Publié par : Published by : Publicación de la :
Faculté des sciences de l’administration Université Laval Québec (Québec) Canada G1K 7P4 Tél. Ph. Tel. : (418) 656-3644 Fax : (418) 656-7047
Édition électronique : Electronic publishing : Edición electrónica :
Aline Guimont Vice-décanat - Recherche et partenariats Faculté des sciences de l’administration
Disponible sur Internet : Available on Internet Disponible por Internet :
http ://www.fsa.ulaval.ca/rd [email protected]
DOCUMENT DE TRAVAIL 2002-008
SHOPPING CENTER RENTS AND AGGLOMERATION ECONOMIES : PRELIMINARY FINDINGS FROM EMPIRICAL EVIDENCE François Des Rosiers Marius Thériault Ünsal Özdilek
Version originale : Original manuscript : Version original :
ISBN – 2-89524-145-7
Série électronique mise à jour : On-line publication updated : Seria electrónica, puesta al dia
06-2002
SSHHOOPPPPIINNGG CCEENNTTEERR RREENNTTSS AANNDD AAGGGGLLOOMMEERRAATTIIOONN EECCOONNOOMMIIEESS::
PPRREELLIIMMIINNAARRYY FFIINNDDIINNGGSS FFRROOMM EEMMPPIIRRIICCAALL EEVVIIDDEENNCCEE
Paper presented
at the 9th European Real Estate Society Conference, Glasgow, Scotland, June 4-7, 2002
by
François Des Rosiers, Ph.D., Faculty of Business Administration, Marius Thériault, Ph.D., Director, Land Planning Research Center,
and Ünsal Özdilek, Ph.D. Candidate,
Laval University, Quebec City, Canada
ABSTRACT This study is an attempt to model shopping centre rents and to investigate whether agglomeration economies are a driving force in the rent determination process. It is part of a research program based on physical and financial information obtained from seven super-regional, regional and community shopping centre managers in Quebec City for the 1998-2000 period. At this point, some 784 shops are covered in the study with gross base rent, operating expenses and gross leasable area being available for all of them. Partial information on lease duration as well as yearly sales is also available. While still preliminary, findings clearly suggest that larger centres generate substantial agglomeration economies for several categories of shops. Retail structure internal to shopping centres as well as cyclical factors also play a significant role in the rent determination process.
KEY WORDS:
Shopping Centres, Rents, Agglomeration Economies, Commercial Mix
____________________________________
1. OBJECTIVE AND CONTEXT OF RESEARCH
This study is an attempt to model shopping centre rents and to investigate whether
agglomeration economies are a driving force in the rent determination process. It is part of
a research program based on physical and financial information obtained from seven super-
regional, regional and community shopping centre managers in Quebec City for the 1998-
2000 period. At this point, some 784 shops are covered in the study with gross base rent,
operating expenses and gross leasable area being available for all of them. Partial
information on store frontage, lease duration as well as yearly sales is also available. When
2
completed, the data bank will include a series of shopping centre amenities and design
features (physical configuration, parking facilities, landscaping, etc.) as well as
neighbourhood characteristics (household composition and income) and urban externalities
(accessibility to services, proximity of competitors, etc.) in order to account for all major
internal and external determinants of shopping centre rents.
2. LITERATURE REVIEW
The academic literature on shopping malls has evolved around various theories of urban
spatial structure (Hotelling, 1929; Christaller, 1933; Lösh, 1940 and Alonso, 1964) with
strategies relating to mall configuration and store location within shopping centres
replicating those observed at the urban level (Vandell and Lane, 1987; Pearson, 1991;
Brueckner, 1993; Roulac, 1996; Brown, 1999). In contrast to what prevails in the
residential market (Follain and Malpezzi, 1980; Noland, 1980; Sirmans and Benjamin,
1991; Benjamin and Sirmans, 1994; Jud et al., 1996; Des Rosiers and Thériault, 1994 &
1996; Chinloy and Maribojoc, 1998) and office sector (Rosen, 1984; Hekman, 1985;
Gabriel and Nothaft, 1988; Wheaton and Torto, 1995; Sivitanides, 1997) where the rent
issue has been widely investigated, studies on the dynamics of commercial rent structuring
remain embryonic because of the confidential nature of the required information.
The mechanics underlying additional, or overage rents - expressed as a percentage of yearly
sales over and above a given, pre-negotiated threshold - are among the issues raised by
authors (Hartzell et al., 1987; Benjamin et al., 1990; Bruecker, 1993; Colwell et al., 1998;
Wheaton, 2000; Chun et al, 2001). Benjamin et al. (1990) were the first to apply hedonics
to the analysis of commercial rent. In their study, base rents derived from 103 commercial
leases pertaining to national, local and independent stores are regressed against sales,
discount rates, overage rents, lease terms, lease provisions, etc. Results suggest that while
base rents are lower where higher overage rent rates apply, they rise with higher sales
thresholds. In another similar study, Sirmans and Guidry (1993) point out that higher
consumer traffic levels are a prerequisite for the success of a store. In a totally different
urban context, Tay et al. (1991) investigate the Hong Kong commercial market. Their data
base includes 405 stores distributed among nine high-rise shopping centres. In contrast with
the literature, their study namely reveals that rent level is positively related to the age of a
shopping centre due to both customers’ fidelity - which tends to grow with time - and
3
continuous improvements to buildings. It also suggests that while the unit rent of a store is
positively correlated with the size of a centre, it is inversely related to its own size. This
latter finding brings us to the core of the current paper, that is agglomeration economies.
