Productivity Growth in the Retail Sector:
Entry, Exit, and Reallocation
Janghee Cho1, Hyunbae Chun2, and Yoonsoo Lee3
March 14, 2014
Abstract
The massive restructuring and reallocation driven by the entry of large discount stores
substantially contributed to the productivity growth in the retail trade sector. We measure the
entry, exit, and reallocation process and its contribution to the productivity growth in the Korean
retail trade sector. We find substantial heterogeneity in the effects of large entrants on
productivity dynamics. Entry of small and specialized stores (i.e., non-GMS) was more active in
the counties where a large discount store opened. Moreover, new entrants in counties with a
large discount store are more productive than those that entered in counties without a large
discount store. However, the productivity increase was not uniformly observed among all new
entrants. We find that such productivity gains in new entrants are concentrated among small,
chain stores, which shares the new, modernized format of the retail service. Our finding suggests
that opening of a large discount store may stimulate the reallocation process in the local retail
sector, by not just driving out less efficient stores but also attracting new, modernized stores into
the neighborhood.
JEL Classification: L81, O47, R12
Keywords: Entry and Exit, Large Discount Stores, Productivity, Reallocation, Retail Trade
1
2
3
Department of Economics, Sogang University, Seoul 121-742, Korea. Tel: +82-2-705-8266. Fax: +82-2-704-8599.
E-mail: [email protected].
Department of Economics, Sogang University, Seoul 121–742, Korea. Tel: +82–2–705–8515. Fax: +82–2–704–
8599. E-mail: [email protected]. Corresponding author.
Department of Economics, Sogang University, Seoul 121–742, Korea. Tel: +82–2–705–8516. Fax: +82–2–704–
8599. E-mail: [email protected].
1
1. Introduction
The massive restructuring and reallocation driven by the entry of large discount stores
substantially contributed to productivity growth in the retail sector. Recent studies find that
productivity growth in the retail sector shows quite a different pattern from that in manufacturing.
While innovation and technological progress is a key factor for productivity growth of existing
and continuing plants in the manufacturing sector, efficient resource reallocation between new
entrants and existing stores account for most of productivity growth in the retail sector. As
resources from less productive stores are reallocated to new stores with higher productivity, the
aggregate productivity increases. For example, Foster, Haltiwanger, and Krizan (2006) find that
entry and exit account for about 30% of labor productivity growth in manufacturing between
1977 and 1987. In the retail sector, they find that virtually all of productivity growth between
1987 and 1997 is contributed to entry and exit. In this reallocation process, entry of large
discount stores with new technology, such as scanners and rapid credit card processing, plays an
important role. However, in studies focusing on other countries, the evidence on the role of
large discount stores in productivity growth is not very strong (e.g., Haskel and Sadun (2009) on
the U.K. and de Vries (2008) on Brazil). Such differences across countries may be due to the
degree of regulations and the effectiveness of reallocation and market selection in promoting
productivity growth.
This study provides new evidence of the role of reallocations in productivity growth in the
Korean retail trade sector. So far relatively few studies have examined productivity dynamics in
2
the retail sector (Foster, Haltiwanger, and Krizan, 2006; Maican and Orth, 2013). 1 Using
establishment-level micro data from the Economic Census in 2005 and 2010, we measure the
entry, exit, and reallocation process and their contributions to productivity growth in the Korea
retail trade industry. Along with the rapid expansion of big-box stores since the mid-1990s, the
Korean retail sector underwent significant structural changes. Before large chain discount stores
were introduced, the retail sector in Korea was long dominated by small shops in traditional
market districts and small and medium-sized local supermarkets. Since the first large discount
store (E-mart) opened in Seoul in 1993, large discount stores and small and medium-sized chain
stores such as convenient stores and super-supermarkets (SSM) have rapidly expanded all over
the country. These new stores, the first modern format of national retailer chains with
modernized shopping infrastructures and advanced technologies, led structural change and
productivity growth in the retail industry.
We contribute to the growing literature of productivity dynamics by providing new insights
on the role of reallocations. While we share the conclusion of existing studies that the
reallocation process is at the core of productivity gains in the retail industry, we find the
underlying mechanism is quite different. In particular, we emphasize the role of large stores in
inducing entry of different types of stores. So far, previous studies focusing on productivity
dynamics have found that increased competition from large entrants with higher productivity
forces low productivity incumbents to improve their productivity and induces exit of low
productivity firms. For example, Maican and Orth (2013) find that large entrants force low
productivity stores out of the market and increase productivity among survivors.
1
Foster,
So far most studies on the large discount stores have focused on the impact on the labor market. For example,
see Basker (2005), Neumark, Zhang, and Ciccarella (2008), Jia (2008) and Sobel and Dean (2008) for the U.S.,
Rivero and Vergara (2008) for Chile, Igami (2011) for Japan, and Schivardi and Viviano (2011) for Italy.
