Store Location in Shopping Centers: 15 Years After Dr. Charles C. Carter and Dimitri M. Teddone, MAI Frank Lloyd Wright: The Elevator & the Automobile at the Turn of the Century With the advent of the elevator, downtown shopping districts with its department stores dominated the retail shopping experience near the end of the nineteenth century. Electric streetcar lines then connected the downtown retail district with residential areas like the spokes of a wheel. Chicago: Marshall Field’s; Detroit: Hudson’s; New York: Macy’s; Minneapolis: Dayton’s; Portland: Meier & Frank Frank Lloyd Wright (1867-1959) knew what to expect when he said: “The city of the future will depend on the race between the car and the elevator, and anyone who bets of the elevator is crazy.” Rise of the Shopping Center Automobile use expanded during the 1920’s and 1930’s & residential uses expanded rapidly into the suburbs. Levitt-towns first started on Long Island, NY after World War II & these became the model of residential uses. Suburban subdivisions consisting of single-story, detached houses on single lots became the norm spreading out from metropolitan areas throughout the U.S. Two- or three-story row houses along boulevards near the downtown were no longer the custom. Shopping centers came to revolutionize retail shopping during the latter half of the 20th century. Fall of the Downtown Retail District Downtown retail districts of the 1950’s were eventually drained of retail uses as suburban shopping centers took over the retail trade. Early suburban centers were built by department stores, i.e., Hudson Co. built Northland Center in Detroit (1954) & Dayton Co. built Southdale Center in Minneapolis (1956). Office uses came to dominate the downtown after the 1960’s Suburban residences met the demand for more living space after the time to cover distances (using cars) shortened. ULI’s Shopping Center Development Handbook, 2nd ed. graphs the increase of US shopping centers from 1950 to 1985. Growth of Shopping Centers From 1950 to 1955 the graph shows a slow increase in the number of centers; from 1955 to 1980 the rate grows at 6 times as much. By 1985 about 50% of all retail sales were made at shopping centers, not including automobile sales & gasoline. The rate of retail sales in shopping centers since 1985 has declined: 1974: 25%, 1985: 50%, 1996: 52% In 1998 87% of all US shopping centers were less than 200,000 SF. GLA (smaller than regional or super-regional centers). In the late 1970’s & early 1980’s regional shopping returned Shopping Centers Slow to the downtown, e.g., Faneuil Hall, Boston, Harborplace, Baltimore, & South Street Seaport, New York By the1990’s shopping center growth had slowed considerably, evidence that the market had become saturated Big box or “category killer” retailers such as Lowe’s, Home Depot, Costco, Best Buy, & K-Mart became popular during the 1980’s; online retailing has expanded since about 2000 By 2005 construction of new, enclosed shopping centers had just about stopped. Today shopping centers are very much with us, and, unlike the downtown retail district, are not likely to disappear soon. What Did Shopping Centers have that the Central Business District Didn’t? Studies of microeconomics of shopping centers started in earnest about 1989 and continues today. Research eventually showed that the structure of rent, space, leases, and store location under a unified plan & management results in positive demand externalities that a CBD cannot reproduce. What are positive consumer demand externalities? They get customers to spend their money by providing a place & circumstances where consumers want to shop. Economists call this “internalizing” positive externalities; the ULI calls this process synergistic & we can leave it at that. Original Data: 1992-93 Shopping Centers, 8 Regional & Super-Regional Original data was gathered from 9 regional (300K-900K GLA) and super-regional (500K-1.5M GLA) shopping centers scattered throughout the U.S. (avg. 828.6K GLA) 6 were single-level, 2 double-level, & 