for live performances due
to a reduction in physical and financial costs). The coefficient of income would be
positive according to primary demand theory, which implies ability to pay for arts
participation (see Heilbrun & Gray 2001; Throsby & Withers 1979). For the coefficients
of education of the youngest child, childcare expenses have negative effects upon arts
participation because of time and budget constraints and other barriers to participating
in arts activities, which is expected to provide similar information as the number of
children. As for occupation, the more professional, the higher the participation, as
examined in Baumol and Bowen (1966), Kurabayashi and Ito (1992) , Nagayama
(1998) and Borgonovi (2004).




















351
316 Asia Pacific Journal of Arts and Cultural Management Vol. 5 Issue 1 August 2008 pp 343-361 University of South Australia ISSN 1449-1184


4. Estimation Results

Table 3 presents the estimated results of the logit model for selected arts and cultural
activities and a few other amusements. The exponent of the estimates presents
covariates of the odds ratio using explanatory variables. In the following discussion,
certain results for exponentiated estimated coefficients of the logit model, , are
summarised using graphs of factors. Note that the magnitude of the coefficients does
not correspond to the marginal effect of an explanatory variable and means
.
































)

exp(
k

1 )

exp(
k

352
Table 3 Estimation Result of the Logit Model (2001)
GENDER -0.4542 ** 0.0141 -0.8593 ** 0.0163 -0.7998 ** 0.0210 -0.6977 ** 0.0174 -0.2842 ** 0.0104
AGE10 -0.4760 ** 0.0639 -0.6240 ** 0.0729 -0.3202 ** 0.0941 0.6972 ** 0.0651 1.7179 ** 0.0439
AGE20 -0.4455 ** 0.0311 -0.7531 ** 0.0344 -0.4943 ** 0.0505 0.5969 ** 0.0395 1.5284 ** 0.0334
AGE30 -0.1638 ** 0.0315 -0.3984 ** 0.0344 -0.1923 ** 0.0505 0.5554 ** 0.0415 1.4012 ** 0.0342
AGE40 0.1244 ** 0.0305 -0.1406 ** 0.0326 0.1873 ** 0.0485 0.6024 ** 0.0412 1.2991 ** 0.0346
AGE50 0.3250 ** 0.0269 0.0982 ** 0.0273 0.3816 ** 0.0439 0.6370 ** 0.0371 0.9349 ** 0.0338
AGE60 0.4872 ** 0.0260 0.3241 ** 0.0256 0.6451 ** 0.0427 0.6709 ** 0.0364 0.6634 ** 0.0353
EDUG_JUN -0.2349 * 0.1111 0.0806 0.1041 -0.6996 ** 0.1609 0.0373 0.1261 -0.1714 0.0942
EDUG_HIGH 0.6766 ** 0.1095 0.3612 ** 0.1033 0.4582 ** 0.1569 0.4015 ** 0.1248 0.4380 ** 0.0926
EDUG_COL 1.0449 ** 0.1103 0.5703 ** 0.1047 0.9866 ** 0.1580 0.5500 ** 0.1261 0.5684 ** 0.0932
EDUG_UNI 1.0669 ** 0.1104 0.5643 ** 0.1051 1.1540 ** 0.1582 0.4272 ** 0.1266 0.4954 ** 0.0933
EDUA_ELE 1.0483 ** 0.1308 1.1239 ** 0.1300 1.7046 ** 0.1834 0.1155 0.1454 0.7714 ** 0.0999
EDUA_JUN 1.0827 ** 0.1292 0.9873 ** 0.1292 1.7598 ** 0.1822 0.3836 ** 0.1420 0.6612 ** 0.0996
EDUA_HIGH 1.0163 ** 0.1291 1.1799 ** 0.1280 1.5049 ** 0.1827 0.6853 ** 0.1404 0.6474 ** 0.0994
EDUA_COL 0.9484 ** 0.1276 0.5655 ** 0.1289 1.0161 ** 0.1816 0.4927 ** 0.1422 0.5923 ** 0.1004
EDUA_UNI 1.2032 ** 0.1173 0.6691 ** 0.1176 1.3172 ** 0.1689 0.5112 ** 0.1337 0.4684 ** 0.0966