3
Haltiwanger, and Krizan (2006) find that the dominant role of net entry is associated with the
entry of more productive stores that are part of large, national chains. The restructuring process
of large, national chain stores replacing less productive, independent stores is the key factor in
explaining the he productivity gains in the U.S. retail industry.
We find substantial heterogeneity in the effects of large entrants on productivity. In contrast
to the finding in the U.S., we find that the net entry effect in the Large GMS (General
Merchandise Stores) sector, to which large discount stores belong, is negative. On the other
hand, in the group of Other Retails, which include very small (1-4 workers), small (5-9), and
medium (10 or more) sized GMS and all non-GMS stores, net entry effects, in particular entry
effect, account for the majority of productivity growth. Further analyzing the heterogeneous
net-entry effects, we find that entry of small and specialized stores (i.e., non-GMS) was more
active in the counties where a large discount store opened. This finding suggests that a large
discount store may have a spillover effect on the local retail sector, thereby attracting small,
specialized shops in the neighborhood. Moreover, new entrants in counties with a large discount
store are more productive than those that entered in counties without a large discount store.
However, the productivity increase was not uniformly observed among all new entrants. We
find that such productivity gains in new entrants are concentrated among small, chain stores,
which shares the new, modernized format of the retail trade.
Our finding suggests that opening of a large discount store may stimulate the reallocation
process in the local retail sector, through not just driving out less efficient stores but also
attracting new, modernized stores into the neighborhood. Our finding that large entrants may
encourage the entry of new stores is a new perspective of the structural changes in the retail
industry. Previous studies overlooked such a dynamic complementarity that exists between a
4
large discount store and new entrants that benefit from modern shopping infrastructure. A large
discount store provided convenient, modern shopping amenities, such as indoor shopping areas
with air conditioners or heaters, food courts, and convenient parking, which are shared by small,
specialized shops. 2 The finding of higher productivity of entrants suggests that the selection
process driven by the large discount stores may work at the entry margin as well as at the exit
margin. The selection process is dynamic in a sense that it takes time to attract new stores to the
neighborhood.
Our paper contributes to another strand of literature focusing on the development. Detailed
micro data on all retail establishments in Korea provide a unique opportunity to explore the
impact of modern large discount stores on the retail sector in a developing county. The ongoing
structural change in retail trade is a worldwide phenomenon. However, the pattern of structural
change and its impact may vary across countries depending upon different development stages.
The spillover effect of the entry of big-box stores in developing countries may occur beyond the
retail sector such as agriculture and manufacturing sectors. For example, Iacovaone et al. (2011)
find that the diffusion of Wal-Mart in Mexico resulted in product upgrading by upstream
manufacturers.
Javorcik and Li (2013) also find that the expansion of global retail chains
increased the productivity in the supplying manufacturing industries in Romania. Our study on
the impact of large discount stores on modernization of the retail sector and its productivity
growth will contribute knowledge on the ongoing evolution of the retail sector around the world.
The remainder of this paper is organized as follows. Section 2 provides a brief background of
2
In a study examining the impact of discount store entry on the local supermarket, Zhu, Singh, and
Dukes (2011) show that the entry generates positive demand externality to the incumbents located in the
same shopping plaza, by attracting consumers. However, the positive externality created by the entry of
large discount store in Korea is not just limited to an increase in the traffic. The entry of large discount
store accompanied the build-up of a modern shopping plaza, which attracted small, specialized shops.
5
large discount stores in Korea and the data. Section 3 explains the productivity decomposition
methods and the results. Section 4 provides further analysis on the net entry effects across
different industries and size groups. Section 5 concludes.
2. Large Discount Stores and Modernization of the Retail Sector in Korea3
We utilize establishment-level micro data from the 2005 Census of Service Industry and 2010
Economic Census. Statistics Korea conducts a survey of all establishments with at least one
worker every five years, collecting data on the kind of business, location, sales, and employment.
[need update the data description in more detail]
Information on the locations and opening dates of large discount stores is obtained from the
Yearbook of Retail Industry published by the Korea Chain Stores Association.4 A typical large
discount store in Korea shares the format similar to a hypermarket or superstore because food
products, including fresh food, comprise approximately 50% of store sales. Thus, discount stores
play the roles of both supermarkets and discount stores that sell general merchandises at low
prices. In order to focus on the impact of modern, large retail chains, we include only national
chains with at least 10 stores in 3 provinces or more (among the total of 16 provinces in Korea).
