Professional Documents
Culture Documents
FERNÁNDEZ-BLANCO, V. y BAÑOS-PINO, J. (1997)
FERNÁNDEZ-BLANCO, V. y BAÑOS-PINO, J. (1997)
57
c 1997 Kluwer Academic Publishers. Printed in the Netherlands.
Abstract. In this paper we explain the decline in cinema attendance in Spain since 1968. We use a
cointegration analysis to estimate an individual demand function for cinema, with annual data and in
terms of average attendance per inhabitant. We find that cinema is a luxury good and that its demand
is elastic with respect to its price. We also show that television acts as a deterrent with regard to
cinema attendance.
Key words: cultural economics, cinema, demand, elasticity, cointegration analysis
At the end of the nineteenth century, Louis Loumière warned Georges Mélies that
the cinematograph could be regarded as a scientific curiosity but that there was
no commercial future in it. He meant to give Mr. Mélies well-meaning advice but
little did he know to what extent history would prove him wrong.1 Nowadays the
cinema holds a prominent position in the world of mass culture and entertainment,
not only in Spain but also on an international scale. For example, in Spain 36,300
million pesetas were obtained in cinema box-office receipts in 1992, making it the
largest cultural business, surpassing all the other performing arts which as a whole
generated 24,500 million pesetas.
This paper studies the market for cinema in Spain from 1968 to 1992.2 This
twenty-five year period represents a period of crisis. The number of film-goers, the
average attendance per person, and box-office receipts all dropped considerably.
The main objective of this paper is to study the major factors driving cinema
attendance in Spain. This paper is structured as follows. In Section 1 we examine
the historic evolution of the film exhibition trade, focusing mainly on demand.
Section 2 outlines and defines the theoretical framework. In Section 3 we explore
the variables used in this analysis and the statistical sources. Section 4 specifies the
econometric model. In Section 5 a cointegration analysis is used to estimate a long
and short run demand function for cinema in Spain during the period 1968–1992.
Finally, Section 6 presents a summary of this work.
During the period 1968–1992 the market for cinema in Spain has experienced a
gradual decline, which over the 1980–1988 period was more dramatic. In 1988
this crisis touched rock bottom and from then onwards it was stable, and a slight
improvement was even detected. The evolution of the trade’s principle variables
(number of film-goers, average attendance per citizen, and year, and receipts in
constant pesetas) can be seen in Figures 1, 2 and 3.
The evolution of the number of attendees and the average attendance in the first
two figures illustrate identical dynamics: until 1988 there was a progressive decline
and from this point onwards the series remains stationary. From 1968 to 1992 the
number of attendees has decreased by a 77%, from a total of 376 million it has
dropped to scarcely 86 million. With respect to attendance, a similar phenomenon
can be observed. In 1968 each Spanish citizen would go to the cinema an average
of 11.33 times a year, whereas in 1992 they would go only twice a year, which
represents an 80% decrease. An in-depth study of both series indicate that the
period 1980–1988 proved to be specially critical.
Figure 3 shows the pattern for box-office receipts, measured in constant pesetas.
Therefore in Spain, during the sample period receipts decreased by a 54%, from
54,000 million in 1968 to 24,800 in 1992. However, two stages can be clearly
defined. Until the end of the 70s receipts remained fairly stable, albeit with episodic
ups and downs. From 1979 and all through the 80s there is a real and steady decline
in the film exhibition trade turnover.
We should also pay attention to the behaviour of other interesting variables, such
as the average expenditure per attendee or the number of cinemas. The first is the
quotient of receipts divided by the number of attendees and it could be considered
as the average ticket price. This price, measured in real terms, has increased by
a 100% in the sample period. However, we must underline that since 1989 it has
become stable and even decreased. Regarding the number of cinemas, whilst not
having an immediate influence on the demand, its evolution accentuates a decline
in the trade; 5,954 cinemas disappeared, a 77% decline, between 1968 and 1992.
Here we will describe the general framework, which is built on two foundations:
conventional demand theory and similar studies on the demand for the performing
arts.
