Internal agglomeration and productivity: Evidence from microdata

• Author: Evan Rawley,Robert Seamans
• Date: 01 October 2020
• DOI: http://doi.org/10.1002/smj.3200
• Published date: 01 October 2020

RESEARCH ARTICLE

Internal agglomeration and productivity:

Evidence from microdata

Evan Rawley

1

| Robert Seamans

2

1

Carlson School of Management,

University of Minnesota, Minneapolis,

Minnesota, USA

2

Stern School of Business, New York

University, New York, New York, USA

Correspondence

Robert Seamans, Stern School of

Business, New York University, New

York, NY, USA.

Email: rseamans@stern.nyu.edu

Funding information

Ewing Marion Kauffman Foundation

Abstract

Research Summary: We study how internal agglomera-

tion—geographic clustering of business establishments

owned by the same parent company—influences estab-

lishment productivity. Using Census microdata on the

population of U.S. hotels from 1987-2007, we find that

doubling the intensity of internal agglomeration is asso-

ciated with a productivity increase of about 2% in pre-

existing establishments. We consider several mecha-

nisms that may be driving the productivity effect and

find evidence consistent with the idea that an economi-

cally meaningful component of the productivity effect is

due to knowledge transfer between internally agglomer-

ated establishments. We replicate our main findings with

Census microdata on the full population of U.S. restau-

rants from 1987-2007, suggesting that the internal

agglomeration effects we document may generalize

broadly to other industries with multi-unit firms.

Managerial Summary: Internal agglomeration is the

geographic clustering of business establishments

owned by the same parent company. This paper uses

detailed Census data on hotels and restaurants to show

how internal agglomeration influences performance.

Interestingly, knowledge sharing between owned estab-

lishments in the same metropolitan area appears to be

a key driver of the internal agglomeration effect.

KEYWORDS

agglomeration, corporate strategy, knowledge spillovers, productivity

Received: 8 June 2018 Revised: 17 April 2020 Accepted: 20 April 2020 Published on: 22 June 2020

DOI: 10.1002/smj.3200

1770 © 2020 Strategic Management Society Strat Mgmt J. 2020;41:1770–1798.wileyonlinelibrary.com/journal/smj

1|INTRODUCTION

Location decisions are crucial for multi-unit firms. One key element of location choice is

whether to agglomerate internally—cluster business establishments owned by the same parent

firm in close geographical proximity. While previous scholarship has suggested that internal

agglomeration can be important (Alcacer & Delgado, 2016; Darr, Argote, & Epple, 1995), no

large sample studies to date have linked internal agglomeration to establishment-level produc-

tivity effects, even though establishment productivity is known to be a particularly important

determinant of firm performance (Syverson, 2011).

To address this lacuna in the literature, this paper uses establishment-level data from the

Economic Census to evaluate whether there is large sample evidence of an internal agglomera-

tion and productivity effect; and explores, conceptually and empirically, several potential mech-

anisms behind the hypothesized effect. The results show that internal agglomeration is

associated with a 2% increase in productivity. Alternative specifications and increasingly

demanding econometric tests, including matching on observables and instrumental variables

techniques, all find a similar relationship between internal agglomeration and productivity.

Taken together, a series of extensions and robustness checks suggest that an economically

meaningful component of the productivity effect is due to knowledge transfer between inter-

nally agglomerated establishments, though we cannot precisely measure the relative contribu-

tion of all the potential mechanisms at play. We replicate our main findings on the U.S. food

service (“restaurant”) industry 1987–2007, suggesting that our findings generalize to other

industries with multi-unit firms.

The paper makes two key contributions. First, we document a large sample, empirically rig-

orous relationship between internal agglomeration and productivity in two important multi-

establishment industries. Second, we test a number of potential mechanisms behind the inter-

nal agglomeration effect and provide some evidence of intrafirm learning amongst internally

agglomerated establishments. While much of the agglomeration literature focuses on the con-

temporaneous “flow”benefits of agglomeration—those benefits that accrue to the focal firm so

long as establishments are agglomerated—such as complementary demand (Canina, Enz, &

Harrison, 2005; Fabrizio & Thomas, 2012; McCann & Vroom, 2010) and resource sharing

(Feldman, 2000; Folta, Cooper, & Baik, 2006; Krugman, 1991), this paper provides evidence that

persistent “stock”benefits associated with internal agglomeration—those benefits that accrue

to the focal firm, and remain in effect, even if internal agglomeration ceases—are also economi-

cally meaningful.

1

Our analyses are compelling for several reasons. The data are unusually broad and rich—

we have establishment-level data on a complete census of every hotel ever operated (for at least

5 years) in the United States 1987–2007. We use the data to offer relatively precise estimates of

the magnitude and persistence of internal agglomeration. Moreover, by exploiting features of

the empirical setting in our research design—particularly the fact that we study hundreds of

distinct localized markets—we offer results that are robust to a wide range of alternative expla-

nations. Furthermore, we triangulate our quantitative findings with qualitative insights gleaned

from interviews with industry experts, which offer supporting evidence about the nature of the

mechanisms behind the internal agglomeration effects we study.

1

“Contemporaneous”versus “persistent”agglomeration benefits differ conceptually from “static”versus “dynamic”

agglomeration benefits in that the former are defined with respect to a state of being—internally agglomerated or not—

whereas the latter are defined with respect to time (Rosenthal & Strange, 2004).

RAWLEY AND SEAMANS 1771