Therefore, we classified the following seven brands as large discount stores: E-mart, Homeplus
(Tesco), Lotte Mart, Hanaro, Wal-Mart, Homever (Carrefour), and Aram Mart. Not all the
foreign retail transnational corporations (TNCs) that entered Korea were successful. For example,
both Carrefour (which entered in 1996) and Wal-Mart (which entered in 1998) failed to attract
local customers and withdrew from the Korean market in 2006; on the other hand, Tesco, a late
3
See Cho, Chun, and Lee (2013) for the detailed description on the diffusion of large discount stores in Korea.
According to Korean Law on the retail industry, a retail chain store is classified as a “large discount store” if it
operates in an area that is over 3,000 square meters and sells items at lower prices than small retail stores.
4
6
entrant (allied with Samsung in 1999) became one of the three leading discount store chains.5 In
2010, these seven national chains accounted for over 95% of all large chain discount stores. Most
of these chains were active during our sample period between 1997 and 2010; however, WalMart was merged to E-mart in 2006 and Aram Mart and Carrefour were merged to Homeplus in
2005 and 2008, respectively.
Before large chain discount stores were introduced in the mid-1990s, the retail sector in
Korea was long dominated by small shops in traditional market districts and small and mediumsized local supermarkets. In contrast to the experience in advanced countries, neither national nor
regional chains of supermarkets (as well as specialized retailers) were established in Korea at the
time when large discount stores were introduced. This is in sharp contrast with the case of WalMart that competes with both incumbent chain retailers as well as mom-and-pop stores. Since the
first E-mart store opened in Seoul in November 1993, large discount stores have rapidly
expanded all over the country. The growth of national retail chain stores is one of the key
features of the evolution of the Korean retail industry during the past two decades. The entry of
large discount store usually accompanied the build-up of a modern shopping plaza, which
attracted small, specialized shops. In 2010, over 60% of the counties in the country had at least
one large discount store.
3. Productivity Dispersion, Reallocation, and Productivity Growth
3.1 Productivity Dispersion and Reallocation
In this section, we present basic facts about the shape and evolution of the productivity
5
See Coe and Lee (2013) for the detailed discussion about the success and failure of global retail chains in Korea
and their strategies.
7
distribution across establishments. Following Foster, Haltiwanger, and Krizan (2006) we
examine the percentiles of the labor productivity distribution across businesses after removing
four-digit industry fixed effects.
In order to analyze the dynamics of establishment-level
productivity, Table 1 presents the transition of individual stores in the distribution of productivity
between 2005 and 2010. In each year, establishments are classified into quintiles of the labor
productivity distribution. The table shows where the establishments in 2005 end up in 2010 in
the productivity distribution and where the establishments in 2010 came from (in italics).
[Table 1 about here]
Foster, Haltiwanger, and Krizan (2006) document the important role of entry and exit in
the transition of productivity distribution. We find that entry and exit rates are at about 52% in
the Korea retail sector, even higher than those found in the U.S. While Foster, Haltiwanger, and
Krizan find that exiting stores are concentrated in lower quintiles, exiting stores in Korea are
uniformly distributed throughout the five quintiles. On the other hand, entrants are more likely
to arrive with relatively higher productivity. 57.4% of stores in the highest quintile in 2010 are
new entrant, while 42.6% stores in the lowest quintile are new entrants.
3.2 Productivity Decomposition
Labor productivity of the retail sector in Korea grew at 4% per year on average between 2005
and 2010. Using establishment-level data, we examine the extent to which entry, exit, and shifts
in the share of inputs across stores affect the aggregate productivity growth. Following Foster,
Haltiwanger, and Krizan (2006), we decompose the changes in industry-level productivity (𝑃! ! )
into components that reflect a within-store effect and other effects that reflect the reallocation of
shares across stores including the effect of entry and exit:
8
∆𝑃! ! =
𝜃! !!! ∆𝑃! ! +
!∈!"#$
+
𝑃! !!! − 𝑃! !!! ∆𝜃! ! +
!∈!"#$
∆𝜃! ! ∆𝑃! !
!∈!"#$
!∈!"# 𝜃! ! 𝑃! ! − 𝑃! !!! −
!∈!"# 𝜃! !!! 𝑃! !!! − 𝑃! !!! ,
where Cont denotes continuing stores, Ent denotes entering stores, and Ext denotes exiting stores.
The first term in the equation reflects changes in productivity from continuing stores, holding
output shares fixed (often interpreted as a “within” effect). The second term reflects changes in
output shares from continuing stores for fixed levels of productivity (often interpreted as a
“between” effect) and the third term represents a cross term that shows whether stores with
positive productivity changes are more likely to have increased employment or not. The last two
terms represent the contribution of entering and exiting stores, respectively. These two terms,
together constituting the net entry effect, along with the between and cross effects, represent the
effect of reallocations across stores on aggregate productivity changes.