From the point of view of demand there are important aspects that the cinema
has in common with the performing arts and other cultural events. In both cases
when consumers purchase their tickets, they are rendered a cultural service that,
once inside the theatre, is a nonrival consumption. Due to this, it is possible to
establish some methodological similarity when analysing the demand for cinema
and the demand for the performing arts.3 For this reason we take as a starting point
the general framework presented by Throsby and Withers (1979) and the prior
studies on theatre demand by Moore (1966), Withers (1980), Gapinsky (1984 and
1986), Abbé-Deccarroux (1994) and Krebs and Pommerehne (1995), and cinema
demand by Cameron (1986, 1990).
Following the works of those authors and according to the fundamental princi-
ples of the economic theory, the following individual demand function for cinema
can be written, and defined in terms of average attendance per inhabitant and year.
Qt = f (Pt; Pst; Yt ; A)
According to this function, the amount of cinema demanded in the period t (Qt )
will depend on the admission price (Pt ), on the price of substitute or complementary
goods (Pst ), on income (Yt ) and on a vector of variables (A) which comprises the
influence exerted by other factors, which in some way can reflect any changes in
the audience’s preferences. We would expect the amount of cinema demanded to be
positively affected by income and the price of substitutes and negatively influenced
by the price of cinema, the price of complementary goods and the influence of TV
and video. With an estimate of that demand function we will be able to calculate
the values of price and income elasticities and to check two customary hypotheses.
The first indicates that the demand for cinema is price elastic. According to this
hypothesis, cinema would be an immediately accessible entertainment such as a
popular musical or a circus which, as Throsby (1994, p. 7) points out, have a less
discriminating nature of demand and subsequently, substitutes are more readily
available than in the case of “higher” performing arts (such as opera or classical
music, among others). Therefore, own-price responsiveness would be greater. In
the second hypothesis we check whether cinema is a luxury product, that is to say,
whether its income elasticity is greater than one. In the case of arts this result is
justified in the long-run because consumption of the arts “is linked with education,
which in turn is a significant determinant of income.”
This demand function has other two characteristics which should not go unmen-
tioned. First, in accordance with the general approach already presented, the amount
of cinema demanded can be defined as the average attendance per citizen in period
t; it is, therefore, a per capita variable and consequently, the income variable needs
to be expressed in per capita terms. Secondly, all the money variables – prices and
income – will be measured in real terms, hence their current values will be deflated
by the GDP Implicit Deflator. This involves accepting the assumption, quite usual
in economic theory, that demand function is homogeneous of degree zero in prices
and income.
Finally, from a theoretical viewpoint, the quality of the product must also be a
significant determinant of the demand for cinema attendance, as well as the demand
for performing arts. Although its influence has been empirically demonstrated
(Abbé-Deccarroux, 1994; Lévy-Garboua and Montmarquette, 1996), it is hard to
include quality explicitly in an empirical analysis. In the case of cinema in Spain it
is difficult to define a unit of standard quality for the sample period. Besides in an
analysis of time series like this, we might expect that distribution between bad and
good movies is very similar each year and, consequently, quality is more or less
homogeneous for the whole period. For these two reasons, the demand function
proposed does not include any quality variables.
With respect to supply, the film exhibition trade finds itself in a rather peculiar
situation. In each cinema every ticket is sold at a uniform price which is fixed, and
the only restriction is the capacity of the closed venue. That is to say, supply is
perfectly elastic at the fixed price until the maximum capacity has been reached. 4
This characteristic of supply plays an essential role in the model because, under
those conditions, the equilibrium attendance is determined exclusively by demand.
The amount sold in a period coincides with the amount demanded and consequently,
the model does not pose an identification problem.
is not a film, but a different theatre play. On the other hand, this finding is relevant in
the case of disaggregated terms. However, it could be argued that in general terms
there is substitution. Nonetheless, when dealing with the Spanish case, cinemas
are available all over the country whereas theatres are only available in Madrid and
Barcelona. Consequently, it would be difficult to maintain that theatre adequately
plays the role of a substitute.
A second alternative would be to regard any leisure activity as a potential sub-
stitute for cinema. To include this alternative, a generic leisure price has been
incorporated by using average earning per working hour (Owen, 1960). No signifi-
cant results have been obtained so the generic leisure price has been excluded from
the final specification of the model.