INCOME 0.2223 ** 0.0093 0.2390 ** 0.0101 0.2254 ** 0.0137 0.1985 ** 0.0110 0.1427 ** 0.0071
EYC_BEF -0.5488 ** 0.0381 -0.5130 ** 0.0466 -0.6326 ** 0.0628 -0.8605 ** 0.0496 -0.5893 ** 0.0262
EYC_KIND -0.2904 ** 0.0369 -0.1598 ** 0.0425 -0.0642 0.0529 -0.3958 ** 0.0454 0.0267 0.0232
ETC_ELE -0.1362 ** 0.0283 -0.1125 ** 0.0338 0.2550 ** 0.0384 -0.1838 ** 0.0360 0.2336 ** 0.0197
EYC_JUN -0.1230 ** 0.0352 -0.2146 ** 0.0438 0.1884 ** 0.0480 -0.0809 0.0449 0.1052 ** 0.0278
EYC_HIGH -0.0614 0.0339 -0.1079 ** 0.0409 0.0565 0.0502 -0.0856 0.0450 0.0263 0.0300
EYC_UNI -0.1666 ** 0.0427 -0.1712 ** 0.0506 -0.1080 0.0641 -0.1620 ** 0.0584 -0.0290 0.0413
OCU_MANAGER 0.2336 ** 0.0367 0.2963 ** 0.0469 0.1556 ** 0.0586 0.2328 ** 0.0562 0.2516 ** 0.0376
OCU_CLERK 0.1595 ** 0.0206 0.2411 ** 0.0237 0.1349 ** 0.0307 0.2465 ** 0.0253 0.2154 ** 0.0170
OCU_SALES -0.0637 * 0.0253 -0.0045 0.0296 -0.2122 ** 0.0404 0.0606 * 0.0308 0.0984 ** 0.0197
OCU_SERVICE -0.1372 ** 0.0300 -0.0113 0.0323 -0.1654 ** 0.0452 0.0351 0.0334 0.1028 ** 0.0216
OCU_PROTECT -0.0217 0.0763 0.0220 0.1008 -0.4829 ** 0.1569 -0.0346 0.1082 0.1437 * 0.0585
OCU_AGRI -0.0221 0.0402 0.4124 ** 0.0347 -0.1878 ** 0.0705 0.2842 ** 0.0453 -0.0106 0.0406
OCU_TRANS -0.4797 ** 0.0692 -0.2923 ** 0.0803 -0.3541 ** 0.1120 -0.1511 * 0.0768 -0.0823 0.0447
OCU_MANU -0.1882 ** 0.0233 -0.0773 ** 0.0257 -0.2182 ** 0.0367 0.0379 0.0271 0.0784 ** 0.0178
OCU_TEACH 0.5690 ** 0.0380 0.8740 ** 0.0440 0.6775 ** 0.0508 0.4744 ** 0.0566 0.4421 ** 0.0356
OCU_ARTIST 0.3605 ** 0.0566 0.4738 ** 0.0684 0.7709 ** 0.0680 0.3000 ** 0.0787 0.0535 0.0553
OCU_PROF 0.1455 ** 0.0251 0.2387 ** 0.0303 0.1588 ** 0.0368 0.1474 ** 0.0330 0.1961 ** 0.0205
Constant Term -21.5065 ** 0.1233 -21.2909 ** 0.1199 -22.2285 ** 0.1782 -21.9231 ** 0.1429 -21.5986 ** 0.1051
* significant at 5% significant level, ** significant at 1% significant level
denotes estimated standard error of estimator.
See Table 2 for variable names.
watching works of art
watching plays,
vaudelvilles and dances
going classical music
concerts
going popular music
concerts
watching movies
Variables

s
316 Asia Pacific Journal of Arts and Cultural Management Vol. 5 Issue 1 August 2008 pp 343-361 University of South Australia ISSN 1449-1184


























First, the results for the coefficients of the gender variables (GENDER = 0 for female)
are shown in Figure 2. Overall, females tend to participate more than males in arts and
cultural activities (in particular, watching plays and attending classical music concerts),
but not in amusements (playing TV games). The coefficient of travel is also relatively
small regarding cultural activities. This supports the effects of capital accumulation in
childhood because the experience of arts lessons cannot be included in this analysis,
as described in Section 2. As in Western countries, girls tend to attend piano (or other
cultural) lessons more than boys, which is related to the early acculturation process
described in Gray (2003).












