Effects of Chain Affiliation in the Movie Theater Industry

Abstract

In this paper, I empirically study the effect of chain affiliation on product variety and price in the movie theater industry. Using longitudinal data on Korean movie theaters, I find that movie variety in a theater increases by 3.2–5.5% after the theater joins a chain. Admission price, however, does not change after chain affiliation. These findings imply that consumers may benefit from the growing dominance of chain-affiliated theaters in recent decades. The results also suggest that the regulatory authorities should carefully examine the trade-off between the increase in market power and efficiency gains when evaluating the implications of chain affiliation.

Keywords:

• Organizational form
• franchising
• product variety

JEL Classification Numbers:

• K20
• L22
• L82

Acknowledgments

I am grateful to seminar attendants at the 2019 International Industrial Organization Conference and 2019 MSSE April Workshop.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Data Availability Statement

The data for the key empirical findings of the paper are available at GitHub (https://github.com/kimik0121/chain_affiliation).

Code Availability Statement

STATA code available upon request.

Notes

1 Franchise business economic outlook for 2018 (https://www.franchise.org/sites/default/files/Franchise_Business_Outlook_Jan_2018.pdf).

2 Specifically, a franchisor should not open a new bakery within 500 m of any independent bakery or its own incumbent, while the number of new bakeries in a year should not exceed 2% of the total number of its incumbents. In 2011, there were 4.8 thousand franchising bakeries in Korea and most of them (97%) were franchisee-owned.

3 In this paper, ‘movie’ implies ‘movie title’.

4 Given the trade-off between movie variety and the number of showings or seats for each movie, one may worry that chain affiliation can cause more consumers to be turned away from theaters especially on weekends. However, since they can always choose another time or date, consumers care more about the availability of a movie than the availability of a seat for the movie.

5 The positive consequences of the spread of chain theaters are well recognized in Korea. For instance, see the following news article ‘Screen buffet changed the movie world’, Weekly donga, September 25, 2003 (http://weekly.donga.com/List/3/all/11/72107/1).

6 See Ashenfelter et al. (2014) for an overview.

7 Primus was owned by CGV.

8 Table A3 in the Appendix shows that chains operated 174 theaters with 1336 screens out of 314 theaters with 1975 screens in total in 2007.

9 Source: MPAA Theatrical Market Statistics 2015, Korean Film Council Annual Report 2015.

10 The weekly average price is adjusted using monthly regional CPI.

11 During the sample period, Korea was made up of 16 first-tier administrative divisions: seven metropolitan cities and nine provinces. Each metropolitan city is subdivided into ‘gu's. For example, there are 25 gus in Seoul whose average surface area is 24.2 km${}^2$. The nine provinces have several administrative subdivisions called either ‘si’ or ‘goon’ depending on population size.

12 Alternatively, I consider the local market to be within a one-mile radius of each theater. The estimation results (available from the author upon request) are qualitatively the same as those reported in Section 3.

13 The data analyzed in this paper come from the following sources: the Korea Box Office Information System (http://www.kobis.or.kr/), KOFIC (http://www.kofic.or.kr/), and the Korean Statistical Information Service (http://kosis.kr/eng/)

14 The number of first-run movies available for screening in a given week ranges from 35 to 132 during the sample period.

15 For instance, a supermarket chain may sell private brand products.

16 1 is added to the number of competitors' screens before taking log, because there are theaters without any competitors in the local market. Ninety theaters opened, and 53 theaters closed during the sample period. There were also 21 cases where an operating theater's screen schedule information was missing for two months or longer. I define such a case as a temporary shutdown, which occurs when a theater renovates its rooms, replaces equipment, or changes its organizational form. A temporary shutdown accompanies 5 out of the 21 organizational changes. Estimation results are not affected when more or less strict definitions of the temporary shutdown are applied.

17 On one hand, an incumbent retailer may reduce its product range to soften price competition after the entry of a competitor. On the other hand, heightened competition may induce the retailer to increase the product range so that it can attract more consumers.

18 There are 16 regions (16 first-tier administrative units) and 36 months (12 months a year × 3 years) in the data.

19 As a matter of fact, in 16 out of the 21 cases, the theater kept opening for business without any closing upon or after the event.

20 Instead, they are considered in the analysis of the market-level implication of chain affiliation in Section 3.3.

21 I estimate the model with robust standard errors clustered by theater.

22 The results are qualitatively the same when the number of competitors is used as a measure of the degree of competition instead of their screen counts.

23 For instance, the sales rate of a movie (the number of tickets sold for a movie divided by total number of seats allocated to the movie nationwide) is 13% on average with a maximum of 23% in Korea during the sample period.

24 The two cases where a chain theater became independent are not considered in this analysis.

25 There are 36 months (12 months a year × 3 years) in the data.

26 As described in Subsection 2.1, I observe the admission price at each theater screen at each ti me slot during a day.

27 Standard errors are assumed to be clustered by market.

28 There are two markets where two independent theaters became chain-affiliated. Here, I disregard the second chain affiliation in these markets. Estimation results are qualitatively the same, when I let $MKTChain$ take the value of two after chain affiliation of the second theater.

29 As noted in Section 2 of the manuscript, information on the organizational form of chain theaters, either company-owned or franchised, is available for the end of 2008.

30 There were 93 company-owned theaters at the end of 2008. I use them in the construction of the instrumental variable.

31 Specifically, I use the average of the weekly movie varieties and the average of weekly admission prices in December 2008.

32 Indicators for opening and closing weeks, and weeks right before and after a temporary shutdown are not included in the model, as there are no such cases in December 2008. Also, due to data availability, I use demographic information for each local market in 2010.

33 In the estimation of the model, I assume that standard errors are clustered by market.

34 Specifically, I first estimate a probit model and obtain the fitted probabilities, $\widehat{Franchisee}$, and then estimate model (A1(A1) $\begin{array}{r}{y}_i=\delta Franchise{e}_i+{\mathbf{x}}_i\lambda +u_i,\end{array}$(A1) ) by 2SLS using $\widehat{Franchisee}$ as an instrument. I estimate and report the usual 2SLS standard errors (clustered by market), as they are known to be asymptotically valid (Wooldridge, 2010). Bootstrap estimates of the standard errors (200 cluster bootstrap replications) are qualitatively the same.

Additional information

Notes on contributors

In Kyung Kim

In Kyung Kim is an assistant professor of economics at Kyungpook National University. His research focuses on industrial organization and quantitative marketing.