[Table 2 about here]
Column [1] of Table 2 reports the productivity decomposition results for the entire retail
(e.g., Total Retail).
As in Foster, Haltiwanger, and Krizan (2006), net entry play an important
role in productivity growth, accounting for more than 2/3 of productivity gains.
In a study
analyzing the impact of large discount store entry on local retail employment, Cho, Chun, Lee
(2013) find that it is important to distinguish the spillover effect of the entry of large retail chain
stores on other retail industries from the direct effect on the industry that large discount stores
belong to. Following the method, we divide Total Retail into two groups: i) Large GMS
including large discount stores and department stores and ii) Other Retail including small (1–9
employees) and medium-sized (10–49 employees) GMS (mom-and-pop groceries, convenience
stores, and supermarkets) as well as all non-GMS stores (e.g., clothing, electronics, bakery,
9
butcher shops, etc.).
Column [2] reports the results for the group of Large GMS, which
accounts for 6.6% of employment in Total Retail and Column [3] reports the results for Other
Retail, which accounts 93.4% of employment in Total Retail.
Examining the productivity
growth separately, we find that most of productivity growth comes from Other Retail. In fact, net
entry effects in the Large GMS is negative, which suggests that entrants in the Large GMS have
lower productivity that existing stores in the same industry.
This finding is due to the fact that
the entrants in our sample is late comers: most of large discount stores have already entered the
county by 2005, well before the sample period we examine. On the other hand, net entry plays a
critical role in the productivity growth in Other Retail. In particular, the contribution of entry
accounts for virtually all of net-entry effect.
4. Further Analysis of Net-Entry Effect
[Incomplete and in progress]
4.1 Entry in counties with large discount store
In the previous section, we find that understanding the productivity gains in Other Retail is
crucial to understand the productivity dynamics in the retail sector. In this section, , we quantify
the net entry effects in a regression context and explore the extent to which net entry effects vary
across store sizes and locations. In particular, we focus on the role of large discount stores in the
reallocation process and examine the difference in net entry effects between counties with a large
discount store and other counties. That is, we consider a simple regression of labor productivity
on a set of dummies that account for the characteristics of the store (e.g., entrants, exiting stores,
continuing stores, size, and industry) and its location (e.g., the presence of large discount store).
In 2010, over 60% of the counties in the country had at least one large discount store. By
10
providing modern shopping infrastructure and attracting new small stores into the neighborhood,
large discount stores transformed local retail sectors away from traditional shopping
environments in the county. In Figure 1, we find that there exists a noticeable difference in
productivity between entrants in counties with a large discount store (LDS Area) and those in
other counties (non-LDS Area).
[Figure 1 and Table 3 here]
In order to analyze the role of entry and exit in productivity dynamics, we consider a
simple regression of labor productivity on a set of dummies indicating the status of the
establishment (continuing, entering, or exiting), the presence of large discount store in the county,
the year dummy (2010) to control the effect of the recession in 2010, industry, and provinces.
Table 3 reports the summary statics for the variables used in the regression and the specification
is given by:
ln 𝐿𝑃
!"
= 𝛼 + 𝜏! 𝑌𝑅2010! +
𝛽! 𝐶𝑜𝑛!" ∙ 𝑀 𝑘
!∈!"
+
𝛿! 𝐸𝑛𝑡!" ∙ 𝑀 𝑘
!
+
!∈!"",!"
, 𝑀 𝑎𝑙𝑙
!
+
𝛾! 𝐶𝑜𝑛!" ∙ 𝑌𝑅2010! ∙ 𝑀 𝑘
!
!∈!"
𝜋! 𝐸𝑥𝑡!" ∙ 𝑀 𝑘
!
+
𝜑! 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦! +
𝜃! 𝑃𝑟𝑜𝑣𝑖𝑛𝑐𝑒! + 𝜀!"
!∈!"",!" !
= 1, 𝑀 05
!
= 1 if Large retail store entered before 2005 ; 0 if otherwise.
The results of the base model, reported in Table 4, show that entrants are more
productive than continuing stores. Exiting stores are less productive once we control for cities
and provinces. In Panel B, we report the results from the model with interaction with locations
with large discount stores.
Overall stores in counties with large discount stores are more
productive. Entrants in counties with large discount stores are also more productive than those
11
in other areas.
[Table 4 about here]
A natural question is what makes entrants in counties with large discount store are more
productive. Given that large discount store brings new, modern shopping infrastructure to the
town, it is interesting to examine which entrants get the benefit from the modernized shopping
environment. To explore the productivity differences across the types of entrants (in terms of
size and national chains) are, we regress the entrants on productivity on dummies indicating the
size and the interactions with the location dummies. We also interact with dummies indicating
the national chains versus single-unit stores, to explore the role of national chains.
ln 𝐿𝑃
!"#
!
