What Can We Learn by Analyzing
Business Location Using
Establishment and Firm Data?
Ron Jarmin
U.S. Census Bureau
Prepared for the Symposium
“The Role of Intangible Assets in the Improvement
of Firm Performance”
CAED 2009, Tokyo
The problem with intangibles is that…
…..they’re intangible!!
2
Making intangibles tangible
• How can the data, tools and methods of
the CAED help on this front?
– Direct measurement of investments in
intangible assets
• R&D, patents, advertising…
– Indirect inference
• Productivity/Market value
3
Commonly available data asset:
Location as an intangible
• Business decisions concerning location affect
firm performance
– Competitive factors (e.g., retail)
– Innovation/Productivity factors (e.g., clusters)
• Location typically not a factor in models of
firm performance
– Not of first-order importance
• Included only to soak up regional variation
– Data not available
• Network of locations a means of capitalizing
on other intangible assets (e.g., Wal-Mart
organizational capital/business model)
4
How can we use “location” to better
understand firm performance?
• Analyzing firm performance across
regions
– Clusters, regional characteristics
• Analyzing firm performance within
regions
– The focus of this talk
5
Illustrative Example: Impact of
Big-Box Store Entry
• Does entry of a Big-Box retailer impact
mom-and-pop and smaller chain stores
in local retail markets?
– Haltiwanger, Jarmin, Krizan (NBER
Working Paper 15348, forthcoming
Journal of Urban Economics)
6
Detailed location and retail firm
performance
• Retail mantra “Location, Location, Location”
• That is a retail firm’s pattern of location(s) is
a critical, first-order part of its overall
business strategy.
• How do we best incorporate this into models
of retail firm performance?
– What aspects of location do we need to capture?
– How do we parsimoniously capture these
aspects?
7
Incorporating Location
• Proximity to customers
– Number
– Characteristics
• Proximity to other retailers
– Substitutes
– Complements
• Proximity to infrastructure
– Highways
– Public Transit
8
Measuring distance between retail
firms within local markets
• Data requirements
– Establishment (store) level data
• Detailed location (either small geography or,
better, latitude/longitude coordinates)
• Retail sub-sector (hardware, clothing…)
• Firm ownership and characteristics
– Mom-and-pop (single location)
– Large chain (15+ states) / Small Chain (<15 states)
– Big-Box
9
Background
• Primary measure of performance:
– Employment growth (measured at the
establishment level
• git = (Xit- Xit-1)/ ((Xit+Xit-1)/2)
• Sample:
–
–
–
–
Washington, DC metro area retail
1976-2005
from the Longitudinal Business Database
focus on mom-and-pop and small chain stores
10
Shares of Retail Employment in
D.C. Metro Counties by store type
Shares of Employment by Establishment Type
Single Unit
Small Chain
Large Chain
0.5
Big Box
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
1975
1980
1985
1990
1995
2000
2005
11
Net Employment Growth by Type
Ne t Em ploym e nt Grow th Ra te s by Esta blishm e nt Type
(3-ye a r M A)
0.2
0.15
0.1
Single Unit
Small Chain
0.05
0
1975
Large Chain
Big Box
1980
1985
1990
1995
2000
2005
-0.05
-0.1
12
Descriptive Findings
• Large increase in the presence of bigbox stores in the DC area during the
1990’s
• Small chains appear to be more
adversely affected than mom-and-pop
stores at the aggregate metro area
level.
13
Results from formal analysis
• Single unit (mom-and-pop) and small chain
store regressions control for:
– Local population characteristics (income, age,
gender)
– Proximity to Interstate exits
– Proximity to Metro stations
– Establishment (store) age
– Year effects
– Retail sector based on detailed industry codes
14
Base results for Mom-and-Pop
Stores
0.1
Impact on Mom and Pop Employment
Growth from Prior Period Initial Big-Box
Entry
0
Same Sector Single
Units
-0.1
Other Sector Single
Units
-0.2
-0.3
Within 1 mile
Betw een 1 and 5 Betw een 5 and 10
miles
miles
Distance to Big-Box Entrant
15
Impact felt largely through store
closings
Impact on Mom and Pop Job Destrustion
from Establishment Exit from Prior Period
Initial Big-Box Entry
0.35
0.3
0.25
0.2
Same Sector
Single Units
0.15
Other Sector
Single Units
0.1
0.05
0
-0.05
Within 1 mile
Betw een 1
and 5 miles
Betw een 5
and 10 miles
Distance to Big-Box Entrant
16
Impact differs by type of area
Impact of Same Sector Big Big Box Growth on SU Employment
Growth Across Different Areas
0.1
0.05
0
Impact
High Pop/Low Inc
Low Pop/High Inc
High Pop/High Inc
Low Pop/Low Inc
-0.05
-0.1
-0.15
-0.2
1 mile
1 to 5 miles
5 to 10 miles
17
Summary of results
• Impact of Big-Box entry and growth on momand-pop and small chain stores limited to
those in the same retail industry and in close
proximity to the Big-Box.
• Store exit is the primary margin of adjustment.
• Impacts vary across the DC area according
to income and population density.
18
Implications for future work
• At least for some sectors (e.g., retail,
services), detailed location is critical to
understanding firm performance.
– In the DC retail example, to understand
why some mom-and-pops perform better
than other requires knowing their relative
proximity to other types of stores.
19
Implications (continued)
• Detailed location (geography) is often
available on establishment level data
• Yet it is under-utilized.
• Geocoded establishment data can easily be
linked to other geocoded datasets to add
addition covariates to analyses of firm
performance.
• By adding detailed location, CAED data,
tools and methods can be applied to broad
range of literatures (e.g., economic
development, urban economics,
entrepreneurship).
20