With location theories as the conceptual background (Weber, 1929), sales potential in
shopping centres are looked upon through the concepts of agglomeration economies and
externalities derived from the presence of anchor tenants (Eaton and Lipsey, 1983;
Mulligan, 1983; West et al., 1985; Ghosh, 1986; Ingene and Ghosh, 1990; Fisher and
Yezer, 1993; Eppli and Benjamin, 1993; Mejia and Benjamin, 2002) as well as from tenant
mix and product diversity (Konrad, 1984; Pashigan and Gould, 1998). Behind the concept
of agglomeration economies lies the reduction of consumer search and of uncertainty costs.
Such advantages allow major tenants to negotiate lower rents with shopping centres’
owners (Anderson, 1985), the fact that their departure may cause rental income to drop by
as much as 25% (Gatzlaff et al., 1994) greatly enhancing their bargaining power.
According to Nelson (1958) and Eppli and Shilling (1996), the clustering of similar stores
leads to an increase in their total sales level, thereby contributing to the success of the
shopping centre.
The image of a shopping centre may also impact upon sales level (Brown, 1992; Kirkup
and Rafiq, 1994; Anikeeff, 1996). It stems from consumers’ perception of major occupants
(Nevin and Houston, 1980), shopping centre size and configuration as well as the quality of
goods and services offered. In this respect, image is increasingly dependent upon fashion
(James et al, 1976; Jain and Etgar, 1976; Mazursky and Jacoby, 1986; Grewal, 1998).
Similarly, it affects tenants in their negotiation for an optimal location (Mejia et al., 2001).
Finally, accounting for all these features raises the spatial autocorrelation issue, recently
addressed by Carter and Haloupek (2000) on the grounds of previous work performed
mainly on the residential market (Griffith, 1987; Pace and Guilley, 1998; Dubin et al.,
1999).
4
3. DATA BANK AND ANALYTICAL APPROACH
3.1 Original Data Bank
As mentioned earlier, this study is based on physical and financial information obtained for
seven super-regional (2), regional (2) and community (3) shopping centres in Quebec City.
While financial data apply to the 1998-2000 period, leases were negotiated over a much
longer period that extends from 1976 to 2000. At this point, 784 stores are accounted for in
the study. Whereas net rent – defined as total yearly rent minus operating expenses - and
gross leasable area (GLA) is available for all retail units, partial information only is
available with respect to yearly sales (526 stores), store frontage (231) and lease duration
(621). Eight retail categories are distinguished in the analysis, namely jewellery stores (58),
clothing stores (264), shoeshops (57), restaurants (128), personal services (74) specialty
stores (151), kiosks (36) and anchor stores (16), all of which totalling nearly 2,4 million
square feet of GLA. As for shopping centre size, it is accounted for via dummy variables
used in interaction with other descriptors to generate hedonic coefficients that are specific
to each shopping centre category. Community (COM – 79 stores and less), regional (REG –
80 to 149 stores) and super-regional (SREG – 150 stores and over) centres’ rented space
roughly amount to 283 000 sq. ft., 612 000 sq. ft. and 1 489 000 sq. ft., respectively. GLA
distribution by retail category and shopping centre size is displayed in Exhibit 1 while
Exhibit 2 provides an operational definition for each variable used in the modelling process.
3.2 Structural and Cyclical Attributes
In addition to basic financial and physical characteristics, a series of five market structure
and cyclical attributes are developed in order to capture the effect on rent of both
commercial dynamics within shopping centres and overall economic situation. These are:
Lease duration (LEASEDURN): It might be expected that risk reduction deriving
from a longer lease term should result in a rent discount for the tenant, provided that
regional commercial rents at the time of negotiation are relatively stable or
declining; otherwise, a longer than average term may penalize the landlord.
An agglomeration index (AGGLOMINDX): Defined as the ratio of GLA for a
given retail category within a shopping centre to total GLA in that centre, this index
5
measures agglomeration economies resulting from the amalgamation of a large
number of tenants in a given field of retail activity and, therefore, from the
increased consumer traffic it implies. It should be expected that the larger the index,
the higher the rent that the landlord can extract from tenants in that category.
A concentration index (CONCNINDX): While the landlord can take advantage of
the abundance of retailers in a given commercial category, he will in contrast be
prone to grant a rent discount where relatively few tenants are in control of the field,
since any defection might induce a severe drop in yearly revenues. The Herfindahl
Index, designed at measuring industrial concentration, is used here. Since it
accounts for both the number of stores in a retail category and their relative share of
the total GLA, it mirrors the level of competition in a field and the bargaining power
of tenants. It can be assumed that the larger the index, the lower the rent paid.
The GLA open to negotiation (OPENLEASE) in any given year is also a good
indication of the bargaining power of tenants in a shopping centre. Indeed, where a
substantial proportion of total GLA is under negotiation at the same time, the
landlord should be expected to grant rent discounts in order to avoid massive
departures to competitors. Therefore, the higher the index, the lower the rent.
Finally, economic cycles (CYCLE91_92) should theoretically affect the negotiation
process. An analysis of the consumption expenses of households and retail sales in
the Quebec metropolitan area between 1981 and 2001 (expressed in 1992 constant
dollars) indicates that the 1991-92 period is characterized by a drop in commercial
activity. Thus, the coefficient of this dummy variable should display a negative
sign.
3.3 Analytical approach
Two functional forms are used in the study, namely the linear and log-linear ones; a
logarithmic transformation is also applied to the GLA variable. Exhibit 3, displaying basic
descriptive statistics for all variables, clearly indicates that neither NETRENT nor GLA are
normally distributed; Figure 1 provides further evidence of this. Hence the rationale for
such transformations. While included in the data base and available for some 526 stores,
unit sales (SALES/SQFT) are not included in the analysis at this stage of the research. The
6
final calibration of the equations involves the removal of 15 extreme residuals considered
to be highly detrimental to the model performances; these represents less than 2% of the
initial data base. Finally, the regression procedure adopted is a combination of both
standard and stepwise approaches.