=𝛼+
𝛽!" 𝐸𝑚𝑝𝐺 𝑠
!
∙𝑀 𝑘
!
+ 𝛾 ln (𝑝𝑜𝑝05)! + 𝛿 ln (𝑝𝑜𝑝𝑑𝑒𝑛05)! + 𝜀!"
!∈!,!,! !∈!"",!",!"#!
, 𝑀 𝑎𝑙𝑙
= 1, 𝑀 05 ! = 1 if Large retail store entered before 2005 ; 0 if otherwise,
𝐸𝑚𝑝𝐺 𝑠 ! is employment size group (1 − 4 , 5 − 9,
and 10+)
!
The results are reported in Table 5. In Columns (1) and (2), we explore the extent to
which productivity of entrants varies across size, locations and whether the store belongs to a
national chain or not. Overall larger stores have higher productivity. For smaller entrants, the
productivity level is higher for entrants in locations with a large discount store. Moreover such a
positive effect observed among small stores with less than five employees are concentrated
among chain stores located in the counties with large discount store. Overall our finding
suggests that entry of stores that provide retail service complementary to a large discount store
increased after a large discount store opened in town (e.g., small, specialized stores). The
finding of higher productivity for small, chain stores in counties with large discount stores
suggests that these stores get the benefit of locating with large discount store, possibly from
12
increased demand for modern shopping environment.
[Table 5 about here]
4.2 Robustness
The results using the expanded sample (entry of large discount store between 2006 and 2010) are
reported in Table 6.
The results are reported for the small GMS (excluding large GMS) and
non-GMS separately in Table 7.
The analysis using TFP is in progress.
[Tables 6 and 7 about here]
5. Conclusion
The ongoing structural change in retail trade—that is, the shift from single-store retailers toward
big-box national chains such as large discount stores and hypermarkets—is a worldwide
phenomenon. We provide insights on the role of such reallocations in productivity growth in the
retail industry. We find that opening of a large discount store may stimulate the reallocation
process in the local retail sector, by not only driving out less efficient stores but also attracting
new, modernized stores into the neighborhood. Productivity of small, chain stores are higher in
locations with a large discount store, possibly thanks to the spillover effects generated by
convenient, modern shopping amenities large discount stores provided. Our finding suggests that
the selection process driven by the large discount stores may work at the entry margin as well as
at the exit margin. Further analysis will help enlighten the selection process in entry and
productivity dynamics of the retail sector.
13
References
Basker, Emek, “Job Creation or Destruction? Labor Market Effects of Wal-Mart Expansion,” Review of
Economics and Statistics, 87(1), 2005, 174–183.
Bertrand, Marianne, Esther Duflo, and Sendhil Mullainathan, “How Much Should We Trust Differencein-Difference Estimates?” Quarterly Journal of Economics, 119(1), 2004, 249–275.
Cho, Janghee, Hyunbae Chun, and Yoonsoo Lee, “The Entry of Large Discount Store and the Evolution
of Employment in the Korean Retail Sector,” 2013 CAED Conference Paper, 2013.
Chun, Hyunbae, Seoji Heo, and Yoonsoo Lee, “Competition in Large Discount Retailers,” Sogang
University Working Paper, 2013.
Coe, Neil M., and Yong-Sook Lee, “We’ve Learnt How to Be Local: The Deepening Territorial
Embeddedness of Samsung-Tesco in South Korea,” Journal of Economic Geography, 13(2), 2013,
327–356.
Ellickson, Paul B., “The Evolution of the Supermarket Industry: From A&P to Wal-Mart,” Simon School
Working Paper Series No. FR 11–17, 2011.
Ellickson, Paul B., and Paul L.E. Grieco, “Wal-Mart and the Geography of Grocery Retailing,” Journal of
Urban Economics, 75, 2013, 1–14.
Foster, Lucia, John Haltiwanger, and C.J. Krizan, “Market Selection, Reallocation, and Restructuring in
the U.S. Retail Trade Sector in the 1990s,” Review of Economics and Statistics, 88(4), 2006, 748–758.
Haltiwanger John, Ron Jarmin, and C.J. Krizan, “Mom-and-Pop Meet Big-Box: Complements or
Substitutes?” Journal of Urban Economics, 67(1), 2010, 116–134.
Haskel, Jonathan, Ron S. Jarmin, Kazuyuki Motohashi, and Raffaella Sadun, “Retail Market Structure
and Dynamics: A Three Country Comparison of Japan, the U.K. and the U.S.,” Working Paper
prepared for National Bureau of Economic Research and the Conference on Research in Income and
Wealth, 2007.
