The Determination of Rent in Shopping Centers: Some Evidence from Hong Kong

Journal of Real Estate Literature, 7: 183±196 (1999)

# 1999 American Real Estate Society

The Determination of Rent in Shopping Centers:Some Evidence from Hong Kong

RICHARD S. TAY

Department of Economics and Marketing, Lincoln University, Canterbury, New Zealand,e-mail: [email protected]

CLEMENT K. LAU

Land Development, SAR Government, Hong Kong

MARIE S. LEUNG

Kintel International, Hong Kong

Abstract

The purpose of this article is to assess the generality of previous empirical ®ndings on the determinants of retail

rent in shopping centers and validate their substantive robustness using data from Hong Kong. Consistent with

previous ®ndings, the rental rate of a retail unit is positively related to its customer-generating power and the size

of the shopping center but negatively related to its own size. In contrast to previous ®ndings, it is positively

associated with chain stores and the age of the shopping center but not signi®cantly related to several provisions in

the lease.

Introduction

The robust economic performance of Hong Kong over the last decade has increased the

population's standard of living. As a result, the retail sales in Hong Kong have witnessed a

strong growth, with an annual compounded growth rate of 12.79 percent in the past ®ve

years, reaching a total of U.S. $27 billion in 1995.1 Furthermore, this growth in the retail

sector is accompanied by a shift in the shopping habits of the people from the traditional

department stores to shopping centers. After the completion of the ®rst shopping center in

Hong Kong, Swire House, in 1960, shopping centers of various sizes were built across the

territory. By 1995, the total gross ¯oor area of shopping centers in Hong Kong, with sizes

greater than 100,000 sq. ft., amounted to about 26 million sq. ft., with another 11 million

sq. ft. scheduled to be completed between 1996 and 2000.2

Despite its increasing importance, relatively little research has been conducted to

analyze the rental market for shopping center space in Hong Kong. Empirical studies on

the property market in Hong Kong have mainly focused on the of®ce and residential

sectors (Cheung, Tsang, and Mak, 1995; Mok, Chen, and Cho, 1995). In a review of the

literature, Eppli and Benjamin (1994) concluded that empirical literature on leasing in

general was scant, probably because data on leases were not easily obtainable.

Furthermore, most of the empirical literature on shopping center leases was on the

United States market.

Several stylized facts, however, have emerged in the literature. First, retail space in

larger shopping centers commands higher rent due to its higher customer-generating

power (Benjamin, Boyle, and Sirmans 1992; Sirmans and Gauidry, 1993; Gatzlaff,

Sirmans, and Diskin, 1994). Second, the age of a shopping center is inversely related to the

rent charged because older centers suffer physical neglect, inappropriate tenant mix, and

older facilities (Sirmans and Gauidry, 1993; Gatzlaff, Sirmans, and Diskin, 1994). Third,

owing to economies of scale in leasing, the rent charged is inversely related to the size of

the retail space (Benjamin, Boyle, and Sirmans, 1990, 1992). Fourth, tenants with greater

ability to generate customer traf®c (often referred to as anchor tenants), are charged lower

rent because of their positive externalities on overall sales in the shopping center (Gatzlaff,

Sirmans, and Diskin, 1994). Fifth, chain stores are charged lower rent since they have

lower probability of default (Benjamin, Boyle, and Sirmans, 1990, 1992). Lastly, rent

liability is also dependent on lease provisions such as the lease term, which is found to

have a negative impact on rent (Benjamin, Boyle, and Sirmans, 1990, 1992).

The purpose of this article is to assess the generality of these stylized facts in the context

of a different business environment and setting, and validate its substantive robustness.

More important we attempt to provide additional explanations for the factors that are

hypothesized to have an impact on rent by adopting a bilateral bargaining approach to rent

determination and relaxing the assumption of homogeneous retail space. Some of the

hypotheses developed from our model are consistent with the stylized facts, while others

are in contrast. We then estimate a model of rent determination for shopping centers in the

Central District of Hong Kong and ®nd that most of our hypotheses are supported by the

data.

Retail real estate market in Hong Kong

The retail real estate market in Hong Kong possesses several characteristics that are

different from the market in the United States. Most shopping centers in the United States

are single-¯oor buildings located in suburban areas and are very car-oriented. The tenant

mix in the shopping centers is usually very diversi®ed with respect to both the size of retail

space and the type of businesses. Most suburban shopping centers are fairly standard in

design and within the shopping center, retail space are relatively homogeneous. Many

empirical research on retail lease valuation assume that retail space is homogeneous (e.g.

Benjamin, Boyle, and Sirmans, 1990, 1992).