Finally, it has been considered that television and video are the closest substi-
tutes for cinema. This possibility is faced with a great disadvantage: there are no
published data on the price of television consumption and on the price of video
films that are for sale and to be rented. Although some alternative variables have
been tested (in particular, we use the price of electrical appliances – according to
the Industrial Price Index – as a “proxy” variable of the price of television sets
and of video recorders), the results proved unsatisfactory. Therefore, instead of
measuring the influence of television and video by means of price, we will opt to
convey this influence by other means, which are comprised in vector A.
Income (Yt ). Disposable per capita income is used, according to the data sup-
plied by the Informes (Reports) published by the Banco Bilbao Vizcaya (Bilbao
Vizcaya Bank), and which will be referred to as INCOME. Both income and price
were deflated by the GDP Implicit Deflator, published in the Contabilidad Nacional
de España (Spanish National Accounts).
Other factors (A). This vector comprises other factors, not relating to price and
income, that may exert some influence on cinema demand, probably modifying
the consumers’ leisure preferences. More specifically, we try to proxy the impact
of television and video, two activities which work as substitutes for cinema, but
whose influence has not been possible to measure via prices.
Video, and especially, purchasing or rental of films or videotapes, probably
plays an important restrictive role as far as the decision of cinema attendance is
concerned. Although it is highly difficult to have access to public information on
the number of films which have been purchased or rented as well as on their price,6
the Anuarios (Yearbooks) published by the national newspaper “El Paı́s”, supply
some data on the evolution of the number of home video cassette recorders. VCRs
first appeared in 1980 and since 1983 have expanded on a large scale. The influence
of video is introduced by means of a binary variable (VIDEO) which has value
zero until 1980 and value one onwards. This variable did not prove to be significant
and it was excluded from the final specification of the model.
The fact that television acts as a deterrent is a phenomenon which is well-known
(Dale, 1984, p. 21; Dunnet, 1990, p. 43). Since its birth cinema became a leading
activity in the mass entertainment trade, in almost all countries, it could not be
overshadowed by any other similar alternative. But the arrival of television and its
progressive expansion in households modified this situation: cinema faced a direct
competitor which offered a similar product, as well as advantages such as comfort
or privacy. The final outcome was a change in the individual’s decision as Cuevas
(1976, p. 68) indicates: “people do not frequent cinemas as before to fill their spare
time; it’s what they have chosen among many other ways of spending that free
time”.
There are several ways, other than prices, to convey the effects of television.
The first possibility is to take into account the number of households which have
television sets. As some international research has shown, the largest influence
television exerts on people’s decisions on leisure occurred during the first years
of its arrival and expansion among the population (Cuevas, 1976, p. 69). If one
bears in mind that, as reported by Cuevas (1976, p. 71), in 1976 approximately
90% of the Spanish households had a television set, it can be assumed that the first
great impact on cinema had already taken place. In this situation, it would certainly
prove difficult for a measure of the number of television sets per household to
yield positive results for the sample period. This is what occurs when we include
in the model the number of television sets per 1,000 inhabitants, according to the
data supplied by UNESCO’s Statistical Yearbook, which does not prove to be a
significant variable of the film-goers’ behaviour.
A second way is to use the data on television audience supplied by the Anuarios
(Yearbooks) published by the aforementioned newspaper “El Paı́s”. However, the
data available do not cover the whole sample period and the results obtained are
not statistically significant either.
A third option would be to consider that the effect brought about by television
on cinema depends not only on the availability of television sets but also on the
supply of television programmes within range of the viewer. If, as already stated,
the use of television had been widespread in Spain since the beginning of the
seventies, it is likely that in the sample period, its influence on cinema should
respond essentially to changes in television programming. In the eighties there
are a series of event which make this television programming to viewers increase
substantially. The TVE (Public Spanish Television) second channel is available
almost all over the country, and the broadcast time increased considerably and,
more especially in 1984 which saw the end of the state television monopoly, as
new regional television appeared for the first time, and later, in 1989 when private
television channels emerged (Jones, 1994). To include this variation of television
programming in the model, the variable TV has been constructed. It is a binary
variable whose value is zero until 1984 and it is one from then onwards.7
Normalized t-values
coefficients
ASIST 1.0000
PRECINE 3.0420 3.55
INCOME –1.2645 –3.18
Note: All the variables are in log form. Critical values are from
Osterwald-Lenum (1992).
attendance per inhabitant, is elastic with regard to its price (the coefficient value
is –3.042) and that cinema is clearly a luxury product, since its income elasticity
has a value of 1.265. The fact that in the last years of the sample period cinema
attendance was steady and even increased shightly can be explained by the high
values of price and income elasticities, by a fall in prices in the last years and a
continuous increase in disposable per capita income.
