1. Key Issues Elements of risk Quantitative methods of location analysis
Buying Power Index
Index of Retail Saturation
Market expansion potential
Gravity models: Reilly, Huff
Regression as used in trade area analysis

2. Selecting a Retail Site
The Expansion Stage
Use real estate firms, local personnel, etc.
to identify a large number of possible sites
The Qualitative Stage
Use checklists or examine performance of analogous
stores to screen possible sites for the best sites
The Quantitative Stage
Use quantitative modeling to further screen the likely
sites by generating forecasted potentials for each site
The Decision Stage
Make a site selection decision based on the results of
both the quantitative qualitative assessments

3. Location Analysis
Regional
Analysis
Area
Analysis
Site Analysis

4. Information Minimizes Risk
Risk Criteria

5. Factors Affecting Interest in a Region or Trade Area
Source: Levy & Weitz

6. Methods: The “Quantitative” Stage
Measuring Demand
Decennial Census of the U.S.
Buying Power Index (BPI)
Private Firms
Issues Affecting Competition
Index of Retail Saturation
Market Expansion Potential
Measuring Trade Areas
Analog (similar store) Approach
Customer Spotting (trade-area mapping)
Gravity Models
Reilly’s Law of Retail Gravitation
Converse’s Breaking Point Model
Huff’s Model of Intermarket Attraction
Multiple Regression Analysis

7. Decennial Census Data Complete Count Data Items
Population Items Housing Items
Relationship to household head Number of units at this address
Color or race Telephone
Age Private entrance to living quarters
Sex Complete kitchen facilities
Marital status Rooms
Water supply
Flush toilet
Bathtub or shower
Basement
Tenure (owner/renter)
Property Value
Contract rent
Vacancy status
Months vacant
Typical Sampled Data Items Population Items
State or country of birth
Years of schooling
Number of children
Employment status
Hours worked last week, year
Last year in which worked
Occupation
Income, by type
Country of birth & of parents
Mother tongue
Year moved into this house
Means of transportation .
Source: The Census and You, U.S. Department of Commerce, Bureau of the Census.

8. Measuring Demand BPI: Buying Power Index Decennial Census of the U.S.
Retail demand in an area as a % of total retail demand in the U.S.
BPI = (.5 x Income) + (.3 x Sales) + (.2 x Population)
BPI
Salt Lake City .471
San Diego .935
New York City
Decennial Census of the U.S.
SMSA: Standard Metropolitan Statistical Area
“a central city of 50,000+ population, the counties in which it is located, and other contiguous metropolitan counties that are economically and socially integrated with the central city”
Private Firms
Urban Decision Systems
Map Objects (W3)
Microvision Zip Code System
Claritas Corporation’s PRIZM database

9. Issues Affecting Competition: Index of Retail Saturation
Shows the relationship between demand and supply for a store type -- the average retail spending per square foot of selling space
IRS = population of area x per capita retail spending
retail selling space (in sq. feet)
Or just …
IRS = area retail spending
retail selling space

10. Issues Affecting Competition: Index of Retail Saturation
IRS = area retail spending
retail selling space
???
E.g., in Utah Valley (2000):
Population = 358,000
Annual per capita retail spending = $10,377
IRS = 358,000 x $10,377 = $ X

11. Utah Valley Retail Selling Space
Provo CBD 1,400,000
University Mall 1,260,000
Provo Towne Centre 950,000
Shops at Riverwoods 190,000
Carillon Square 525,000
American Fork CBD 320,000
Parkway Village 443,000
Pason CBD 360,000
Provo Big Lots Strip Mall 100,000
Orem Center Street 260,000
Springville CBD 175,000
Pleasant Grove CBD 120,000
Lehi CBD 100,000
Brigham’s Landing 120,000
East Bay 300,000
Riverside Shopping Center 250,000
Am Fork Shopping Center 115,000
Northgate Shopping Center 110,000
Park Place 220,000
Macey’s Orem 100,000
Timp Plaza 100,000
University Festival Mall 125,000
Expressway Square 85,000
Edgemont Center 40,000
North Park Shopping Center 125,000
TOTAL 8,013,000
IRS = 358,000 x $10,377
8,013,000
= $464
Interpretation?
Source: Colliers & Clarke, Provo, 2002;
Utah Valley Economic Development,
Provo/Orem Chamber of Commerce
This number is too low