In contrast, all shopping centers in Hong Kong are located in urban areas (Hong Kong is

almost 100 percent urbanized), and most of them occupy the ®rst few levels of a high-rise

building in which the upper levels consist of either of®ce space or residential units.3 Many

of these building are interconnected by subways and overhead pedestrian walkways (see

Figures 1 and 2).4 Within the shopping center, retail space is often quite heterogeneous

with regard to pedestrian traf®c and accessibility. Therefore, the assumption of

homogeneous retail space is rather restrictive.

Also, most shopping centers in Hong Kong have very limited car-parking facilities that

are usually reserved for season ticket holders. As a result, parking charges are very high

compared to the United States, where parking is often free. This parking restriction is part

of the government's plan to discourage driving in Hong Kong. As a result, most shopping

centers are served by public buses and located within walking distance from a public

transit station (see Figure 2).5

Another interesting feature of the retail real estate market in Hong Kong is that anchor

tenants of the size found in the United States are very rare because land is extremely

scarce.6 Instead, restaurants are often considered as ``magnet shops'' because they are able

to draw customers into the shopping centers.7 Unlike the typical anchor tenant in the

United States, however, restaurants in Hong Kong often occupy relatively larger areas that

are located in less prominent positions.8

In addition, besides the normal shopping centers with a diverse mix of tenants, there

exist many shopping centers that are specialized and cater almost exclusively to a

particular retail sector. For example, Golden Shopping Center consists mainly of small

retail stores dealing in personal computer±related goods and services. The optimal mix of

the center appears to be homogeneous with respect to both the size of retail space and the

type of business.9 The existence of such centers casts some doubts, at least in the empirical

184 RICHARD TAY, CLEMENT LAU, AND MARIE LEUNG

Figure 1. Shopping center in Hong Kong.

Figure 2. Location of shopping centers and transit stations.

DETERMINATION OF RENT IN HONG KONG SHOPPING CENTERS 185

literature, on the concept of optimal tenant mix and the rationale for having a diverse mix.

This observation, however, provides some support for hypothesis, developed by the De

Palma, Ginsburg, Papageorgiou, and Thisse (1985), that positive externality is not

restricted to anchor tenants but can be generated by small homogeneous retailers as well.

A simple model of rent determination

In economics, the price of a product, in general, will depend on the market demand and

supply. However, in retail real estate, supply is ®xed in the short run, and the equilibrium

price will ultimately depend on demand and the market structure in which ®rms operate.10

Without loss of generality, consider a landlord with a ®xed amount Q of retail space to let

and facing two tenants A and B with demands shown in Figure 3. Consistent with the law

of demand, the higher the rent per unit, the lower will be the quantity of retail space

demanded by a tenant.11 If the retail space market is competitive, then the market price

will be set at P* and aQ amount of the space will be leased to A with the remaining space,

�1ÿ a�Q, leased to B.12

Benjamin, Boyle, and Sirmans (1992), however, argue that price discrimination is often

encountered in the setting of rent in shopping centers because concurrent selling markets

frequently do not exist and releasing is often prohibited so that lessees are effectively

segmented. The authors therefore developed and estimated a simple price-discrimination

model. More recently, Brueckner (1993) analyzed the retail market assuming that the

landlord has the ability to practice ®rst-degree or perfect price discrimination.

Because most leases are entered through a bilateral bargaining process, self-selection by

retailers is not a problem, and the simple price discrimination model may be overly

restrictive. On the other hand, perfect price discrimination is very rare in practice because

retailers have varying degrees of bargaining power. The price a consumer pays will

therefore depend mainly on his demand and the ability of the landlord to extract as much of

the consumer surplus from the retailers as possible. The bargaining power of the landlord,

relative to the tenant, will depend, among other things, on the ease of ®nding other tenants

who are interested in renting the retail space and the dif®culty of tenants in ®nding an

alternative site.

If retailers have little or no bargaining power, then the landlord will still lease aQ to Aand �1ÿ a�Q to B but charge the tenant A an amount equal to

RPa�Qa�dQa, which is the

maximum amount A is willing to pay to have the entire space and charge tenant B a similar

Figure 3. Rent determination.


amount equal toR

Pb�Qb�dQb. This pricing scheme is similar to the perfect price

discrimination discussed by Brueckner (1993) and is depicted in Figure 3 by the areas

given by GHIJ for tenant A and HIKL for tenant B.13 It is clear that, all else held constant,

since A has a higher demand than B, A will pay a higher rent per square foot than B.

The above scenario assumes that both tenants A and B have no bargaining power.