4. MAJOR FINDINGS
4.1 The Linear Form – Model A
As can be seen from Exhibit 4, the linear model yields interesting results in spite of the non-
linearity of the dependent variable. While the explanatory performance of Model A, which
displays an adjusted R-Square of 0,714 and a F value of 84,2, is quite fair, its predictive
performance remains rather weak: considering the global average rent ($64), the relative
SEE stands at 38,4%, which was to be expected in the light of the highly skewed
distribution of the NETRENT/SQFT variable. Examination of the regression coefficients
suggests that no inconsistency can be detected with respect to either their magnitude or
sign, with the exception of the ANCHOR parameter estimate which is unexpectedly
positively signed whereas it should be negative. Safe for a few coefficients whose statistical
significance do not meet the 0,10 (CYCLE91_92; SPECIALTY*SREG) or the 0,05
(SHOESTORE*COM; AGGLOMINDX*REG) threshold, all other descriptors exhibit
strong t values. Most interestingly, the two structural attributes indicative of tenants’
bargaining power (CONCNINDX*SREG and OPENLEASE*REG) emerge as significant
and with the right sign.
Considering that the linear model may not provide the best indications on the hedonic
shopping centre rent function, let us turn to Model B for a more reliable interpretation of
the coefficients obtained via the log-linear form.
4.2 The Log-Linear Form – Model B
Regression results for Model B are shown in Exhibit 5. While overall performances
obtained with the log-linear form are quite similar to those derived from the linear one1, the
1 It should be kept in mind that performance tests obtained with a linear form cannot be compared with those derived from a non-linear function without first applying the test suggested by Box and Cox (1964). The latter consists in computing the linear model using the transformed dependent variable Yi / YG , with YG being the geometric mean of the Y.
7
statistical significance of parameter estimates is substantially improved for most
descriptors. Only one coefficient (KIOSK*REG) is not significant at the 0,05 level while
all coefficient signs – including that of the ANCHOR variable - are in line with theoretical
expectations. Most important though, four out of five structural and cyclical variables
emerge as statistically significant, lease duration (LEASEDURN) alone being rejected from
the model. As for excessive multicollinearity, VIFs suggest that there is none, the highest
VIF value standing at 2,89, which is well below the critical threshold of 10. In spite of that,
it is interesting to note that there might be a structural link between store agglomeration
(AGGLOMINDX*REG) and the clothing retail sector (CLOTHING*REG) in regional
shopping centres.
In order to make an adequate interpretation of the implicit rents of commercial attributes
derived from Model B, it is appropriate to keep in mind the retail structure of shopping
centres. In that respect, Exhibit 1 may prove very helpful.
4.3 Interpreting the coefficients of basic physical and retail category attributes
As expected, the prominent role of gross leasable area in the retail rent determination
process clearly emerges from the model, the negative sign of the Ln_GLA coefficient
indicating that unit rent decreases with store size. More precisely, increasing average store
size (3 000 sq. ft.) by 10% will result in a 2,4% drop in unit rent2. As shown by the Beta,
standardized coefficient, frontage is also a major determinant3 and exerts a positive
influence on rent due to the enhanced visibility and attractiveness it confers to a store.
Thus, raising the mean frontage (34 ft.) by 10% increases unit rent by roughly 1,5%.
Quite interestingly, the presence of a jewellery impacts differently on rents depending on
the size category of the shopping centre it belongs to. In community shopping centres
(JEWELRY*COM), the effect is negative whereas it is positive in both regional
(JEWELRY*REG) and super-regional (JEWELRY*SREG) structures. A twofold
explanation can be brought forward: on the one hand, the relative scarcity of this retail
activity in community centres (6,8% of stores and 1,6% of GLA) as opposed to regional
2 This marginal impact is obtained by applying mean values to all model variables and then simulating a 10% increase in GLA, from 3 000 to 3 300 sq. ft. As a result, unit rent drops by 2,4%, from $38,53 to $37,59. 3 In the current state of the data base, this attribute is only available for super-regional centres.
8
centres where jewellers are in excess (25,7% of stores and 3,0% of GLA) translates into a
rent discount in the former case and in a location premium in the latter case. On the other
hand, jewellery stores located in community centres typically offer cheaper products than
those found in more glamorous, either regional4 or super-regional ones that also benefit
from agglomeration economies; hence the positive sign of the latter coefficients. The same
rationale, combining relative scarcity, prestige location and agglomeration economies,
applies – with even greater relevance – to clothing and shoe stores located in community
(negative sign) and regional (positive sign) shopping centres.
Agglomeration economies are clearly at stake when it comes to interpreting the marginal
contribution of restaurants to unit rents, with regional and super-regional centres
commanding substantially higher rents than community ones. Rent patterns differ though
with respect to personal services and specialty stores. In spite of their relative abundance,
the negative coefficients assigned to the SERVICES*COM and SPECIALTY*COM
estimates reflect the bargaining power of local retailers that are at the very core of
community centres’ mission, namely proximity services. In contrast to all other statistically
significant retail categories, specialty stores located in super-regional centres
(SPECIALTY*SREG) also command lower unit rents: according to the “combined” and
“compared” purchase concept, consumers are attracted by high-tech specialty stores
(computer, audio and video stores) which, for that reason, benefit from rent discounts;
hence the negative sign attached to the related coefficient.
Kiosks are found only in regional and super-regional shopping centres. Since they occupy
the central alley of shopping malls where consumer traffic is optimal, they generate very
high sales per square foot and, consequently, command the highest unit rents of all retail
categories. Their contribution to the retail rent determination process is among the
interesting findings of this study. While the KIOSK*REG coefficient is only significant at
the 0,10 level, the parameter estimate of the KIOSK*SREG descriptor (Beta coefficient of
0,286) undoubtedly suggests that kiosks located in super-regional centres substantially
increase their overall profitability; actually, the unit rent they pay represents a 144%
4 It is worth noting that among the two regional shopping centres included in the study, one is located in a rather low-class, popular environment while the other attracts high-income, most sophisticated customers. Considering that bias, the magnitude of the regional centres-related coefficients would project a different picture if the “low-profile” centre were distinguished from the “high-profile” one.