12. Issues Affecting Competition: Market Expansion Potential
MEP - indicates an area’s potential for creating new demand
MEP = Expected Sales
Actual Sales

13. Measuring Trade Areas: Analog Approach
Information about customers, competition, and sales of current similar stores are used to predict sales of a new store or center.
Steps:
1. Determine trade area for successful stores (customer spotting)
2. Characterize their primary, secondary, fringe trade areas
3. Match the characteristics of these stores with potential new store locations to determine the best site
Store Average White- % of Predominant Level of
Location Household Collar Residents PRIZM Profile Competition
Income Occup. Age 45+
Optics City $ 99,999 High 38% Blue Blood Est. Low
Site A ,000 High Young Suburbia Medium
Site B ,000 Low Gray Power Low
Site C ,000 High Young Literati Low
Site D ,000 High Money & Brains Medium
Successful Store

14. Measuring Trade Areas Gravity Models

15. Gravity Models: Reilly’s Law of Retail Gravitation
Breaking Point
Dist(A)
Dist(B) A B
At what point between A & B will shoppers go to A? B? A
Provo
pop’n = 358K B
Salt Lake
pop’n = 1275K
45 miles

16. How could we use a Gravity modeling approach here?
Gravity Models A B C
Site Possibility D
Competitors
How could we use a Gravity modeling approach here?

17. Gravity Models: Huff’s Intramarket Area Model
Prob(shoppingj) = f(selling spacej, travel timeij)
Based on the premise that the probability that a given customer will shop in a particular store or shopping center becomes larger as the size of store or center grows and distance or travel time from customer shrinks
Probability that
consumer living
in area i will
shop at store j
Sq feet of selling
space in j
Travel time
from i to j
Sq ft of selling space
in other stores
in trade area
Travel time from i
to other stores
Relative Attractive-
ness of store j
Total Attractiveness
of other stores

18. Gravity Models: Huff’s Intramarket Area Model
Probability that
consumer living
in area i will
shop at store j
Sq feet of selling
space in j
Travel time
from i to j
Sq ft of selling space
in other stores
in trade area
Travel time from i
to other stores
Relative Attractive-
ness of store j
Total Attractiveness
of other stores S1
(Ti1)  Sj
(Tij)  n 
j=1
P(Cij) =
where:
P(Cij) = probability that consumer C living in area i will visit shopping center j
Sj = square footage of selling space in j devoted to a particular class of merchandise
Tij = travel time from area i to shopping center j
 = an estimated parameter to reflect the effect of travel time on various kinds of shopping trips (larger reflects greater weight for travel time)

19. Gravity Models: Huff’s Intramarket Area Model
University Mall
2 mi
Riverwoods Shopping Center
Provo Town Centre
3 mi i
4 mi
where:
S1 = 10,000 square feet
S2 = 15,000 square feet
S3 = 20,000 square feet
Ti1 = 2 miles
Ti2 = 3 miles
Ti3 = 4 miles
 = 2
thus:
10,000
P(Ci1) = = .46
10, , ,000
Interpretation?

20. Regression Analysis Approaches
Can use data from current successful stores to predict the probability of success for stores were they placed on sites being considered.
E.g.,
predicted sales on site A = f(visibility of store from street, no. of competitors in trade area, population density in trade area, average age in trade area, average income in trade area, number cars per hour past site, etc.)

