The economic advantages of chain organization

• Date: 01 December 2017
• Published date: 01 December 2017
• DOI: http://doi.org/10.1111/1756-2171.12214
• Author: Brett Hollenbeck

RAND Journal of Economics

Vol.48, No. 4, Winter 2017

pp. 1103–1135

The economic advantages of chain

organization

Brett Hollenbeck∗

This article considers the rapid spread of chain firms in many industries. The conventional

explanation is that chains generate economies of scale in costs. Alternatively, the structure

of chains may enhance demand by helping firms develop reputations, among other reasons. I

quantify the value of these explanations empirically with a large, detailed data set on the hotel

industry, combining a reduced-form analysis of revenues with a structural estimation of firm

costs. Revenue analysis shows substantial evidence of a large chain premium. Cost estimation

shows that after accounting for unobserved heterogeneity, chain-affiliated firms receive no cost

advantage relative to independent firms.

1. Introduction

Over the past severaldecades, the chain business model has come to dominate most retail and

service industries. A striking growth in shares of sales, employment,and number of establishments

organized as chains has been well documented (Foster, Haltiwanger, and Krizan, 2006; Jarmin,

Klimek, and Miranda, 2009; Foster et al., 2015).1This phenomenon is not just widespread but

has been shown by economists to matter for issues ranging from employment and wages, to

competition, to aggregate productivity.2This broad scope reflects the fact that organizational

form matters for a variety of economic outcomes. The success of the horizontal chain business

model has been striking, especially considering the large fees paid to affiliatewith chains, but what

∗University of California, Los Angeles; brett.hollenbeck@gmail.com.

This article has previously circulated under the title “The spread of horizontal chains: efficiencyor infor mation?” I would

like to thank Eugenio Miravete and Stephen Ryan for their guidance and support. I also thank Allan Collard-Wexler,

Stephanie Houghton, Mike Mazzeo, Sanjog Misra, Peter Rossi, David Sibley, Haiqing Xu, and Rostislav Bogoslovskiy

for their help as well as all participants in the UT-Austin weekly IO Seminar and seminar participants at CalTech,

UCLA, Chicago-Booth School of Management, FTC-Bureau of Economics, DOJ Antitrust Division, Stanford GSB, the

University of British Columbia, the Stanford Institute for Theoretical Economics, and the 2015 Yale Marketing-Industrial

Organization Conference.

1I use “chains” to mean any business that operates multiple outlets offering similar goods or services under the

same banner. The spread of chains can be seen in Figure 1, taken from Fosteret al. (2015), which uses a slightly more

restrictive definition.

2Examples of work on wages and employment include Basker (2005) and Jarmin, Klimek, and Miranda (2009).

For work on competition, see Baskerand Noel (2009) and Jia (2008), among others. Research on agg regate productivity

includes Foster, Haltiwanger, and Krizan (2006) and Bloom, Sadun, and Reenan (2013). This is just a sample of the

available literature, for an overviewof work related to chains and franchising, see Kosova and Lafontaine (2012).

C2017, The RAND Corporation. 1103

1104 / THE RAND JOURNAL OF ECONOMICS

explains this success? The conventional explanationis that fir ms form chains to take advantage of

economies of scale and operate with lower costs. An alternative explanationis that the advantage

to chain affiliation operates through revenues and that the chain structure itself provides this

revenue advantage. These effects have different potential implications for welfare and for the

future of competition in these industries. This article seeks to examine the revenue effects of

chain affiliation in comparison to cost-side effects in the context of the hotel industry in Texas.

A natural explanation for why independent businesses would join together into a chain

network is that this allows them to exploit economies of scale to lower costs and enhance

efficiency. Fixed costs can be spread over more firms, inputs can be purchased with greater

bargaining power and distributed using an efficient network. Firms may have substantially lower

capital costs and insurance costs when affiliated with a known and successful chain. These types

of economies of scale have been shown to be important drivers of the “big box”retail chains that

have been the focus of the extant literature on the economics of chain firms.3

Alternatively, for a given level of costs, chain affiliation maybenefit fir ms byincreasing rev-

enues. In many industries, particularly industries with little repeat business or that sell experience

goods, consumers have very little prior information on the quality of the product or service that

firms offer. If consumers are risk averse, they favor firms whose reputation they know and may

find search and experimentation costly.4Firms in settings where little information is available may

affiliate with one another and operate under one banner to generate a reputation for consistent

quality to attract risk-averse consumers. Firms affiliated with a well-known chain could thus earn

a substantial premium over independent firms offering uncertain quality even if the underlying

product, and hence costs, is identical. In essence, chains may be offering a solution to a lemons

problem by facilitating repeat interactions that could not otherwise occur by providing uniform

services in many settings. This is, in a way, economies of scale on the demand side, as the larger

a chain is, the more consumers have the opportunity to interact with it, increasing its value. The

same effect could be seen as firms invest in national advertising to develop a reputation over time.

Finally, information learned by firms on revenue management strategies and shared within chains

might also increase revenues, as could access to centralized booking systems.

This article is the first to consider in depth this demand-side explanation for chain affiliation.

I examine both cost and revenue explanations empirically over time within the same industry and

attempt to answer the question: which of these is primarilyresponsible for the success and spread

of the chain model in this nonretail setting? The answer is of interest for several reasons. First,

although cost-side advantages have been shown in large scale retail, it is unclear if the results

from that literature hold outside that industry. Second, the implications for consumers may differ.