9
premium over the base (intercept) rent. Finally, anchor tenants do benefit, as expected,
from a major rent discount; while it amounts to roughly 24% on average, it can be much
higher in some cases.
4.4 Interpreting the coefficients of structural and cyclical descriptors
Central to this study, findings pertaining to agglomeration economies in shopping centres
corroborate the literature on the subject. They clearly suggest that while larger size
shopping centres involve greater competition among tenants, store clustering also generate
substantially higher sales volumes, which translates into higher rents. When applied to
regional shopping centres (AGGLOMINDX*REG), a 10% increase in the mean
agglomeration index (0,14) results in a 1,3% rise in unit rent. In contrast, and according to
theoretical expectations, concentration levels in retail activity, as measured by the
Herfindahl Index, impact negatively on rents. In as much as super-regional shopping
centres are concerned (CONCNINDX*SREG), a 10% rise in the average concentration
index (0,11) translates into a 1,5% drop in rent. Tenants who negotiate their lease while a
large number of retailers do the same do experience, as expected, a strengthening of their
bargaining power. In regional shopping centres, every 10% increase in the proportion of
open leases (OPENLEASE*REG) results in a 0,8% reduction in unit rent. This raises the
importance of lease renewal strategy for shopping centre owners. Finally, a recessionary
economic cycle affects retail rents negatively. Over the 1991-92 period, overall shopping
centre rents in the Quebec metropolitan region dropped by 12%.
5. SUMMARY OF FINDINGS AND SUGGESTIONS FOR FURTHER RESEARCH
5.1 Summary of findings
This study is an attempt to model shopping centre rents and to investigate whether
agglomeration economies are a driving force in the rent determination process. It is based
on physical and financial information obtained for seven super-regional, regional and
community shopping centres in Quebec City. While financial data apply to the 1998-2000
period, leases were negotiated over a longer period that extends from 1976 to 2000. At this
point, 784 stores are accounted for in the study. In addition to basic financial and physical
characteristics, a series of five market structure and cyclical attributes are developed in
10
order to capture the effect on rent of both commercial dynamics within shopping centres
and overall economic situation. Main findings can be summarized as follows:
As expected, the prominent role of gross leasable area in the retail rent determination process clearly emerges from the model, the negative sign of the Ln_GLA coefficient indicating that unit rent decreases with store size;
Frontage is also a major determinant and exerts a positive influence on rent due to the enhanced visibility and attractiveness it confers to a store;
Quite interestingly, the presence of a jewellery impacts differently on rents depending on the size category of the shopping centre it belongs to. In community shopping centres, the effect is negative whereas it is positive in both regional and super-regional structures. A combination of relative scarcity, prestige location and agglomeration economies provides a sensible explanation to such findings;
A similar rationale applies – with even greater relevance – to clothing and shoe stores located in community and regional shopping centres;
Agglomeration economies are clearly at stake when it comes to interpreting the marginal contribution of restaurants to unit rents, with regional and super-regional centres commanding substantially higher rents than community ones;
Rent patterns differ though with respect to personal services and specialty stores. In spite of their relative abundance, the negative contribution of this retail category in community shopping centres reflects the bargaining power of local retailers that are at the very core of community centres’ mission, namely proximity services;
Specialty stores located in super-regional centres also command lower unit rents due to the attraction they exert on high-tech goods consumers;
Since they occupy the central alley of shopping malls where consumer traffic is optimal, kiosks command the highest unit rents of all retail categories;
Anchor tenants do benefit, as expected, from a major rent discount which amounts, on average, to 24% of base rent;
Findings clearly suggest that while larger size shopping centres involve greater competition among tenants, store clustering generates substantial agglomeration economies that translate into higher rents;
Concentration levels in retail activity, as measured by the Herfindahl Index, impact negatively on rents;
The proportion of leasable space under negotiation in a given year strengthens the bargaining power of tenants, who consequently benefit from a rent discount;
Finally, a recessionary economic cycle affects retail rents negatively.
11
In short, this study corroborates previous research findings about the critical role played by
anchor tenants in the generation of agglomeration economies as well as the importance of
tenant mix, product diversity, quality of goods and image in the retail rent structuring
process. While increased attractiveness has an overall positive effect on both sales and unit
rents for most stores - namely in the case of kiosks - , they allow major tenants and other
specialty goods retailers to negotiate lower rents with shopping centres’ owners. Such
factors however impact differently on rents depending on shopping centre size and profile.
Finally, retail structure within shopping centres and lease renewal strategies by owners will
affect tenants’ bargaining power and, therefore, rent levels.
5.2 Suggestion for further research
While this study is but a preliminary investigation into the retail rent dynamics, it raises
several interesting issues namely with respect to the measurement of externalities and
agglomeration, as opposed to dispersion, effects. While shopping centre configuration and
design deserve further attention, resorting to concepts and analytical tools found in
geography and spatial economics to measure such phenomena may lead to major
developments in the field.
12
REFERENCES Alonso, W., (1964). Location and Land Use. Toward a General Theory of Land Rent, Cambridge, Ma.,
Harvard University Press, 205 pages. Anderson, P. M. (1985). Association of Shopping Center Anchors with Performance of Non-Anchor
Speciality Chain’s Stores, Journal of Retailing, 61, pp. 61-74. Anikeeff, M. (1996). Shopping Center Tenant Selection and Mix: a Review, in J. Benjamin, ed., Research
Issues in Real Estate, Vol. 3 : Megatrends in Retail Real Estate, Norwell : Kluwer Academic Publishers. Benjamin, J. D. and Sirmans, G. S. (1994). Apartment Rent : Rent Control and Other Determinants, Journal
of Property Research, 11, pp. 27-50. Benjamin, J. D., Boyle, G. W. and Sirmans, C. F. (1990). Retail Leasing : Determinants of Shopping Center
Rents, Journal of the American Real Estate and Urban Economics Association (AREUEA), 18 :3, pp. 302-312.