Next, it was tested whether the error correction mechanism was significant
(Table II). This mechanism proved to be significant at a level of confidence of over
90%, and with the correct sign, in the attendance demand equation. However, its
value is small (–0.065), which shows that each year there is hardly 6.5% correction
last years’ disequilibrium. To study the goodness of this specification several tests
have been performed: the Bera-Jarque (1981) test for the normality of the regression
residuals and the Ljung-Box statistical test for autocorrelation (see Box and Pierce,
1970). The results of the tests lead us to the conclusion that the specification is
appropriate. Moreover, the values of the error mechanism in the price and income
equations are noticeably lower (–0.02 and –0.01, respectively), for this reason, and
according to Banerjee et al. (1994, pp. 268–271), the hypothesis of weak exogenity
cannot be rejected.
We analyse the stability over time of our decision to accept the presence of one
cointegrating relationship. In order to do this, we estimate recursively the VAR
model and calculate the values of the trace and maximum eigenvalue tests. The
recursive analysis starts up with the first 14 observations, which generate an initial
estimation of the likelihood ratio tests of cointegration (LR). Then, one observation
is added and new values of the trace and maximum eigenvalues tests are computed.
Figure 4 illustrates the results of this recursive analysis together with their 5%
critical values, and indicates that one cointegrating relationship is held for the
whole sample period.
Table II. Vector error correction estimates applying the Johansen technique
Included observations 23
Adjusted R-squared 0.1307
Log likelihood 28.1911
Jarque-Bera N (2) 0.654
Ljung-Box Q(1) 0.0219
Ljung-Box Q(2) 1.8431
Ljung-Box Q(3) 2.1514
Ljung-Box Q(4) 3.1328
Note: is first-difference operator. The t-ratio for each parameter appears in brackets
in the Table. All the variables are in log form. The Jarque-Bera (N ) and the Ljung-Box
(Q) statistics have the 2 distribution with degrees of freedom as shown in brackets.
In Figures 5 and 6 two new recursive analysis have been included. Figure 5 shows
a plot of the test for constancy of the cointegration space. The test statistic has been
scaled by the 95% quantity in the 2 distribution such that unity corresponds to a
test with 5% significance level (Hansen and Juselius, 1994). This Figure supports
the hypothesis of parameter constancy for the period we investigate.
Figure 6 illustrates the time paths of the non-zero eigenvalues, which correspond
to eigenvectors normalised, together with the asymptotic 95% error bounds for each
sub-sample. This plot does not indicate non-constancy in our model.
Variable Coefficient
PRECINE –3.8687
(–14.27)
INCOME 0.9779
(2.65)
INTERCEPT 9.5739
(2.64)
Included observations 25
Durbin-Watson 0.7336
Augmented Dickey-Fuller: ADF(3)a –3.8748
Durbin-Watsonb 2.1664
Included observations 24
Adjusted R-squared 0.9909
F -statistic 419.047
Durbin-Watson 1.8369
Jarque-Bera N (2) 0.4498
Ljung-Box Q(1) 0.2741
Ljung-Box Q(2) 1.1495
Ljung-Box Q(3) 5.1814
Ljung-Box Q(4) 7.0727
Note: is first-difference operator. The t-ratio for each parameter appears in brackets in the Table. All the
variables are in log form. The Jarque-Bera (N ) and the Ljung-Box (Q) statistics have the 2 distribution with
degrees of freedom as shown in brackets.
of the effect of income and own price on the demand takes place in the first year,
and almost 90% of this adjustment will occur in three years’ time (as Table A.IV
in Appendix 2 shows).
Fourth, the negative and statistically significant influence of television on cinema
attendance has been checked. Accordingly, it is possible to suspect that in the
eighties people’s leisure preferences experienced a change which has made them
move away from the cinema screens.
Finally, the analysis of the equation residuals is also acceptable. However, there
is evidence of some third order correlation, which according to the corresponding
Ljung-Box statistic lies on the border-line of acceptability. Other diagnostic test
can be noticed in Appendix 2.