353
Table 3 (continued)
GENDER -0.5611 ** 0.0181 -0.6256 ** 0.0491 0.4116 * 0.0115 -0.1032 ** 0.0081
AGE10 1.3121 ** 0.0720 -1.0553 ** 0.2071 3.5974 * 0.0709 -0.2911 ** 0.0367
AGE20 0.8227 ** 0.0526 -1.2230 ** 0.1034 3.2686 * 0.0662 -0.2275 ** 0.0176
AGE30 0.5860 ** 0.0553 -1.0877 ** 0.1103 2.8955 * 0.0670 -0.1467 ** 0.0183
AGE40 0.5773 ** 0.0563 -0.8296 ** 0.1056 2.4194 * 0.0681 -0.0666 ** 0.0181
AGE50 0.4366 ** 0.0534 -0.2716 ** 0.0812 1.5180 * 0.0691 0.0223 0.0156
AGE60 0.4595 ** 0.0538 0.0336 0.0744 1.2037 * 0.0717 0.2287 ** 0.0147
EDUG_JUN -0.3333 * 0.1646 0.6622 0.5032 0.2236 0.1266 0.2459 ** 0.0592
EDUG_HIGH 0.1858 0.1617 1.3665 ** 0.5010 0.4616 * 0.1250 0.3899 ** 0.0589
EDUG_COL 0.6346 ** 0.1627 1.6696 ** 0.5040 0.4006 * 0.1258 0.4302 ** 0.0597
EDUG_UNI 0.6736 ** 0.1630 1.5841 ** 0.5050 0.2772 * 0.1258 0.3024 ** 0.0597
EDUA_ELE 1.3007 ** 0.1729 1.9941 ** 0.5535 0.8190 * 0.1298 0.9763 ** 0.0711
EDUA_JUN 1.1707 ** 0.1727 2.2555 ** 0.5475 0.7823 * 0.1296 0.8193 ** 0.0708
EDUA_HIGH 0.9829 ** 0.1727 2.5266 ** 0.5446 0.6310 * 0.1295 0.5618 ** 0.0711
EDUA_COL 0.7482 ** 0.1748 2.0193 ** 0.5405 0.4639 * 0.1318 0.4375 ** 0.0719
EDUA_UNI 0.8149 ** 0.1683 1.9743 ** 0.5245 0.2545 * 0.1282 0.3660 ** 0.0653
INCOME 0.1437 ** 0.0122 0.1175 ** 0.0306 0.0547 * 0.0078 0.2213 ** 0.0056
EYC_BEF 0.0081 0.0431 -0.4746 ** 0.1569 -0.0069 0.0228 -0.0669 ** 0.0195
EYC_KIND 0.2871 ** 0.0439 -0.1210 0.1461 0.2767 * 0.0246 0.0926 ** 0.0200
ETC_ELE 0.3468 ** 0.0396 -0.2401 0.1255 0.4836 * 0.0234 0.1392 ** 0.0169
EYC_JUN 0.2416 ** 0.0540 -0.1142 0.1466 0.1537 * 0.0386 -0.0088 0.0227
EYC_HIGH -0.0862 0.0639 -0.3163 * 0.1527 -0.0447 0.0462 -0.0689 ** 0.0229
EYC_UNI -0.2984 ** 0.0905 -0.6323 ** 0.2045 -0.1101 0.0707 -0.1571 ** 0.0299
OCU_MANAGER -0.0073 0.0773 0.2414 0.1464 0.2155 * 0.0498 0.2614 ** 0.0246
OCU_CLERK -0.0421 0.0329 0.0612 0.0818 0.0526 * 0.0215 0.2577 ** 0.0135
OCU_SALES -0.1297 ** 0.0386 -0.0604 0.0957 0.0059 0.0230 0.1145 ** 0.0155
OCU_SERVICE -0.0636 0.0412 0.1175 0.0987 -0.0739 * 0.0264 0.0708 ** 0.0179
OCU_PROTECT -0.0902 0.1318 -0.5818 0.4121 0.1312 * 0.0602 0.1883 ** 0.0420
OCU_AGRI 0.1626 * 0.0684 0.2216 0.1177 -0.0468 0.0530 0.5877 ** 0.0180
OCU_TRANS 0.0437 0.0878 -0.1049 0.2317 0.1154 * 0.0402 0.1085 ** 0.0296
OCU_MANU -0.0718 * 0.0348 -0.0746 0.0831 0.1306 * 0.0197 0.1679 ** 0.0130
OCU_TEACH 0.8520 ** 0.0528 1.0268 ** 0.1385 0.1016 0.0526 0.4621 ** 0.0299
OCU_ARTIST 0.8349 ** 0.0693 1.0950 ** 0.1649 -0.2184 * 0.0745 -0.1008 * 0.0508
OCU_PROF 0.2412 ** 0.0363 0.3431 ** 0.0939 0.0297 0.0254 0.2193 ** 0.0168
Constant Term -22.1076 ** 0.1801 -23.4724 ** 0.5316 -23.0710 * 0.1471 -20.3872 ** 0.0679
travel (domestic,
sightseeing)
Variables
playing musical
instruments
traditional Japanese
music
Playing TV games