Success due to lower costs are unambiguously positive for consumers, improving competition

and lowering prices. Where the chain model is successful mainlydue to revenue effects, however,

the effects are more ambiguous. Consumers may find themselves paying higher prices for the

same quality good, but they may also find more high quality goods available.

Third, the answerhas implications for competition in the future. Over the past decade, online

review and rating websites have allowed consumers to document and share their experiences with

almost every conceivable type of firm.5Many industries are therefore transitioning from a very

low-to a relatively high-information environment, and the path this transition will take depends

on what has driven the chain model’s success to this point.

3See, for instance, Holmes (2011) and Jia (2008), who show the importance of distribution costs to Wal–Mart’s

expansion strategy.

4Throughout, I use the term “reputation” to denote the signal a firm’s brand name provides to consumers. This

signal should at minimum reduce variance in a consumer’s prior over firm quality,something risk-averse consumers value.

It may also increase or decrease mean expected quality. In a seminal article, Kreps (1990) discussed the importance of

firm reputation, Horner (2002) further examines the role of firm reputation in competition, and Cai and Obara (2009)

explicitly model reputation building in low-information environmentsas an incentivefor horizontal expansion.

5According to the Local Consumer ReviewSur vey2012, 85% of consumers checked online reviews before making

purchasing decisions in 2012.

C

The RAND Corporation 2017.

HOLLENBECK / 1105

This article uses a unique firm-level data set on the hotel industry to examine empirically

the nature of chain affiliation and the relative contribution of each factor to firm profits. Two

features of these data are particularly useful. I observe the full population of firms; this allows

a direct comparison between chain-affiliated and independent firms rather than simply relying

on observations of a single chain like McDonald’s or Walmart. In addition, I observe firm

revenue, making it possible to separate demand-side factors from cost-side factors. These data is

supplemented with online reviews data and market-level demand shifters.

The lodging industry is an ideal setting to examine the questions presented above for several

reasons. First, hotels compete in a large number of geographicallydistinct markets. Second, unlike

retailers or firms in other service industries, they offer close to a single product, a night’s stay in a

room. This product is differentiated betweenfirms almost entirely on universallyag reed-on quality

and not other factors. These result in a relatively straightforward problem with few confounding

factors. Third, the trend toward chain firms in this industry closely matches the aggregate trend.6

Finally, franchising and the hotel industry are significant sectors of the economy in their own

right, with hotels generating $196 billion in sales in 2012 and employing 2 million individuals

according to the 2012 Economic Census, and franchises were responsible for $1.3 trillion sales

(9.2% of Gross Domestic Product) and 7.9 million employed (Kosova and Lafontaine, 2012).

The empirical strategy combines a structural estimation of firm costs with a detailed, reduced-

form examination of hotel revenues to quantify the value to firms of chain affiliation, and the

share of this value that comes through lower costs versus higher revenues. To do so, I model the

dynamic game of market entry, exit,and type choice, and use these decisions to identify sunk and

fixed costs. In the first-stage estimation of revenue, I examine the impact on revenues of hotels

that add or drop affiliation during the sample period. I find that conditional on firm and market

characteristics, chain-affiliated properties earn 20% higher revenue per room than independent

firms. This is a substantial advantage and it is robust to a variety of specification tests.

I then examine the conventional explanation that the chain-affiliatedfir ms are more efficient

than independent firms by examining their operating costs. These costs are unobserved; instead,

I estimate a dynamic model in the style of Arcidiacono and Miller (2011) to recover the cost

structure of the different firm types to examine what cost advantage is associated with chain

affiliation. This is one of the first applications of the Arcidiacono-Miller estimator, which allows

for flexible and persistent unobserved heterogeneity, and I show that allowing for this significantly

changes the results from a more restrictive estimator.7The dynamic model produces realistic

estimates of firm operating costs. They suggest, however, that chain firms gain no cost advantage

from their affiliation after controlling for quality and unobserved market-level heterogeneity. I

solve and simulate the model using the estimated parameters and find that it fits well.

I next examine patterns in revenue to consider how organizing as chains generates this

revenue advantage. Potentialsources of the increase in revenue include reputation effects, loyalty

effects, and centralized bookings. Tests of revenues are consistent with a variety of predictions

of a model where chains signal quality to consumers with low information. The chain premium

declines over the past decade as online reputation mechanisms become more widely used. The

chain premium appears immediately when a firm joins a chain, as opposed to slowly phasing in,

and it is positivelycor related with chain size. Finally, online reviewsdata show strong correlations

between customer information and independent firm success, and among firms with large numbers

of online reviews, the chain premium disappears completely. Although other explanations such

as loyalty programs or centralized bookings likely play a role as well, these patterns suggest at

minimum that a large share of the revenue premium is due to reputation effects.

A growing literature addresses the spread of chains. In part this reflects the success of

large retailers such as Walmart (Basker, 2005; Foster, Haltiwanger, and Krizan, 2006; Holmes,

2011; Jia, 2008; Ellickson, Houghton, and Timmins, 2013; Foster et al., 2015; Zheng, 2016).

6This can be seen by comparing Figures 1 and 4.

7Scott (2013) also employs this estimator in a study of agricultural land use.

C

The RAND Corporation 2017.