Box, G. E. P., and D. R. Cox (1964) An analysis of transformations, Journal of the Royal Statistical Society, 26, 211-243.
Brown, S. (1992). Tenant Mix, Tenant Placement, and Shopper Behavior in a Planned Shopping Centre, The Services Industries Journal, 12 :3, pp. 384-403.
Brown, G. M. (1999). Design and Value : Spatial Form and the Economic Failure of a Mall, Journal of Real Estate Research, 17:2, pp. 189-225.
Brueckner, J. (1993). Inter Store Externalities and Space Allocation in Shopping Centers, Journal of Real Estate Finance and Economics, 7:1, pp. 5-17.
Carter, C. and Haloupek, W. J. (2000). Spatial Autocorrelation in a Retail Context, International Real Estate Review, 3:1, pp. 34-48.
Chinloy, P. and Maribojoc, E. (1998). Expense and Rent Strategies in Real Estate Management, Journal of Real Estate Research, 15:3, pp. 267-282.
Christaller, W. (1933). Die Zentralen Orte in Suddeutschland, Verlag. Traduit par Baskin, C.W., (1966), Central Places in Southern Germany, Englewood Cliffs, N. J., Prentice-Hall, 230 pages.
Chun, G. H., Epli, M. J. and Shilling, J. D. (2001). A simulation analysis of the relationship between retail sales and shopping center rents, The Journal of Real Estate Research, 21:3, pp. 163-186.
Colwell, Peter F. and Munneke, H. J. (1998). Percentage Leases and the Advantages of Regional Malls, Journal of Real Estate Research, 15 :3, pp. 239-252.
Des Rosiers, F. and Thériault, M. (1994). Implicit Prices of Rental Services : Modeling the Quebec Market, Assessment Journal, 1: 4, pp. 47-60.
Des Rosiers, F. and Thériault, M. (1996). Rental Amenities and the Stability of Hedonic Prices : A Comparative Analysis of Five Market Segments, The Journal of Real Estate Research, 12 :1, pp. 17-36.
Dubin, R. A., Pace, R. K. and Thibodeau, T. G. (1999). Spatial Autoregression Techniques for Real Estate Data, Journal of Real Estate Literature, 7: 2, pp. 79-95.
Eaton, B. C. and Lipsey, R. G. (1983). An Economic Theory of Central Places, Economic Journal, 92, pp. 56-72.
Eppli, M. J. and Benjamin, J. D. (1993). The Evolution of Shopping Center Research : A Review and Analysis, Journal of Real Estate Research, 9 :1, pp. 5-32.
Eppli, M. J. and Shilling, J.D. (1996). How Critical is a Good Location to a Regional Shopping Centre ?, Journal of Real Estate Research, 12:3, pp. 459-469.
Fisher, J. D. and Yezer, A. (1993). Spatial Structure and Rents in Shopping Centers, Working paper, Indiana University) et de Losh (Losch, A. (1938). The Nature of Economic Regions, Southern Economic Journal, 5, pp. 71-78.
Follain, J. R. and Malpezzi, S. (1980). Dissecting Housing Value and Rent, Washington, D. C. : Urban Institute, Working paper UI249-17.
Gabriel, S. and Nothaft, F. (1988). Rental Housing Markets and the Natural Vacancy Rate, AREUEA Journal, 16, pp. 419-429.
13
Gatzlaff, D. H., Sirmans, G. S. and Diskin, B. A. (1994). The Effect of Anchor Tenant Loss on Shopping Center Rents, The Journal of Real Estate Research, 9 :1, pp. 99-110.
Ghosh, A. (1986). The Value of a Mall and Other Insights from a Revised Central Place Model, Journal of Retailing, 62, pp. 244-249.
Grewal, D. R., Krishnan, R., Baker, J. and Borin, N. (1998). The Effect of Store Name, Brand Name and Price Discounts on Consumers’ Evaluations and Purchase Intentions, Journal of Retailing, 74 :3, pp. 331-352.
Griffith, D. A. (1987). Spatial Autocorrelation : A Primer, Washington D. C. : Association of American Geographers.
Hartzell, D. J., Shulman D. G. and Wurtzebach, C. H. (1987). Refining the Analysis of Regional Diversification for Income-Producing Real estate. The Journal of Real Estate Research 2 :2, pp. 85-95.
Heckman, J. (1985). Rental Price Adjustment and Investment in the Office Market, AREUEA Journal, 13, pp. 23-42.