s
316 Asia Pacific Journal of Arts and Cultural Management Vol. 5 Issue 1 August 2008 pp 343-361 University of South Australia ISSN 1449-1184



















Figure 3 presents the coefficients of the age variables. As noted earlier, there are two
theoretical possibilities for this tendency: the positive linear relationship according to
human capital theory, and the nonlinear relationship according to the life cycle effects. It
is clear that human capital theory explains the relationship between age and
participation for attending classical concerts, visiting galleries and watching plays etc.,
namely those activities classified as so-called high culture (as for the rise in the
youngest group, see the explanation for school attendance). In contrast, people watch
movies less as age increases because much of the content of these activities is
directed towards the younger market. Within popular music concerts, U-curved
tendencies can be observed. Although it seems misplaced, popular music includes
traditional Japanese folk songs such as enka, which are preferred by older people.
Since Japan is rapidly becoming an ageing society, classical music concerts and art
museums will become more important and accordingly the content of popular music
and movies should turn its attention to the needs of older people.

















354
Fi gure 2 Exponenti ated Coeffi ci ents of Gender
0 1
TV games
travel
watchi ng movi es
vi si ti ng art museums
pl ayi ng musi cal i nstruments
tradi ti onal Japanese musi c
popul ar musi c concerts
cl assi cal musi c concerts
watchi ng pl ays etc
)

exp(
Note: Calculated from for the gender variable in Table 3.

Figure 3 Exponentiated Coefficients of Age


0
1
2
3
4
5
6
10-19 20-29 30-39 40-49 50-59 60-69 age (years)
popular music concerts
visiting art museums
classical music concerts
watching plays etc.
watching movies
playing musical instruments
)

exp(
Note: Calculated from for the age variables in Table 3.

316 Asia Pacific Journal of Arts and Cultural Management Vol. 5 Issue 1 August 2008 pp 343-361 University of South Australia ISSN 1449-1184
Figure 4 represents the income elasticity of participation (ratio between percentage
change in income and participation rate). All the coefficients are positive and
statistically significant (see Table 3), which is interpreted as the income effects as
described in the usual demand theory. In arts and cultural activities, attending
performing arts (classical concerts and plays etc.) and visiting museums demonstrate
relatively large elasticity, which means that the effect of income is more pronounced
because of the high price of tickets, particularly for classical concerts and musicals in
Japan. These high prices in Japan are sometimes accounted for by the expensive
rental fee of theatres, low level of both public and private funding, limited number of
concerts held and so on. The effect of income level on classical music concerts is
greater than on popular music, as was expected. Moreover, the effect of income on
attending performing arts is generally greater than that on positive (acting) activities,
such as playing musical instruments. The coefficients for entertainment activities (TV
games as well as sports) are relatively small compared with those for cultural activities,
which stresses the importance of the ability to pay in arts participation.





