To conclude, taking into account the results these estimations have yielded, we
can suggest some factors that explain the progressive decline in cinema attendance
in the sample period. The increase of the cinema ticket price must have had a
considerable effect on this decline, as cinema attendance has a noticeably price
elastic nature with regard to its price. In addition to this effect, we can also add
the impact of an increase of television programming, which has probably been
responsible for a change in the audience’s preferences to the detriment of cinema.
These negative effects have been partly mitigated by the positive influence of an
increase in the disposable per capita income on cinema attendance.
6. Conclusions
In this paper we have analysed for the first time the cinema demand in Spain. Some
of the most important parameters have been calculated, such as the elasticities of
price and income, and the impact of television on audience attendance.
In the first part a descriptive revision was made of the evolution of the principal
variables of the Spanish cinematography market during the period 1968 to 1992.
A significant crisis was observed in the sector, which was reflected in the decline
of the number of attendees, in the average attendance per inhabitant, and the low
box-office receipts in constant pesetas during the sample period. After 1988, we
also show that this situation presents symptoms of a slight improvement.
In the second part, we employed the Engle-Granger test and the Johansen
Maximum Likelihood procedure to identify the cointegrating relationships among
the variables of the cinema demand. Previously we checked that all the variables
have a unit root. Having obtained only one long-term cointegration relation, we
reformulated the cinema demand in the error-correction form (ECM), but without
imposing common factor restrictions.
We have found that in Spain the cinema demand is highly elastic with respect
to its price, especially in the long-term. Income, both positively and significantly
affects cinema attendance, which in the long-term analysis, proves that cinema is a
luxury product. Finally, we have evaluated the impact of the increase of television
To determine the order of integration of the model variables, we used the Dickey-
Fuller and the augmented Dickey-Fuller (1979, 1981) tests to test for orders of
integration, according to the equation
xt = +
xt,1 +
XP x + "
i t, t
1
i=1
where denotes the first-order difference, t is the time trend and p the order of
autoregression.
According to this Table A.I, the fact that cinema attendance and price are
I (2) variables is rejected, but what cannot be rejected is that they are first-order
integrated with non-null constant. In the case of the available income per capita, the
()
Dickey-Fuller technique is used and it cannot be rejected that it is an I 2 variable.
However, as Andrés et al. (1990) suggest, it is possible to consider the existence
of deterministic trends segmented in the mean. In this case, if three segments are
()
being taken, it cannot be accepted that income is I 2 .
+P
=
Fuller; H0 :
0)
= +
xt
xt,1 P x
i=1 i t,1 "t +
Variable ADF Constant Lags DW Q(4)
ASIST –4.21 Yes 0 1.82 4.49
PRECINE 3.11 Yes 1 2.10 0.77
ASIST –1.08 Yes 0 1.88 4.84
PRECINE –2.00 Yes 2 2.02 5.12
Note: All the variables are in log form. The ADF statistic is the
t ratio of the coefficient of xt,1 ; its 5% critical value is –2.99.
The appropriate number of lags for the ADF was determined
on the basis on an examination of the partial autocorrelation
function. The Durbin-Watson and the Ljung-Box Q(4) statistic
have been calculated on the previous equation residuals.
Variable t-ratio
Changes in PRECINE (–1) –0.66
Changes in PRECINE (–2) 0.48
Changes in INCOME (–1) 0.33
Changes in INCOME (–2) 1.43
Changes in INCOME (–3) 0.41
Changes in ASIST (–1) –0.11
Changes in ASIST (–2) 0.64
This Appendix comprises other empirical estimations which relate to the model
described in Table IV.
According to the results displayed in this Table, it can be inferred that in the
suggested model a structural change cannot be observed.
1 0.8954 –1.4600
2 1.1971 –2.3984
3 1.3599 –2.9051
4 1.4475 –3.1787
5 1.4949 –3.3264
6 1.5204 –3.4062
7 1.5342 –3.4493
Finally, a test ARCH (1) (see Engle, 1982) has been used in order to check the
existence of dynamic heteroscedasticity. In this case, the t-statistics corresponding
to the square of the residuals takes value –0.79. Since the corresponding 5% critical
value is 2.12, it can be thought that there are no heteroscedasticity problems.