Hotelling, H., (1929). Stability in Competition, The Economic Journal, 39, pp. 41-57. Ingene, C. A. and Ghosh, A. (1990). Consumer and Producer Behaviour in a Multipurpose Shopping
Environment, Geographic Analysis, 22, pp. 70-91. Jain, A. J. and Etgar, M. (1976). Measuring Store Image Through Multidimensional Scaling of Free Response
Data, Journal of Retailing, 52 :4, pp. 61-70. James, D. L., Durand, R. M. and Dreves, R. A. (1976). The Use of a Multi-Attribute Attitude Model in a Store
Image Study, Journal of Retailing, 52 :2, pp. 23-31. Jud, D. G., Benjamin, J. D. and Sirmans, G. S. (1996). What Do We Know about Apartments and Their
Markets ? Journal of Real Estate Research, 11, pp. 243-258. Kirkup, M. and Rafiq, M. (1994). Managing Tenant Mix in New Shopping Centers, International Journal of
Retail and Distribution Management, 22: 6, pp. 29-37. Konrad, S. (1984). Location and Spatial Pricing with non-convex Transportation Schedules, Bell Journal of
Economics, 12. Lösch, A. (1940). Die räumliche Ordnung der Wirtschaft, Jena. Traduit par Woglom, W.H. et Stolper, W.F.,
(1954), The Economics of Location, New York, Yale University. Mejia, Luis C. and Benjamin, John, D. (2002). What Do We Know About the Determinants of Shopping
Center Sales ? Spatial vs. Non-Spatial Factors, Journal of Real Estate Literature, 10:1, pp. 3-26. Mazursky, D. and Jacoby, J. (1986). Exploring the Development of Store Images, Journal of Retailing, 62 :3,
pp. 145-163. Mejia, Luis C., Eppli, M. J. and Benjamin, J., D. (2001). Retail Tenant Rents in a Market with Shopping Mall
Image Differentiation, Working Paper. Mulligan, G. F. (1983) Consumer Demand and Multipurpose Shopping Behavior, Geographical Analysis, 15,
pp. 76-81. Nelson, R. L. (1958). The Selection of Retail Locations, New York : Dodge. Nevin, J. R. and Houstan, M. J. (1980). Image as a Component of Attraction to Intraurban Shopping Areas,
Journal of Retailing, 56, pp. 77-93. Noland, C. W. (1980). Assessing Hedonic Indexes for Housing, Santa Monica, Calif. : Rand Corporation,
Working paperN-1505-HUD. Pace, R. K. and Gilley, O. W. (1998). Generalizing the OLS and the Grid Estimator, Real Estate Economics,
26: 1, pp. 331-347. Pashigan, B. and Gould, E. (1998). Internalizing Externalities : The Pricing of Space in Shopping Centers,
Journal of Law and Economics, XLI :1, pp. 115-142. Pearson, T. D., (1991). Location ! Location ! Location ! What is Location, The Appraisal Journal, 59 :1, pp.
7-20. Rosen, K. T. (1984). Towards a Model for the Office Building Sector, AREUEA Journal, 12, pp. 261-269. Roulac, S. E. (1996). Foreword State of the Discipline : Malaise or Renaissance ? Journal of Real Estate
Research, 12 :2, pp. 111-123.
14
Sirmans, G. S. and Benjamin, J. D. (1991). Determinants of Market Rent, Journal of Real Estate Research, 6, pp. 357-379.
Sirmans, C. F. and Guidry, K. A. (1993). The determinants of shopping centre rents, The Journal of Real Estate Research, 8 :1, pp. 107-115.
Sivitanides, P. S. (1997). The Rent Adjustment Process and the Structural Vacancy Rate in the Commercial Real Estate Market, Journal of Real Estate Research, 13:2, pp. 195-209.
Tay, R. S., Lau, C. K. and Leung, M. S. (1991) The Determination of Rent in Shopping Centers : Some Evidence from Hong Kong. Journal of Real Estate Literature, Cleveland, 7:2, pp. 183-196.
Vandell, K. D. and Lane, J. S. (1987). The Economics of Architecture and Urban Design : Some Preliminary Findings, Journal of the American Real Estate and Urban Economics Association, 17:2, pp. 235-265.
Weber, A. (1929). Theory of the Location of Industries, Chicago, University of Chicago Press, 259 pages. West, D. S., Von Hohenbalken, B. et Kroner, K. (1985). Tests of Intra urban Central Place Theories,
Economic Journal, 95, pp. 101-117. Wheaton, W. C. and Torto, R., G. (1995). Office Rent Indices and their Behavior Over Time, Journal of
Urban Economics, 35:2, pp. 121-139. Wheaton, W. C. (2000). Percentage Rent in Retail Leasing : The Alignment of Landlord – Tenant Interests,
Real Estate Economics, 28: 2, pp. 185-204.
________________________________________________
15
Exhi
bit 1
: Str
uctu
ring
Dat
a B
ase
by R
etai
l Cat
egor
y an
d Ty
pe o
f Sho
ppin
g C
entr
e
Nb.
%G
LA (s
f)%
Nb.
%G
LA (s
f)%
JEW
ELR
Y5
6,8%
4 63
71,
6%19
8,6%
18 4
573,
0%C
LOTH
ING
1317
,6%
23 7
148,
4%82
37,1
%21
8 12
835
,6%
SHO
ES2
2,7%
4 99
51,
8%22
10,0
%39
796
6,5%
RES
TAU
RAN
T12
16,2
%12
983
4,6%
3415
,4%
33 6
985,
5%SE
RVI
CES
2027
,0%
41 8
2614
,8%
2210
,0%
34 6
205,
7%SP
ECIA
LTY
1925
,7%
120
397
42,6
%36
16,3
%69
749
11,4
%AN
CH
OR
34,
1%74
148
26,2
%5
2,3%
197
699
32,3
%K
IOSK
00,
0%0
0,0%
10,
5%23
80,
0%To
tal
7410
0,0%
282
700
100,
0%22
110
0,0%
612
385
100,
0%
Nb.
%G
LA (s
f)%
Nb.