Figure 5 represents the magnitudes of the coefficients of education (graduate school).
The coefficients for university and junior college graduates are significantly higher than
those for high school graduates for most activities in arts and culture, but not for TV
games. For amusement activities such as TV games, the coefficient for university
graduates is less than that of secondary education graduates. It is confirmed that
Japanese people with higher levels of education tend to display higher participation
rates, which supports the human capital theory in cultural activities. The most prominent
effect of education is recognised in attendance at classical music concerts, because
here the difference between the coefficients of lower and higher levels of education is
the largest, followed by visiting art museums. As a factor of investment, the effect of
education is most pronounced in these cultural activities. However, it is not clear what
the effects of education will be in the future, considering the increasing ratio of people
undertaking university education (nearly 50% even now), due to low birth rate and the
decreasing number of children.


355
Figure 4 Elasticity of Income
0.0 0.1 0.2
Playing TV games
watching movies
travel (domestic,
sightseeing)
traditional Japanese
music
playing musical
instruments
going popular music
concerts
watching works of art
watching plays,
vaudevilles and dances
going classical music
concerts
elasticity of income
Note: Calculated from for the income variables in Table 3 and participation rates.

316 Asia Pacific Journal of Arts and Cultural Management Vol. 5 Issue 1 August 2008 pp 343-361 University of South Australia ISSN 1449-1184




















The classification of attending school does not present similar results as in the case of
having graduated. The lower the level of attending school, the more frequent
opportunity they have for attending classical concerts, plays etc., and visiting art
museums with family and people in their regional community. The effect of the
accumulation of capital is mixed in this situation, which makes the interpretation of the
coefficients complicated. However, even within higher levels of education, art/music
lessons or appreciation courses for concerts or museums should be encouraged, and
this could be related to the above discussion about society having a low birth rate.

In terms of education (or age) of the youngest child, a clear tendency can be observed
in arts participation: the younger the child, the less the parents participate in arts
activities (Figure 6). In particular, having children below kindergarten age would be a
significant barrier to parents participation in arts activities. Having children at home
means both time and monetary constraints on arts participation. It is not easy to take
children to classical concerts, plays or galleries. Katsuura (2005) analysed the
relationship between time use for child care and participation, in which a negative
correlation was observed for such activities, but a positive correlation existed for
listening to music via media (CDs etc.). This points towards a substitution effect
between attending live performances and listening to music via media when children
are very young (before elementary). This result provides a policy implication that
establishing day-care service facilities in concert halls and theatres would be needed if
these parents felt it was a barrier to attending. When a child is at elementary school, the
coefficient for classical concerts (and playing musical instruments) is high, probably
because parents attend these activities with their children.

Figure 7 represents the estimated coefficients for occupation variables in the logit
model only for classical music concerts and visiting art museums. The greatest positive
effects of occupation are found among teachers and artists, which is consistent with
previous works such as Nagayama (1998) in the Japanese case. Following these
occupations, managers and officials, clerical workers, and professional and technical
workers have higher participation rates than manufacturing and transport workers. Such
an observation is similar to those by Baumol and Bowen (1966) and Borgonovi (2004).
356
Figure 5 Exponentiated Coefficients of Education (Graduates)
0
1
2
3
4
elementary, junior
high
high school junior college university
watching works of art
watching plays, vaudelvilles and dances
going classical music concerts
going popular music concerts
watching movies
playing musical instruments
playing TV games
)

exp(
Note: Calculated from for the education variables in Table 3.

316 Asia Pacific Journal of Arts and Cultural Management Vol. 5 Issue 1 August 2008 pp 343-361 University of South Australia ISSN 1449-1184
This is also explained by human capital development, because of the strong
relationship between cultural capital and economic status in such occupations
(Borgonovi, 2004). Teachers and artists might be regarded as the most
professionalised occupations from the viewpoint of culture.



















357
Figure 6 Exponentiated Coefficients of Education of the
Youngest Child
0.0
1.0
before
kindergarten,
etc.
daycare,
kindergarten
elementary junior high high school junior college,
university
watching works of art
watching plays, etc.
classical music concerts
popular music concerts
playing musical instruments
)

exp(
Note: Calculated from for the education of own youngest child in Table3.