Haskel, Jonathan, and Raffaella Sadun, “Entry, Exit and Labor Productivity in UK Retailing: Evidence
from Micro Data,” in Timothy Dunne, J. Bradford Jensen, and Mark J. Roberts (eds.) Producer
Dynamics: New Evidence from Micro Data, Chicago: University of Chicago Press, 2009, 271–302.
Haskel, Jonathan, and Raffaella Sadun. “Entry, Exit and Labor Productivity in UK Retailing: Evidence
from Micro Data,” Timothy Dunne, J. Bradford Jensen, and Mark J. Roberts (Eds.), Producer
Dynamics: New Evidence from Micro Data, University of Chicago Press, 2009, 271-302.
Haskel, Jonathan, and Raffaella Sadun, “Regulation and UK Retailing Productivity: Evidence from
Microdata,” Economica, 79(315), 2012, 425-448.
Hausman, Jerry, and Ephraim Leibtag, “Consumer Benefits from Increased Competition in Shopping
Outlets: Measuring the Effect of Wal-Mart,” Journal of Applied Econometrics, 22(7), 2007, 1157–
14
1177.
Holmes, Thomas J., “The Diffusion of Wal-Mart and Economies of Density,” Econometrica, 79(1), 2011,
253–302.
Iacovaone, Leonardo, Beata Smarzynska Javorcik, Wolfgang Keller, and James R. Tybout, “Supplier
Responses to Wal-Mart’s Invasion of Mexico,” NBER Working Paper No.17204, 2011.
Igami, Mitsuru, “Does Big Drive Out Small? Entry, Exit, and Differentiation in the Supermarket Industry,”
Review of Industrial Organization, 38(1), 2011, 1–21.
Jarmin, Ronald S., Shawn D. Klimek, and Javier Miranda, “The Role of Retail Chains: National,
Regional and Industry Results,” in Timothy Dunne, J. Bradford Jensen, and Mark J. Roberts (eds.)
Producer Dynamics: New Evidence from Micro Data, Chicago: University of Chicago Press, 2009,
237–262.
Javorcik, Beata S., and Yue Li, “Do the Biggest Aaisles Serve a Brighter Future? Global Retail Chains
and Their Implications for Romania,” Journal of International Economics, 90(2), 2013, 348–363.
Jia, Panle, “What Happen When Wal-Mart Comes to Town: An Empirical Analysis the Discount
Retailing Industry,” Econometrica, 76(6), 2008, 1263–1316.
Masta, David A., “Competition and Product Quality in the Supermarket Industry,” Quarterly Journal of
Economics, 126(3), 2011, 1539–1591.
Neumark, David, Junfu Zhang, and Stephen Ciccarella, “The Effects of Wal-Mart on Local Labor
Markets,” Journal of Urban Economics, 63(2), 2008, 405–430.
Rivero, Rosario, and Rodrigo Vergara, “Do Large Retailers Affect Employment? Evidence from an
Emerging Economy,” Applied Economics Letters, 15(1), 2008, 65–68.
Sadun, Raffaella, “Does Planning Regulation Protect Independent Retailers?” NBER Working Paper No.
19797, 2014.
Schivaridi, Fabiano, and Eliana Viviano, “Entry Barriers in Retail Trade,” Economic Journal, 121(551),
2010, 145–170.
Sobel, Russell S., and Andrea M. Dean, “Has Wal-Mart Buried Mom and Pop? The Impact of Wal-Mart
on Self-Employment and Small Establishments in the United States,” Economic Inquiry, 46(4), 2008,
676–695.
de Vries, Gaaitzen J., “Did Liberalization Start A Retail Revolution In Brazil?” No. GD-105. Groningen
Growth and Development Centre, University of Groningen, 2008.
Zhu, Ting, Vishal Singh, and Anthony Dukes, “Local Competition and Impact of Entry by a Discount
Retailer,” Working Paper, Carnegie Mellon University, 2006.
15
Figure 1. Productivity Distribution of Entrants in Countries with and without Large Discount
Stores
Notes: Figures show the distribution of log labor productivity of entrants in Other Retail (exlcuding LDS). LDS
Area is a county where a large discount store entered before 2005. Non-LDS Area is a county where no large
discount entered as of 2010.