Suppose A, the ``large'' tenant, has some bargaining power because of the limited number

of large buyers in the market.14 Tenant A knows that if the landlord does not enter into an

agreement with A, the landlord has to deal with smaller tenants like B who have a lower

demand and thus a lower willingness to pay. Tenant A can then exercise his bargaining

power to try to minimize the consumer surplus the landlord can extract. For simplicity,

assume that the landlord's alternative is rent aQ to two smaller retailers, C and D, that are

similar to B. As shown in Figure 4, the maximum rent that the landlord can charge is then

given by the area JXWVI. Tenant Awill then try to reduce the rent to as close to this amount

as possible.15 The equilibrium rent charge will depend on the relative bargaining skills of

the tenant and the landlord, but it will be between this amount andR

Pa�Qa�dQa in

Figure 3.

Another assumption implicit in the model is that there is no demand externality between

tenant A and B. Suppose A is an anchor tenant who is able to generate some demand

externality on B. If the landlord cannot enter into an agreement with A, then he has to lease

his space to smaller tenants like C and D, whose demands, like B's, may also be adversely

affected by the absence of A. This can be illustrated in Figure 4 by the demand curves D0b,

D0c, and D0d. The maximum rent that the landlord can charge will then be reduced to

JX0W0V0I, and the landlord will also lose HH0L0L from tenant B. This gives the large tenant,

who already has some market power, more leverage to bargain with the landlord. The

equilibrium rent will then be between the maximum amount A is willing to pay (GHIJ) and

the minimum amount the landlord is willing to accept (JX0W0V0I minus HH0L0L).

Also implicit in the above model is the assumption that the retail space is homogeneous.

The model, however, can easily be extended to incorporate differentiated space. Harvey

(1987) discusses the choice of location by function (business type of tenants):16 ``The

prime sites are occupied by department stores, multiple chain stores, and, because most

Figure 4. Rent without anchor tenant.


shoppers are women (who take note of window displays), by shops selling women's

shopping goods.'' In addition, Harvey noted that `àccessibility and type of shops are also

correlated to sales turnover, shop rents, and. . . department stores and other key traders

occupy the most accessible and valuable locations for they enjoy a high sales turnover and

intensity of shopping.'' The most straightforward way to account for the differentiated

space is to consider two types of space ( prime space and less valuable space) and apply the

above analysis to each market. It should be noted that all retailers would prefer prime

space to nonprime space but they have different willingness to pay (substitution effects).

What determines who gets the prime space, and the size of its premium, is the relative

willingness of different businesses to pay for it.

Discussion of the stylized facts

The main hypotheses or stylized facts examined in this article are summarized in Table 1.

The ®rst two columns list the stylized facts or previous results and the explanation or

reasons given by the corresponding authors to support their hypotheses. The last two

columns present our hypotheses and the additional reasons we provide to support our

hypotheses.

First, Sirmans and Guidry (1993) and Gatzlaff, Sirmans, and Diskin (1994) argue that

large shopping centers have a more diverse portfolio and higher spatial concentration of

tenants, which attract proportionately broader customer bases and more comparison

shoppers. These arguments can be translated, in our model, into a higher demand curve for

retail space in larger shopping centers because of the higher marginal product of these

spaces, which ceteris paribus, will result in higher rent. Our model, however, presents an

additional argument for this positive relationship. Since landlords of larger shopping

Table 1. Summary of the stylized facts

Previous Results Previous Reasons Our Hypotheses Additional Reasons

Retail space in larger centers

commands higher rent.

Higher customer-generating

power

Same Larger centers have more market power

and thus greater ability to extract

consumer surplus.

Lower rent is charged in

older centers.

Physical neglect, older facilities,

and inappropriate tenant mix

Opposite Centers are regularly renovated to update

facilities. Tenant mix can be changed with

new leases instead of renewals. Older

centers are more established.

Lower rent is charged for

larger space.

Economies of scale in leasing Same An inverse relationship is a direct result

of the law of demand. Larger tenants also

have more bargaining power.

Anchor tenants are charged

lower rent.

Positive externality on overall

sales

No anchor tenants

similar to the

United States.

Anchor tenants have higher demand,

especially for prime space, and may be

charged a higher rent.

Same Besides creating positive exernality,

magnet shops like restaurants are charged

lower rent for taking up nonprime land.

Chain stores are charged

lower rent.

Lower probability of default Opposite Lower probability of default implies

higher willingness to pay. Also chain

stores have higher demand for prime space.

Rent liability is dependent

on lease provisions such as

duration of lease, review of

lease, rent-free period,

percentage of turnover

rent/base rent, and number

of leases held by tenant.

Transaction cost, information

cost, default risk, etc.