Acknowledgements
This work has greatly benefited from the comments made by Juan J. Dolado,
Simón Sosvilla and Miguel Sebastián. We are grateful to Marı́a Argente for her
assistance. The authors would also like to express their gratitude to the editors and
two anonymous referees for their useful comments. As usual, a disclaimer applies.
Notes
1. See Navarro (1995).
2. 1968 has been chosen as the starting point because since 1967 there has been a box-office
control; in accordance with the Ministerial Order 26/7/1967, Spain’s Ministry of Culture must
be informed of the number of tickets sold by each cinema.
3. From the point of view of supply, there are considerable reasons to separate the cinema and the
performing arts. This option is very common and can be found in Baumol and Bowen’s (1965)
and in Heilbrun and Gray’s books (1993).
4. Although, for the sake of simplicity, a single price is proposed in the model, in actual fact there
are some other means which enable us to discriminate prices [“Dı́a del espectador”, spectators
day (special discounts on cinema ticket for all viewers on that particular day at a fixed time),
student discount, pensioner discount, etc.].
5. Indubitably the average ticket price and the price of urban transportation could be regarded
as two separate variables. Had we included both separately in the model, there would have
been a methodological contradiction. Following Engle and Granger technique (1987) there
is no cointegration relation among these variables, income and cinema attendance, although
according to the technique proposed by Johansen and Juselius (1991), there is such relation.
6. There are some organisations which, sporadically, supply this type of information, but which is
not publically available, and consequently, it is not possible to obtain a slightly extensive and
homogeneous data series. It has also been considered to use data relating to royalties collected by
the “Sociedad General de Autores de España” (Spain’s Authors General Society), and published
in its annual Memorias (Bulletins). In principle, this source also poses some problems: the series
can only be constructed from 1985 onwards and no distinction can be made among the royalties
collected from the sale of records, audiotapes or videotapes.
7. Other alternatives have been tried in order to comprise the influence of the supply of television
programming. First of all, a synthetic indicator was created in which, each television channel
was given value one proportional to the percentage of population covered by each television
channel. This alternative did not yield good results in the empirical estimation. Secondly, we
tried to measure the effect brought about by new private channels emerging, by means of a
“dummy” variable, with value one since 1989. Once more, the empirical results did not prove
to be satisfactory. Finally, it can be accepted that the “dummy” TV variable may comprise other
factors apart from the influence exerted by television. For example, it could include the effect
of video, specially if we take into account that it is from 1984 onwards that the video recorder
becomes a common domestic appliance in Spanish households (more than a 10% of Spanish
households has a videorecorder). These facts help to explain the reason why the variable VIDEO
mentioned previously is of little significance.
8. Mentioned previously, the Johansen technique was used without a constant, which was not
included in the vector of cointegration either. In this latter case, to justify the absence of a
constant, the test devised by Johansen and Juselius (1990) has been used and a value of 82.81
has been yielded. As the corresponding critical value, which is distributed as a 2 with two
degrees of freedom, is 5.99, the existence of a constant within the cointegration vector can be
rejected.
9. Different methods were tried to correct that anomalous value. Specially, the value corresponding
to 1988 was interpolated and we obtained very similar econometric values to those obtained
after including the dummy variable.
References
Abbé-Decarroux, F. (1994) “The Perception of Quality and the Demand for Services. Empirical
Application to the Performing Arts”. Journal of Economic Behavior and Organization 23: 99–
107.
Andrés, J., Molinas, C., and Taguas, D. (1990) “Una funci ón de consumo privado para la economı́a
española: aplicación del análisis de cointegración”. Cuadernos Económicos de ICE 14: 173–212.
Banerjee, A., Dolado, J., Galbraith, J., and Hendry, D. (1994) Co-Integration, Error Correction and
Econometric Analysis of Non-Stationary Data. Oxford University Press, Oxford.
Baumol, W. and Bowen, W. (1966) Performing Arts – The Economic Dilemma. The Twentieth Century
Found, New York.
Bera, A. K. and Jarque, C. M. (1981) “An Efficient Large Sample Test for Normality of Observations
and Regression Residuals”. Australian National University Working Papers in Econometrics No.
49.