16
Table 1. Transition Matrix of Relative Productivity in 2005 and 2010
Quintile1(2005)
Quintile2(2005)
Quintile3(2005)
Quintile4(2005)
Quintile5(2005)
Entrants
Quintile1
(2010)
Quintile2
(2010)
Quintile3
(2010)
Quintile4
(2010)
Quintile5
(2010)
Exits
25.4
12.2
6.1
3.2
1.2
51.8
26.7
12.6
6.5
3.3
1.2
14.8
13.8
9.5
6.4
2.9
14.5
13.4
9.5
6.3
2.9
9.3
12.1
10.9
9.4
5.2
8.9
11.4
10.6
9.1
5.0
5.4
8.3
10.5
13.5
9.9
4.4
8.2
10.6
13.5
9.9
2.1
4.1
6.5
12.2
24.2
2.0
4.0
6.4
12.1
23.8
9.6
42.6
50.4
56.4
55.8
57.4
52.5
11.6
10.1
8.7
8.9
8.6
17
Row
Total
10.1
52.6
9.3
53.1
9.0
52.5
9.5
51.0
52.2
100.0
Table 2. Decomposition of Labor Productivity Growth, 2005-2010
[1]
[2]
[3]
Total Retail
Large GMS
Other Retail
weight(employment)
1.000
0.066
0.934
Within
4.33
6.03
4.22
Between
1.41
12.14
0.66
Cross
-4.44
-29.86
-2.65
Net entry
7.92
-0.97
8.55
Entry
6.67
-12.11
7.99
Exit
1.25
11.14
0.55
9.23
-12.67
10.77
∆ Labor Productivity
18
Table 3. Summary Statistics
Establishment Type
Continuing ( 479,390)
Entrants (268,872)
Exiters (266,135)
Total (1,014,397)
Labor productivity
Employment
Sales
29.13
1.94
174.23
(53.70)
(2.05)
(674.87)
41.64
2.12
229.06
(172.19)
(2.66)
(851.34)
28.75
1.92
141.84
(46.05)
(2.69)
(532.08)
32.34
1.98
180.26
(99.04)
(2.40)
(694.73)
Notes: The table provides summary statistics of the regression sample. For establishment types, numbers in
parentheses are the number of observations. The table reports averages of labor productivity, employment, and sales.
Numbers in parentheses are standard deviations.
19
Table 4. Productivity Growth of Continuing Establishments, Entrants, and Exiters in 2005-2010
A. Baseline Model
Year 2010
Entrants
Exiters
Sample size
R-squared
Controls
(1)
(2)
-0.227***
(0.010)
0.218***
(0.013)
0.002
(0.012)
1,014,397
0.083
-0.227***
(0.010)
0.190***
(0.012)
-0.022**
(0.010)
1,014,397
0.102
4-digit industry and
provinces dummies
4-digit industry dummies
B. Model interacted with counties entered by large discount stores
(1)
Year 2010
Continuing 2005
M(05)
Continuing 2010
M(05)
Entrants
All area
M(05)
Exiters
All area
M(05)
Sample size
R-squared
Control
(2)
-0.294***
(0.021)
-0.294***
(0.021)
0.294***
(0.040)
0.245***
(0.034)
0.089***
(0.024)
0.089***
(0.024)
0.305***
(0.018)
0.241***
(0.040)
0.277***
(0.020)
0.203***
(0.032)
-0.011
(0.019)
0.290***
(0.054)
1,014,397
0.083
-0.034**
(0.017)
0.246***
(0.044)
1,014,397
0.102
4-digit industry and
provinces dummies
4-digit industry dummies
Notes: Dependent variable is the log productivity of establishments. M(05) indicates a county where a large discount
store enters before 2005. Numbers in parentheses are county-clustered standard errors. * Significant at the 10% level;
** Significant at the 5% level; *** Significant at the 1% level
20
Table 5. Productivity Growth and Entrants: Size and Chains
Small2
Medium
Small1 * M(05)
Small2 * M(05)
Medium * M(05)
Small1 * M(05) * Chain
Small2 * M(05) * Chain
Medium * M(05) * Chain
Ln (Population)
Ln (Population density)
Sample size
R-squared
Labor productivity
of entrants
(1)
(2)
1.111***
1.154***
(0.057)
(0.056)
1.331***
1.377***
(0.053)
(0.055)
0.089***
0.001
(0.032)
(0.033)
-0.124***
-0.237***
(0.062)
(0.059)
-0.028
-0.355***
(0.060)
(0.066)
0.928***
(0.022)
0.259***
(0.034)
0.658***
(0.046)
0.118***
0.104***
(0.027)
(0.027)
0.016
0.018
(0.014)
(0.013)
268,872
268,872
0.114
0.167
Entrants dummy
in 2010
(3)
(4)
1.161***
1.165***
(0.009)
(0.009)
0.209***
0.214***
(0.020)
(0.020)
0.058***
0.047***
(0.014)
(0.014)
-0.010
-0.015
(0.016)
(0.016)
0.006
0.044
(0.026)
(0.027)
0.148***
(0.006)
0.014
(0.011)
-0.068***
(0.015)
0.047***
0.045**
(0.012)
(0.012)
-0.006
-0.006
(0.005)
(0.004)
508,567
508,567
0.082
0.088
Exits dummy
in 2010
(5)
(6)
0.053***
0.052***
(0.012)
(0.012)
0.040
0.039
(0.029)
(0.029)
0.045**
0.047**
(0.019)
(0.019)
0.034*
0.084***
(0.019)
(0.020)
0.044
0.127***
(0.032)
(0.032)
-0.022***
(0.008)
-0.120***
(0.011)
-0.165***
(0.018)
0.011
0.011
(0.016)
(0.016)
0.007*
0.007*
(0.004)
(0.004)
505,830
505,830
0.037
0.037
Notes: Dependent variables are labor productivity of entrants in (1) and (2), entrant dummy in (3) and (4), and exiter
dummy in (5) and (6), respectively. All regressions include 4-digit industry and provinces dummies. Stores are
classified into three groups of Small1 (fewer than 5 workers), Small2 (5 to 9 workers), and Medium (10 or more).