1 three-level (3-level mall not used) anchor stores ranged from 2 to 6 per center (avg. 3.56) Parking ratios ranged from 5.35 to 5.62 stalls per 1,000 sq ft of shopping area (standard = 5.5 stalls per 1,000 sq ft) Years completed ranged from 1952 to 1990, the center built in 1952 had been completely renovated in 1987 (avg. 16 yrs.) Vacancies ranged from 4.4% to 0.4%, avg. 3.1% Original Data: 1992-93 Shopping Centers, 8 Regional & Super-Regional Numbers of non-anchors (including vacancies) ranged from 117 to 158 (avg. 129) Of the 1,010 leases 689 were deemed usable (69%) Leases were unusable: 1) lease either starts or ends in the year sales were reported (insufficient info. on renewed/rolled over tenants) 2) some tenants did not report sales (i.e., service tenants w/out overage rent) 3) cinemas & out-pad tenants (the former often operate on a different schedule & the latter are outside the mall) Findings in First Study 1) Highest customer traffic takes place at the mall’s center (usually at or near the food court), & customer traffic tapers off with distance from the center. 2) Store size increases, & rent per SF decreases, w/distance from the mall’s center (rents per SF ↑ @ +4.0% /100 ft) 3) Stores of the same type that promote comparison shopping (e.g., women’s apparel & men’s apparel) will generally be dispersed as opposed to clustered 4) Rents decline at different rates with distance from the mall center, smaller store types paying higher rents/SF at the center whose rents decline faster than larger store types. Conclusion: There is an Optimal Tenant Mix & Optimal Tenant Locations Stores locate according to what is termed a bid-rent process in microeconomics , so that store types position themselves to receive maximum sales (and pay maximum rents) Store location & size are two of the mall characteristics that, together, go to maximize rents (lease structure does too) Assuming leases are properly structured, the retail market is just-saturated, and zero vacancies, store sizes and locations within the mall are the remaining variables to obtain optimal tenant mix. These variables are best understood as the result of trial & error, Adam Smith’s “invisible hand.” Follow-Up Data: 2006 Shopping Centers, 8 Regional & Super-Regional Follow-up data was gathered from 8 regional (300K-900K GLA) & super-regional (500K-1.5M GLA) shopping centers located along the U.S. Atlantic Coast (avg. 481.5K GLA) 6 were single level & 2 were double level centers anchor stores ranged from 2 to 4 per center (avg. 3.25) Years completed ranged from 1961 to 1992, with 2 centers completely renovated (ones built in 1961 & 1985 in 1993 & 1995, respectively) (avg. 25.6 yrs.) Leases were unusable: 1) 3 shopping centers had too many vacancies (46%, 30%, & 20%) for results to be creditable (previous research shows this) Follow-up Data: 2006 Shopping Centers, 8 Regional & Super-Regional 2) For 3 centers site plans were not readily available (from which to measure distances from stores) For the remaining 2, vacancies were 8.6% and 10%, parking ratios were 5.5 & 6 spaces per 1000 GLA, & the number of usable non-anchor stores were 50 (w/33K GLA vacant) and 41 (w/47K GLA vacant) Of the 91 leases 54 were deemed usable (59%) Comparisons: 481.5K avg. GLA (2006) vs. 828.6K avg. GLA (1992); 3.25 anchors avg. (2006) vs. 3.56 anchors avg. (1992); 25.6 avg. age (2006) vs. 16 avg. age (1992); 17.2% avg. vacancies (2006) vs. 3.1% avg. vacancies (1992); number of usable leases 54 (2006) vs. 689 (1992) How to Get More Observations: Bootstrapping A bootstrapping technique was used to provide more observations (lease data to use in statistical analysis). By this nonparametric method samples from the observations at hand are made as if these samples were from a population. In bootstrapping, sampling with replacement takes place using the data on hand (the 54 store leases) to create a larger number of observations from which to run statistical analysis. From this method 1,200 non-anchor observations were produced, i.e., 60 random samples of 20 leases from