Figure 7 Exponentiated Coefficients of Occupation


0.0 1.0 2.0
managers and officials
clertial and related
sales
service
protective service workers
agricultural, forestry and fishery
transport and communication
manufacturing
teacher
artist
professional and technical
watching works of art
classical music concerts
)

exp(
Note: Calculated from for the occupation variables in Table 3.

330 Asia Pacific Journal of Arts and Cultural Management Vol. 5 Issue 1 August 2008 pp 343-361 University of South Australia ISSN 1449-1184
5. Concluding Remarks

This paper has applied a logit model to the STULA microdata for analysing arts and
cultural participation in Japan. Economic implications are extracted from the results, as
described in the previous section. Although participation data in Japan, i.e. the STULA,
have not been given international attention, most of the results with regard to the effects
of education, income, age, occupation and education of children are now proved to be
similar to previous research in Western countries. They are also well explained in the
literature, such as the human capital theory developed in Stigler and Becker (1977).
The differences between cultural and other activities, are also examined. Attending
classical concerts, visiting art museums and watching plays etc., are the activities that
are greatly affected by age, income and education; fewer effects are experienced by
popular music concerts and movies, and the least are experienced by entertainments.

However, more detailed analysis is possible utilising the data of the STULA. For
example, in addition to socioeconomic and demographic characteristics, other cultural
activities have certain effects upon participation in arts. Most arts and cultural activities
are positively correlated, including classical and popular music. However, participation
in amusement activities presents different trends than participation in arts and cultural
activities in terms of education, income and interrelationship. A simultaneous equation
model for multiple activities must be established and estimated in order to analyse
interrelationship, as developed by Prieto-Rodrguez and Fernndez-Blanco (2000).
Moreover, in addition to whether an individual participated or not, the frequency of
participation can also be analysed. One of the models for considering frequencies is the
ordered probit model, which is applied to the SPPA data in Borgonovi (2004). These
developments could be considered for a future study.

In such a multivariate analysis, one of the most difficult problems is the selection of
explanatory variables. A large number of combinations of variables is possible, and the
definition of variables itself is not simple. A trial and error process is needed to define
the optimal model; however, a model for one activity may not necessarily be suitable for
others. Although this paper utilises a standard model using the explanatory variables in
Table 2, this model is no guarantee of the optimal (the results for other types of model
are available upon request).

The analysis of arts and cultural participation using the STULA (or Japanese data) may
be only an initial attempt. One of the reasons for this is the limited accessibility of the
data, as described in Section 2. There are many barriers and constraints to using micro
-data from the official statistics, partly due to legal matters. The Statistics Law was
revised in 2007, and the improvement of the secondary use of micro-data is now
underway and is hoped to be successful in the near future. If the accessibility to micro-
data is improved, the analysis of participation in arts and culture in Japan could be
further developed.










358
330 Asia Pacific Journal of Arts and Cultural Management Vol. 5 Issue 1 August 2008 pp 343-361 University of South Australia ISSN 1449-1184
References

Abb-Decarroux, F. and F. Grin, (1992), Risk Aversion and the Demand for Performing
Arts, in R. Towse and A. Khakee (eds.), Cultural Economics, Springer-Verlag, Berlin,
125140.

Borgonovi, F. (2004), Performing Arts Attendance: An Economic Approach, Journal of
Cultural Economics, 36, 18711885.

Baumol, W. and Bowen, W. (1966), Performing Arts: The Economic Dilemma, The
Twentieth Century Fund, New York.

DiMaggio, P. and Ostrower, F. (1992), Race, Ethnicity and Arts Participation, Research
Division Report #25, National Endowment for the Arts.

Ford Foundation (1974), The finances of the Performing Arts, vol. 2, A Survey of the
Characteristics and Attitudes of Audiences for Theater, Opera, Symphony and Ballet in
12 U. S. Cities, A Report to the Ford Foundation.

Gray, C. M. (1998a), Trends in Arts Participation 1982 - 1997, Report to the National
Endowment for the Arts.