The average size between 2005 and 2010 is used for continuing stores. M(05) indicates a county where a large
discount store enters before 2005. Numbers in parentheses are county-clustered standard errors. * Significant at the
10% level; ** Significant at the 5% level; *** Significant at the 1% level
21
Table 6. Productivity Growth and Entrants: Expanded Sample
A. Baseline Model
Year 2010
Entrants
Exiters
Observations
R-squared
Control
(1)
-0.226***
(0.009)
0.220***
(0.012)
-0.003
(0.011)
1,137,912
0.082
4-digit industry dummy
B. Model interacted with counties entered by large discount stores
(1)
Year 2010
-0.294***
(0.021)
Continuing 2005
M(05)
0.294***
(0.040)
M(0610)
0.177***
(0.049)
Continuing 2010
M(05)
0.089***
(0.024)
M(0610)
0.075**
(0.029)
Entrants
All area
0.305***
(0.018)
M(05)
0.241***
(0.040)
M(0610)
0.187***
(0.046)
Exiters
All area
-0.011
(0.019)
M(05)
0.289***
(0.054)
M(0610)
0.142***
(0.061)
Observations
1,137,912
R-squared
0.095
Control
4-digit industry dummy
(2)
-0.226***
(0.009)
0.194***
(0.011)
-0.026***
(0.009)
1,137,912
0.101
4-digit industry
and province dummies
(2)
-0.294***
(0.021)
0.244***
(0.033)
0.138***
(0.038)
0.089***
(0.024)
0.075**
(0.029)
0.276***
(0.020)
0.203***
(0.031)
0.159***
(0.037)
-0.034**
(0.017)
0.246***
(0.043)
0.110**
(0.048)
1,137,912
0.110
4-digit industry
and province dummies
Note: M(05) indicates a county where a large discount store enters before 2005. M(0610) indicates a
county where a large discount store enters between 2006 and 2010. Numbers in parentheses are countyclustered standard errors. * Significant at the 10% level; ** Significant at the 5% level; *** Significant at
the 1% level.
22
Table 7. Productivity Growth and Entrants: GMS versus Non-GMS Industries
A. Baseline Model
Year2010
Entrants
Exiters
Sample size
R-squared
Control
(1) GMS
(2) non-GMS
-0.192***
(0.012)
0.351***
(0.016)
-0.142***
(0.008)
195,002
0.065
4-digit industry
and province dummies
-0.238***
(0.010)
0.164***
(0.013)
0.001
(0.010)
819,395
0.119
4-digit industry
and province dummies
B. Model interacted with counties entered by large discount stores
(1)
Year2010
Continuing 2005
M(05)
Continuing 2010
M(05)
Entrants
All area
M(05)
Exiters
All area
M(05)
Sample size
R-squared
Control
(2)
-0.275***
(0.021)
-0.300***
(0.025)
0.318***
(0.039)
0.211***
(0.034)
0.113***
(0.025)
0.082***
(0.027)
0.425***
(0.023)
0.306***
(0.048)
0.234***
(0.024)
0.186***
(0.033)
-0.193***
(0.019)
0.374***
(0.045)
195,002
0.084
4-digit industry
and province dummies
0.001
(0.016)
0.199***
(0.043)
819,395
0.126
4-digit industry
and province dummies
Notes: GMS includes supermarkets and convenience stores. Non-GMS includes specialized stores such as bakery,
electronics, clothing stores, etc. M(05) indicates a county where a large discount store enters before 2005. Numbers
in parentheses for the estimated coefficients are county-clustered standard errors. * Significant at the 10% level; **
Significant at the 5% level; *** Significant at the 1% level.
23