%G
LA (s
f)%
JEW
ELR
Y34
7,0%
25 4
221,
7%58
7,4%
48 5
162,
0%C
LOTH
ING
169
34,6
%39
7 03
726
,7%
264
33,7
%63
8 87
926
,8%
SHO
ES33
6,7%
58 3
423,
9%57
7,3%
103
133
4,3%
RES
TAU
RAN
T82
16,8
%56
923
3,8%
128
16,3
%10
3 60
44,
3%SE
RVI
CES
326,
5%38
048
2,6%
749,
4%11
4 49
44,
8%SP
ECIA
LTY
9619
,6%
307
394
20,6
%15
119
,3%
497
540
20,9
%AN
CH
OR
81,
6%60
0 53
940
,3%
162,
0%87
2 38
636
,6%
KIO
SK35
7,2%
5 61
00,
4%36
4,6%
5 84
80,
2%To
tal
489
100,
0%1
489
315
100,
0%78
410
0,0%
2 38
4 40
010
0,0%
SU
PER
_REG
ION
AL (2
)
A
ll C
ateg
orie
s (7
)St
ore
Cat
egor
y
Stor
e C
ateg
ory
CO
MM
UN
ITY
(3)
R
EGIO
NAL
(2)
16
Exhibit 2: Operational Definition of Variables
N.B.: M = Metric variable; D = Dummy variable
V A R IA B L E
O PE R A T IO N A L D E FIN IT IO N T Y PE
N E T R E N T /SQ F T N et rent / sq . foot (D ependent variable); expressed as total gross rent (base + overage) m inus transferable operating expenses.
M
B asic A ttributes G L A G ross Leasable A rea (sq. feet). M
SA L E S/SQ F T Store yearly realized sales / G LA . M
F R O N TA G E Store frontage (feet) M
JEW E L R Y Jew elry retailer. D
C L O T H IN G C lothing retailer; includes w om en 's, m en 's and children 's read-to -w ear clothes and accessories.
D
SH O ESTO R E Shoe retailer; includes fam illy shoes, w om en 's, m en 's & boy's shoes, athletic footw ear, unisex / jean store.
D
R E ST A U R A N T Food services; includes restaurant & fast food (w ith/out liquor), sandw ich & pizzas stores, candy & nuts stores.
D
SE R V IC E S Personal and financial services; includes banks & insurance, m edical & dental offices, beauty salons, cleaners & dryers, unisex hair, barbershops, travel agents.
D
SP E C IA L TY Specialty and gifts retailer; includes radio, video & m usic centres, cards & gifts, books, decorative accessories, eyeglasses-optician stores.
D
A N C H O R A nchor tenant; large, chain stores having betw een 20 000 et 200 000 sq. feet of G LA - U sually seen as providing both stability and custom er draw ing pow er to the shopping centre (W al-M art, S im ons, T oys' R ' U S, The Bay, Sears, etc.).
D
K IO SK K iosk store; sm all, light structure w ith open sides, usually located at a central position of the m all, close to high pedestrian traffic and having a G LA of betw een 70 and 300 sq. feet (cellular phone, candy and lottery stores).
D
Structural and cyclical attributes LE A SED U R N Lease duration (years) M
A G G L O M IN D X A gglom eration index, expressed as the ratio of G LA for a given retail category w ithin a shopping center to total G LA in that center.
M
C O N C N IN D X H erfindahl Index, used as a m easure of the concentration of a retail category w ithin a given shopping center; it is expressed as the sum of the squared proportion of each store’s G LA in that category.
M
O P EN L EA SE Leasable area open to negotiation during a given year, in a given shopping center, as a proportion of total G LA in that shopping centre; it is used as a m easure of the bargaining pow er of tenants.
M
C Y C L E91_92 Lease w as negotiated over the 1991-1992 recessionary period. D
17
Exhi
bit 3
: Des
crip
tive
Stat
istic
s
VARI
ABLE
SNb
. Com
pute
dTy
peM
ean
Med
ian
Mod
eSt
d. D
ev.
Min
.M
ax.
NETR
ENT/
SQFT
($)
784
M64
5430
460,
430
8G
LA (S
q.ft)
784
M3
041
1 23
32
000
10 8
2922
163
034
SALE
S/SQ
FT1 (
$)52
6M
431
339
214
335
42
627
FRO
NTAG
E*SR
EG2 (F
t.)23
1M
3428
3320
913
6JE
WEL
RY58
D0,
070
1CL
OTH
ING
264
D0,
340
1SH
OES
TORE
57D
0,07
01
REST
AURA
NT12
8D
0,16
01
SERV
ICES
74D
0,09
01
SPEC
IALT
Y15
1D
0,19
01
ANCH
OR
16D
0,02
01
KIO
SK36
D0,
050
1LE
ASED
URN3
(Yea
rs)
621
M10
1010
51
47AG
GLO
MIN
DX78
4M
0,14
0,08
0,22
0,13
0,00
0,86
CONC
NIND
X78
4M
0,11
0,08
0,02
0,13
0,02
1O
PENL
EASE
362
1M
0,08
0,08
0,00
0,06
0,00
0,50
CYCL
E91_
9237
D0,
050
1
1. S
ales
wer
e no
t ava
ilabl
e fo
r one
sup
er-re
gion
al s
hopp
ing
cent
re (2
58 c
ases
mis
sing
).2.
Fro
ntag
e w
as a
vaila
ble
for o
nly
one
supe
r-reg
iona
l sho
ppin
g ce
ntre
(FR
ON
TAG
E*SR
EG, 2
31 c
ases
).3.
Sin
ce y
ear o
f lea
se n
egoc
iatio
n w
as a
vaila
ble
for 6
21 s
tore
s on
ly (1
63 c
ases
mis
sing
), bo
th le
ase
dura
tion
(LEA
SED
UR
N)
an d
pro
porti
on o
f ope
n le
ases
(OPE
NLE
ASE)
wer
e co
mpu
ted
on th
is g
roun
d.
18
Fig
ure
1: N
et U
nit R
ent a
nd G
LA D
istr
ibut
ions
NET
REN
T/SQ
FT
300,0
280,0
260,0
240,0
220,0
200,0
180,0
160,0
140,0
120,0
100,0
80,0
60,0
40,0
20,0
0,0NET
REN
T/SQ
FT
Frequency
160
140
120
100 80 60 40 20 0
Std.