Gray, C. M. (1998b), Hope for the Future? Early Exposure to the Arts and Adult Visits
to Art Museums, Journal of Cultural Economics, 22, 8798.

Gray, C. M. (2003), Participation, in Towse, R. (ed.), A Handbook of Cultural
Economics, Edward Elgar, Cheltenham, 356365.

Greene, W. H. (2003), Econometric Analysis, 5th Edition, Prentice Hall, New Jersey.

Harris, L. (1996), Americans and the Arts VII, Americans for the Arts, Washington DC.

Heilbrun, J. and C. M. Gray (2001), The Economics of Art and Culture, Cambridge
Univ. Press, Cambridge.

Katsuura, M. (2003), Statistical Analysis for Art Activities by Japanese People, Paper
presented at the 2003 Meeting of Japan Association for Cultural Economics, 112115
(in Japanese).

Katsuura, M. (2005), Relationship between Time Use and Leisure Activities in the
Survey on time Use and Leisure Activities, The Meijo Review (Meijo Ronso), 5, 145
156 (in Japanese).

Katsuura, M. (2006), Possibilities of Applying Statistical Methods to Empirical Study of
Arts and Culture, BunkaKeizaigaku, 5, 1725 (in Japansese).

Kurabayashi, Y. and T. Ito (1992), Socio-Economic Characteristics of Audiences for
Western Classical Music in Japan: A Statistical Analysis, in R. Towse and A. Khakee
(eds.), Cultural Economics, Springer-Verlag, Berlin, 275287.

Lvy-Garboua, L and Montmarquette, C. (1996), A Microeconometric Study of Theater
Demand, Journal of Cultural Economics, 20, 2550.

359
330 Asia Pacific Journal of Arts and Cultural Management Vol. 5 Issue 1 August 2008 pp 343-361 University of South Australia ISSN 1449-1184
McCarthy, K. F. and K. Jinnett (2001), A New Framework for Building Participation in
the Arts, RAND, Santa Monica.

Morrison, W. G. and West, E. G. (1986), Child Exposure to the Performing Arts: The
Implications for Adult Demand, Journal of Cultural Economics, 10, 1724.

Nagayama, S. (1998), Cultural Economics and Empirical Analysis, in Ikegami, A. et al.
(eds), Cultural Economics, Yuuhikaku, Tokyo, 249277 (in Japanese).

National Endowment for the Arts (2004), 2002 Survey Of Public Participation in the
Arts, Research Division Report #45, National Endowment for the Arts.

OHagan, J. (1996), Access to and Participation in the Arts: The Case of Those with
Low Incomes/Educational Attainment, Journal of Cultural Economics, 20, 269282.

Peterson, R. A., Hull, P. C. and Kern, R. M. (2000), Age and Arts Participation: 1982
1997, Research Division Report #42, National Endowment for the Arts.

Prieto-Rodrguez, J. and V. Fernndez-Blanco (2000), Are Popular and Classical
Music Listeners the Same People? Journal of Cultural Economics, 24, 147164.

Seaman, B. (2006) Empirical Studies of Demand for the Performing Arts in Ginsburgh,
V. and Throsby, D. (eds.), Handbook of the Economics of Art and Culture, Amsterdam:
North Holland, 415472.

Statistics Bureau Japan (2003), 2001 Survey on Time Use and Leisure Activities, Japan
Statistical Association, Tokyo.

Stigler, G. J. and Becker, G. S. (1977), De Gustibus Non est Disputandum, American
Economic Review, 67, 7690.

Throsby, C. D. and Withers, G. (1979), The Economics of the Performing Arts, St.
Martins Press, New York.
360
332 Asia Pacific Journal of Arts and Cultural Management Vol. 5 Issue 1 August 2008 pp 343-361 University of South Australia ISSN 1449-1184
Acknowledgements

This paper is based on a presentation at the ACEI 14th International Conference on
Cultural Economics, Vienna, 2006. The author is grateful to the discussant and chair of
the session, Prof. C. M. Gray, as well as the editor and two anonymous referees of the
journal, for their helpful comments. This research was partially supported by the
Ministry of Education, Science, Sports and Culture, Grant-in-Aid for Scientific Research
(C), 18530161, 20062008.
361

You might also like