Box, G.E.P. and Pierce, D.A. (1970) “Distribution of Residual Autocorrelations in Autoregressive
Integrated Moving Average Time Series Models”. Journal of the American Statistical Association
65: 1509–1526.
Cameron, S. (1986) “The Supply and Demand for Cinema Tickets: Some U.K. Evidence”. Journal
of Cultural Economics 10: 38–62.
Cameron, S. (1990) “The Demand for Cinema in the United Kingdom”. Journal of Cultural Economics
14: 35–47.
Cuevas, A. (1976) Economı́a Cinematográfica. La producción y el comercio de pelı́culas. Madrid.
Dale, M. (1994) “Esperando al ave fénix. El reto de la industria cinematográfica europea”. Situación
3: 7–52.
Dickey, D.A. and Fuller, W.A. (1981) “Likelihood Ratio Statistics for Autoregressive Time Series
with a Unit Root”. Econometrica 49: 1057–1072.
Dunnett, P.J.S. (1990) The World Television Industry: An Economic Analysis. Routledge, London.
Engle, R.F. (1982) “Autoregressive Conditional Heteroscedasticity with Estimates of the Variance of
United Kingdom Inflations”. Econometrica 50: 987–1008.
Engle, R.F. and Granger, C.W.J. (1987) “Cointegration and Error Correction: Representation, Esti-
mation and Testing”. Econometrica 55: 251–276.
Gapinski, J. (1984) “The Economics of Performing Shakespeare”. American Economic Review 74 (3):
458–466.
Gapinski, J. (1986) “The Lively Arts as Substitutes for the Lively Arts”. American Economic Review
76 (2): 20–25.
Hansen, H. and Juselius, K. (1995) CATS in RATS. Cointegration Analysis of Time Series. Estima,
Evanston.
Heilbrun, J. and Gray, C. M. (1993) The Economics of Art and Culture. An American Perspective.
Cambridge University Press, Cambridge.
Johansen, S. (1988) “Statistical Analysis of Cointegration Vectors”. Journal of Economics Dynamics
and Control 12: 231–254.
Johansen, S. and Juselius, K. (1990) “Maximum Likelihood Estimation and Inference on Cointegra-
tion – with Applications to the Demand for Money”. Oxford Bulletin of Economic and Statistics
52: 169–210.
Jones, D. (1994) “Las empresas de televisión como productoras y exhibidoras de pelı́culas”. Situación
3: 165–177.
Krebs, S. and Pommerehne, W. (1996) “Politico-Economic Interactions of German Public Performing
Arts Institutions”. Journal of Cultural Economics 19 (1): 17–32.
Lévy-Garboua, L. and Montmarquette, C. (1996) “A Microeconometric Study of Theatre Demand”.
Journal of Cultural Economics 20: 25–50.
Mackinon, J. G. (1991) “Critical Values for Co-Integration Tests”. In R.F. Engle and W.J. Granger
(eds.), Long-Run Economic Relationships. Oxford University Press, Oxford, p. 267–276.
Moore, T. G. (1966) “The Demand for Broadway Theatre Tickets”. Review of Economics and Statistics
48: 79–87.
Navarro, A. (1995) “El cine mudo”. Dirigido. No. 237.
Osterwald-Lenum, M. (1992) “A Note with Fractiles of the Asymptotic Distribution of the Maximum
Likelihood Cointegration Rank Test Statistics: Four Cases”. Oxford Bulletin of Economic and
Statistics 54: 461–472.
Owen, J. D. (1969) The Price of Leisure. Rotterdam University Press, Rotterdam.
Phlips, L. (1974) Applied Consumption Analysis. North Holland, Amsterdam.
Throsby, C.D. and Withers, G. (1979) The Economics of Performing Arts. Edward Arnold, London.
Throsby, C.D. (1994) “The Production and Consumption of the Arts: A View of Cultural Economics”.
Journal of Economic Literature XXXII: 1–29.
White, H. (1980) “A Heteroskedasticity-Consistent Covariance Matrix and a Direct Test for Het-
eroskedasticity”. Econometrica 48: 817–838.
Withers, G. (1980) “Unbalanced Growth and the Demand for Performing Arts: An Econometric
Analysis”. Southern Economic Journal 46 (3): 735–742.