Dev
= 4
5,92
M
ean
= 63
,7
N =
784
,00
GLA
1600
00,0
1500
00,0
1400
00,0
1300
00,0
1200
00,0
1100
00,0
1000
00,0
9000
0,0
8000
0,0
7000
0,0
6000
0,0
5000
0,0
4000
0,0
3000
0,0
2000
0,0
1000
0,0
0,0GLA
Frequency
700
600
500
400
300
200
100 0
Std.
Dev
= 1
0829
,43
M
ean
= 30
41,3
N =
784
,00
19
Exhibit 4: Model A - Linear Form / N=769 / K=23
Model Summaryb
,850 ,722 ,714 24,60Model1
R R SquareAdjusted R
SquareStd. Error of the
Estimate
Dependent Variable: NETRENT/SQFTb.
ANOVAb
1171410 23 50931 84,162 ,000450838 745 605
1622248 768
RegressionResidualTotal
Model1
Sum ofSquares df Mean Square F Sig.
Dependent Variable: NETRENT/SQFTb.
Coefficientsa
157,98 8,44 18,71 ,000-14,88 1,12 -,371 -13,25 ,000 2,10
,32 ,05 ,131 6,03 ,000 1,25-29,40 11,21 -,051 -2,62 ,009 1,0325,64 6,20 ,087 4,13 ,000 1,1826,48 4,70 ,119 5,63 ,000 1,19
-30,21 7,37 -,082 -4,10 ,000 1,0615,34 4,89 ,103 3,14 ,002 2,89
-29,64 17,51 -,033 -1,69 ,091 1,0116,67 5,85 ,061 2,85 ,005 1,21
-22,04 7,72 -,057 -2,85 ,004 1,0735,51 5,11 ,159 6,94 ,000 1,4029,69 3,52 ,195 8,42 ,000 1,44
-25,82 5,98 -,087 -4,32 ,000 1,10-25,64 6,29 -,082 -4,08 ,000 1,0915,07 5,54 ,068 2,72 ,007 1,69-4,65 3,16 -,033 -1,47 ,141 1,3657,34 24,77 ,045 2,32 ,021 1,01
122,24 5,14 ,555 23,76 ,000 1,4622,15 8,67 ,067 2,55 ,011 1,8330,62 16,65 ,058 1,84 ,066 2,63
-60,07 19,51 -,081 -3,08 ,002 1,87
-78,57 25,94 -,080 -3,03 ,003 1,86
-4,48 4,30 -,021 -1,04 ,298 1,05
(Constant)
Ln_GLAFRONTAGE*SREGJEWELRY*COMJEWELRY*REGJEWELRY*SREGCLOTHING*COMCLOTHING*REGSHOESTORE*COMSHOESTORE*REGRESTAURANT*COMRESTAURANT*REGRESTAURANT*SREGSERVICES*COMSPECIALTY*COMSPECIALTY*REGSPECIALTY*SREGKIOSK*REGKIOSK*SREGANCHORAGGLOMINDX*REGCONCNINDX*SREGOPENLEASE*REGCYCLE91_92
B Std. Error
Unstand'zd Coeff.
Beta
Stand'zdCoeff.
t Sig. VIF
CollinearityStatistics
Dependent Variable: NETRENT/SQFTa.
20
Exhibit 5: Semi-Log Model / N=769 / K=23
Model Summaryb
,843 ,711 ,702 ,3543Model1
R R SquareAdjusted R
SquareStd. Error of the
Estimate
Dependent Variable: Ln_NETRENT/SQFTb.
ANOVAb
231 23 10,024 79,845 ,00094 745 ,126
324 768
RegressionResidualTotal
Model1
Sum ofSquares df Mean Square F Sig.
Dependent Variable: Ln_NETRENT/SQFTb.
Coefficientsa
5,725 ,122 47,07 ,000-,259 ,016 -,457 -16,04 ,000 2,10,006 ,001 ,177 8,02 ,000 1,25
-,780 ,161 -,097 -4,83 ,000 1,03,345 ,089 ,083 3,86 ,000 1,18,291 ,068 ,092 4,29 ,000 1,19
-,928 ,106 -,177 -8,74 ,000 1,06,206 ,070 ,098 2,92 ,004 2,89
-1,189 ,252 -,093 -4,71 ,000 1,01,295 ,084 ,076 3,50 ,000 1,21
-,518 ,111 -,095 -4,66 ,000 1,07,429 ,074 ,136 5,82 ,000 1,40,361 ,051 ,168 7,12 ,000 1,44
-,800 ,086 -,191 -9,28 ,000 1,10-,819 ,091 -,186 -9,05 ,000 1,09,191 ,080 ,061 2,40 ,017 1,69
-,114 ,045 -,057 -2,50 ,013 1,36,591 ,357 ,033 1,66 ,098 1,01,891 ,074 ,286 12,02 ,000 1,46
-,269 ,125 -,057 -2,15 ,032 1,83,940 ,240 ,125 3,92 ,000 2,63
-1,350 ,281 -,129 -4,80 ,000 1,87
-1,028 ,374 -,074 -2,75 ,006 1,86
-,129 ,062 -,042 -2,08 ,038 1,05
(Constant)
Ln_GLAFRONTAGE*SREGJEWELRY*COMJEWELRY*REGJEWELRY*SREGCLOTHING*COMCLOTHING*REGSHOESTORE*COMSHOESTORE*REGRESTAURANT*COMRESTAURANT*REGRESTAURANT*SREGSERVICES*COMSPECIALTY*COMSPECIALTY*REGSPECIALTY*SREGKIOSK*REGKIOSK*SREGANCHORAGGLOMINDX*REGCONCNINDX*SREGOPENLEASE*REGCYCLE91_92
BStd.Error
Unstand'zd Coeff.
Beta
Stand'zdCoeff.
t Sig. VIF
CollinearityStatistics
Dependent Variable: Ln_NETRENT/SQFTa.