1.5 Methodology: the econometrics of cinema capacity utilisation
 
The efficiency of a cinema site can be measured by seat utilisation rate. We define seat utilisation rate as the number of admissions per seat.
 
We can divide the explanatory variables into three major blocks:  
Country specific factors
 
Table 32 displays the annual average cinema visit per person in 1992. The frequency varies considerably across countries, ranging from 0.9 in the Netherlands to 2.2 in Ireland.
 
Table 32 
Frequency of Cinema Visits and Density of Screens, 1992
Country
Annual Average Number of Visits per Person
Density of Screens 000s of inhabitants per screen
Density of Population (number of inhabitants per km²)
Belgium
1.6
23.3
327
Denmark
1.7
16.4
122
France
2.1
13.0
104
Germany
1.3
22.1
237
Greece
-
25.4
78
Ireland
2.2
18.7
51
Italy
1.4
19.1
192
The Netherlands
0.9
36.4
444
Portugal
1.2
42.4
106
Spain
2.1
21.8
77
UK
1.8
33.0
236
SOURCE: MEDIA Salles, European Cinema Yearbook, 1993
 
As the frequency of cinema visits rises, the number of admissions per seat should increase. An increase in screen density should also lead to a higher seat utilisation rate. Both screen density and frequency of cinema visits vary across the 11 countries sampled, as illustrated in Table 32. We need to control for these different factors using dummy variables. We created 11 dummy variables - C1 to C11 - with each variable representing a different country.
 
 
Location specific factors
 
Location should play an important part in the popularity of a cinema. The larger the catchment area, the higher the utilisation rate is likely to be. For larger towns, it is important to distinguish between screens in town centres and screen in suburban areas; we might expect centre of town locations to be more popular than those in suburban locations.
 
We created 8 dummy variables that control for location effects. These are:
 
PP1 Population less than 25,000 
PP2 Population between 25,000 and 50,000
PP3 Population between 50,000 and 100,000
PP4 Population between 100,000 and 250,000 
PP5 Population between 250,000 and 1 million; 
cinema in town centre 
PP6 Population more than 1 million; 
cinema in town centre 
PP7 Population between 250,000 and 1 million; 
cinema in suburbs 
PP8 Population more than 1 million; 
cinema in suburbs 
 
Site specific factors
 
We have identified six categories of site specific factors that may affect the seat utilisation rate at a cinema.
 
 
Ancillary services
 
Ancillary services are those services provided by the cinema that do not directly affect the screen but may make the cinema more attractive to customers. We have defined seven dummy variables:
 
PR1 Vending machine for drinks/confectionery 
PR2 Drinks/confectionery for sale in the auditorium 
PR3 Coffee bar 
PR4 Restaurant 
PR5 Bookshop 
PR6 Poster shop 
PR7 Video shop 

 
Share of domestic films
 
A cinema programme generally contains films that belong to one of three categories: domestic films, US films and films from the rest of the world. In general, we might expect a high share of US films to enhance the popularity and thus seat utilisation in a cinema. We have therefore defined two variables: FD and FU. FD represents the percentage of domestic films in the film programme, and FU represents the percentage of US films.
 
 
Parking availability
 
Availability of parking adjacent to a cinema will enhance the popularity of a cinema vis-à-vis cinemas without parking facilities. We include two dummy variables:
 
PK1   Availability of a reserved car park for cinema customers
PK2 Availability of a free car park or a car park with preferential rates for cinema customers 
This is against the base case of no parking facilities.
 
 
Number of screens
 
The variable SC tells us whether the film is shown in a single or multi-screen cinema. There are no strong a priori assumptions about this variable.
 
 
Advance booking facility
 
An advanced booking facility may increase seat utilisation rates in two ways. Firstly, it removes the uncertainty experienced by the customer when visiting the cinema. Customers may be more reluctant to visit the cinema if they think that they may not be able to view their preferred film. Hence the availability of advanced booking facilities may make the cinema a more popular choice. Secondly, an advance booking facility gives the cinema the ability to assess the popularity of various films. This may lead to an increase in efficiency: the cinema will be better placed to match films screened to customers' preferences. A dummy variable, AB, was used to capture this effect.
 
 
Years since modernisation
 
We might expect a recently modernised cinema to be more popular, other things being equal, and so to have higher seat utilisation rates. The variable MOD was used to capture this effect. It must be noted, however, that this variable will not show the impact of a brand new cinema, since no recent improvement is recorded in this case.
 
 
A Two Week Sample
 
The data set
 
We have data on a variety of screen and cinema characteristics, as described above, from 11 European countries, taken from a one week sample period.
 
Some of the observations were inconsistent and were dropped from the sample. A further number of observations has weekly performance figures that did not seem realistic. We reduced the sample further to include those observations with a weekly performance figure of 35 performances per week. The sample size was 928.
 
We dealt with the effect of the number of screens in two ways. The first was to simply include the logarithm of the number of screens. The coefficient gives us an elasticity, but as the variable is bound between 1 and 14 we ought to be careful in interpreting that coefficient. The second method was to treat 1-2 screens as a base and include two dummies, MD1 for 3 or more screens and MD2 for 8 screens or more.
 
 
Results
 
Using admissions per screen as the dependent variable, we obtained the results shown below. The LM test for functional form was passed but the test for heteroscedasticity failed. The t-statistics were then corrected using Whites heteroscedasticity consistent standard errors. Table 33 is based on 928 observations and uses the logarithm of screens.
 

Table 33: 
Results of Small Sample Using Ln of Screens 
Variable
Co-efficient
T-statistic
Constant
-1.309
-9.331
Ln (Sc)
.204
5.065
Ln (Performance)
.790
16.526
C4
.337
4.461
C5
-.193
-2.306
C8
.643
3.883
C11
-.198
-2.153
PP6
.325
4.653
PR4
.333
2.183
PR7
.238
2.606
PK2
.130
2.457
Ln (MOD)
-.093
-2.903
R2=.404 Adjusted R2= .397 Standard Error= .728
   
Table 34: 
Results of the Regression with larger sample, and Multiplex Dummy 
Variable
Coefficient
T-statistic
Constant
-1.303
-9.434
3+ screens
0.190
3.076
6+ screens
-0.90
-0.827
8+ screens
0.692
3.565
Ln(performances)
0.821
17.281
C4
0.239
3.232
C5
-0.189
-2.392
C8
0.619
3.829
C11
-0.283
-3.049
PP6
0.273
3.911
PR6
0.216
2.480
PK2
0.122
1.500
Ln(modernisation)
-0.094
-2.995
R2 = .413
Adjusted R2 = .405
Standard Error = .726
 
Table 34 shows the results of using the larger sample with a cut off of 45 weekly performances and uses the multiplex dummy variables instead of the logarithm for screens.
 
The coefficients for the 3+ and 8+ screen dummies are clear evidence for the beneficial effects of multiple screens, and the "multiplex" effect for cinemas with 8 or more screens. The 3+ dummy says that capacity utilisation is significantly higher in cinemas with 3 or more screens. The 8+ dummy says that there is a benefit from having 8 or more screens, over and above the benefit of having more than 2 screens. The 6+ dummy is, however, not significant. The "multiplex" effect kicks in only with 8 or more screens, not with 6 or 7.