the 54 store leases = 1,200 lease observations Store Characteristics for 2006 Follow-Up Data Variables Mean Std. Dev. Min. Max. Square Feet (SF) 3084 2761 590 15,709 Rent per SF ($/SF) $43.64/SF $24.92/SF $4.09/SF $118.30/SF Feet from Mall Center 267.17 130.78 30 540 Variable Store Type Observatio n Frequency Mean Std. Dev. Min. Max. SF Women’s Apparel 11.6% 5,215 1,665.82 3,015 7,773 SF Jewelry 7.5% 1464.5 205.40 1,205 1,673 SF Food Court 13.3% 784 254.64 590 1,226 TRNT /SF Women’s Apparel $18.17 $13.91 $8.44 41.36 TRNT / SF Jewelry $57.29 $17.13 $40.46 $76.13 TRNT / SF Food Court $82.07 $21.55 $60.10 $118.30 Findings in Follow-Up Study 1) Store size increases, & rent per SF decreases, w/distance from the mall’s center (rents @ +5.3% /100 ft, square footage +8.26% / 100 ft) 2) Generally, store types still line up with distance from the mall’s center (higher rents & smaller stores w/distance) the same as they did in the earlier study. 3) Store types have changed significantly from 1992 to 2006, e.g., fewer men’s shoe stores & more electronics stores, which makes testing for location effects of comparison store shopping difficult. Comparisons:1992 & 2006 Data Mean 1992 SF Std. Dev. Difference 2,417.06 ft 1992 SF Jewelry 1,239.57 ft 679.90 ft 1992 SF Fast Fd 874.69 ft 859.13 ft 1992 SF Wom A 3,855.74 ft 2,305.82 ft 3,084 ft 2,761 ft + 27% 2006 SF Jewelry 1,464.5 ft 205.64 ft + 18% 2006 SF Fast Fd 784 ft 254.64 ft - 10% 2006 SF Wom A 5,215 ft 1,665.8 ft + 35% 2006 SF 1992 $Sales $363.72/SF 1992 $Jewelry $656.37/SF $367.37/SF 1992 $Fast Fd $503.18/SF $268.42/SF 1992 $Wom Ap $252.01/SF $130.19/SF 2006 $Sales w/Inflation Comparisons: 1992 & 2006 Data Mean Std Dev Difference % Inflation % 1992 $Rent $37.75/SF $21.52/SF 1992 $Jewelry $63.84/SF 1992 $Fast Fd $59.05/SF 1992 $Wom Ap $30.00/SF 2006 $Rent $43.64/SF $24.92/SF + 15.6% +43.69% 2.5 2006 $Jewelry $57.29/SF $17.13/SF - 10% +43.69% 2006 $Fast Fd $82.07/SF $21.55/SF + 39% + 43.69% 2006 $Wom Ap $18.16/SF $13.91/SF - 39% +43.69% 2006 $Jewelry 2006 $Fast Fd 2006 $Wom Ap 1992 Ft Cent 295.2 ft 1992 Jewelry 281.6 ft 517.8 ft 485.4 ft Comparisons: 1992 & 2006 Data 1992 Fast Fd ft Mean Std Dev 70.5 ft 115.4 ft 1992 Wom Ap ft 312.7 ft 2006 Ft Cent 267.16 ft 2006 Jewelry ft 2006 Fast Fd ft 2006 Wom Ap ft Difference % 519.8 ft 130.78 - 9% Observations on Store Types, 1992 vs 2006 Data 1992 Tenants % of total 2006 Tenants % of total Fast Food 11% Fast Food 12% Jewelry 6.2% Jewelry 6.9% Specialty Food 6.3% Specialty Food 6.9% Men’s Apparel 6.3% Women’s App 18.5% Women’s App 14% Family Apparel 7% Men & Wom Ap 8.6% Women’s Shoes 5% Men Wom Shoes 8.6% Men’s Shoes 6.7% Home Furn 6.2% Home Furnish 3.4% Cards & Gifts 6.5% Cards & Gifts 5.2% Leisure & Enter 22% Leisure & Ent 8.6% Books 2% Services 14% by number of stores not GLA Did Rent per Square Feet Keep Up with Inflation? Definitely not, but it would be premature to say exactly why and how shopping center rents are falling historically. The only obvious conclusion is that regional & superregional shopping centers are not doing as well as they had been to internalize positive demand externalities compared with other retailing . Still I don’t think any other form of retailing surpasses the shopping center experience (if only we could impose sales taxes on online shopping). Retailing in urban areas is best done using regional & superregional models such as Harborplace, Baltimore. Future Research Similar research should be done using recent data from like regional & super-regional shopping centers. The best follow-up data would be for those 8 centers used in the 1st study. The 1992 data was from the best shopping centers then owned by a state pension fund. Choice of malls in 1992 by regional managers was meant to impress the pension fund managers. The follow-up data in 2006 was most probably from a crosssection of centers that the owners or lenders thought might be having financial difficulties. My Address & Phone Number Dr. Charles C. Carter 9034 S.W. 7th Street Boca Raton, FL 33433 954-708-3654 cccarter2010@yahoo.com