21. Regression Fundamentals
R2 = variance in Y explained by X
Equation for a line?
Y = a + bX … a? b?
a = intercept … point where line crosses Y axis
b = slope … Y / X
Interpret:
Y = X
Y = 1 + 1X
Y-axis e.g., sales 5 4 3 2 1
X-axis e.g., population

22. Regression Example from Branch-Bank Performance Model
Regression Cumulative
Variable Coefficient t-value R2
CD: Checking deposits ($1,000):
HI: Median household income ($l00)
SF: Retail square footage (1,000)
C: # of competing banks' branches
SD: Savings deposits ($1,000):
PP: Purchasing power ($1,000)
EL: Employment level
RH: Percentage of renter housing

23. Annual Sales for 10 Home Improvement Centers
Regression Example
Annual Sales for 10 Home Improvement Centers
Store Yearly Sales to 3-Mile Radius
($000) Population
$ ,000
,500
,000
,400
,600
,100
,000
,200
,400
,200
Source: John S. Thompson, Site Selection (New York: Lebar-Friedman), p. 133.

24. 10 Home Improvement Centers
800 --
700 --
600 --
500 --
400 --
366 --
300 --
200 --
100 -- 10 7
Sales ($000)
Change in sales
= slope = 0.007
Change in population 3 6 2 8 1
Change in sales 4 9
Change in population 5
90.97
Population (000)
What is the regression equation?

25. Critiquing a Multiple Regression Equation
Is the equation complete on all important variables?
Do the signs of the independent variables make sense?
Is there logical reason to expect that the performance of a store is related to the independent variables?
Are the regression forecasts kept within the range of the input data?

26. Interpretation?
Suppose Media Play has developed a regression equation to identify attractive retail trade areas and sites, based on a study of 100 of its stores having from $450,000 to $1 million in sales.
Forecasted Annual Sales (in $000) = HHI + 1.2TC CR
where, HHI = mean trade-area household income (in $000)
TC = average no. of cars per minute driving past the site
CR = # competing retailers within a mile of site
For an increase of Sales forecast will change by
$1000 in mean household income _________
50 cars per minute _________
1 retailer in the trade area _________

27. Problems with this Equation?
Forecasted Annual Sales (in $000) = HHI + 1.2TC CR
where, HHI = mean trade-area household income (in $000)
TC = average no. of cars per minute driving past the site
CR = # competing retailers within a mile of site
Is the equation complete on all important variables?
Do the signs of the independent variables make sense?
Is there logical reason to expect that the performance of a store is related to the independent variables?
Are the regression forecasts kept within the range of the input data?

28. Multiple Regression Procedure
Select a group of existing stores
Identify a dependent variable (e.g., revenues)
Select a set of independent variables logically related to the dependent variable
Collect data on all variables for both existing stores (including revenues) and the proposed sites (no revenues of course)
Enter the data & develop the a regression equation
Use equation to forecast sales or share for the proposed sites
Focus further attention on sites having highest forecasts

29. Regression Example: Hogi Yogi
No. competitors: No. of frozen yogurt competitors within 1-mile radius
Visibility: A subjective 1-7 index rating based on how visible the store is within the shopping center or on the street
Clustering: The degree to which the store is clustered with other restaurants (0 to 3)
RGI: Secondary data on Restaurant Growth Index (which shows the relationship between restaurant supply & demand, by market, with average = 100
University: Whether there is a university within 1 mile (0=no, 1=yes)
Local owner: Whether the owner lives near the restaurant (0=no, 1=yes)
Method:
20 Hogi Yogi stores were initially analyzed by means of 70 variables per store, along with sales data.
Data was examined for (1) the first 4 months of a store’s opening and (2) the first year of sales.
Using stepwise regression, all 70 were examined, to develop two equations:
1. Initial 4-month sales
upon opening a new
store.
2. Sustained sales
during the first year
of operation
Source: BYU Site selection study for Hogi Yogi, 1995

30. Regression Example: Hogi Yogi
$ Sales during the
first four months = -$
No. competitors  (-$31330) +
Visibility  ($13297) +
Clustering  ($27566) +
3-mile/capita income  ($6.69)+
Size in sq ft  ($56.7) +
RGI  ($1382) +
University  ($27521) +
Local owner  ($21422)
Actual Sales
Predicted Sales
Source: BYU Site selection study for Hogi Yogi, 1995