Minimal in¯uence Short duration of leases and information

on business type of tenant greatly reduce

these uncertainties, costs, and discounts.


centers control a larger share of the market, they have more market power and therefore

are better able to extract more consumer surplus from the retailers.17 Even when demand is

held constant, landlords of larger shopping centers will still be able to charge relatively

higher rent because they possess more bargaining power than landlords of smaller

shopping centers.18 These effects are captured in the model by LRSIZE, which is the

logarithm of the total retail space in the shopping center and is expected to have a positive

in¯uence on rent.

Second, Sirmans and Guidry (1993) and Gatzlaff, Sirmans, and Diskin (1994) claim that

the age of a shopping center is inversely related to the rent charged because older centers

suffer physical neglect, inappropriate tenant mix, and older facilities. In Hong Kong and

other modern Asian cities like Singapore, however, extensive renovations are carried out

regularly in the shopping centers, especially in the prime shopping districts. This regular

upgrading would negate the adverse impact of age on the desirability of the shopping

centers. Furthermore, since the average duration of a lease is relatively short, tenant mix

can be changed quite easily. On the other hand, older shopping centers are more

established with longer history and tradition, and are more prestigous. Although the

prestige factor cannot be overemphasized in the Hong Kong market, it is also possible that

older centers cater more to the tourist market in terms of pricing, location, and product

range. Therefore, in contrast to previous studies, we expect a positive in¯uence by AGE.

Third, Benjamin, Boyle, and Sirmans (1992) argues that the cost of negotiating and

servicing lease contracts per dollar of rent probably decreases as the square footage of a

unit increases. Although plausible in many cases, this argument may not hold in others.

Since the transaction cost per lease is likely to be independent of the size of retail space,

the justi®cation for economies of scale is more likely to arise from having to negotiate a

smaller number of leases if each retailer takes up a larger retail unit. This savings in cost,

however, may be offset by a larger loss due to higher economic rent forgone. Figure 3

provides a simple counterexample.19 Suppose AC1 and AC2 represent the average costs of

negotiating one and two leases respectively, then the cost savings from leasing the entire Qto A is given by the area between AC2 and AC1 which is smaller than the economic rent

(LHN) forgone. Even though we agree with the negative correlation hypothesized and that

economies of scale in lease negotiation may be important in some cases, we assert,

however, that this negative correlation is a direct result of the Law of Demand.

Furthermore, as discussed earlier, larger tenants also have more bargaining power which

they can exercise to reduce the economic rent extracted by the landlord. In tandem with the

size effect, the greater the number of retail units incorporated in a lease, the greater the

negotiating power a tenant has in concluding a deal with the landlord.

Fourth, Benjamin, Boyle, and Sirmans (1992) and Gatzlaff, Sirmans, and Diskin (1994)

assert that a tenant with high anticipated sales, usually known as anchor tenant in the

literature, is charged less rent for his positive externality on overall sales in the shopping

center. Our model does provide some support for this hypothesis since the anchor tenant is

usually a large tenant with some market power that it can use to minimize the rent that the

landlord can extract, and this bargaining power is also reinforced by the positive

externality it generates on other stores. This hypothesis, however, has to be quali®ed. As

discussed earlier, a tenant with higher anticipated sales will, ceteris paribus, imply that he

has a higher willingness to pay for the retail space and may be charged a higher rent by a

®rst-degree price discriminating landlord. In addition, as suggested by Harvey (1987), as a

key trader, the anchor tenant usually has a relative higher demand for prime space, which

will also translate into a higher rent compared to businesses that have a relatively lower

demand for prime space.

As discussed earlier, anchor tenants of the size found in the United States are very rare

in Hong Kong. Instead, restaurants are often considered as ``magnet shops'' because they

are able to draw customers into the shopping centers. Unlike the typical anchor tenant in


the United States, however, restaurants in Hong Kong often occupy areas that are located

in less prominent positions. This anomaly arises because, due to their customer drawing

power, restaurants do not need to be located in the more prominent and accessible areasÐ

that is, the marginal product of ``premium'' space is not signi®cantly higher than the ``less

desirable'' space. Given his ability to generate positive externality and his low demand for

prime space, a restuarant is therefore expected to pay less for his retail space.

Fifth, Benjamin, Boyle, and Sirmans (1990, 1992) conclude that chain stores are

charged lower rent since they have lower probability of default. Although plausible, this

argument has to be weighed against in¯uences similar to the anchor tenants. Since chain

stores have a lower probability of default, they will also have higher expected pro®ts that

will, ceteris paribus, provide more economic rents that can be paid to acquire the space.20

The abundance of chain stores in Hong Kong (70 percent in our sample) imply that their

bargaining power may be limited.21 In addition, chain stores are willing to pay higher

premiums to secure the more prominently and conveniently located retail units within the

shopping centers because, relative to other tenants, their businesses depend more on the

status and prestige of these locations. Therefore, in contrast to Benjamin, Boyle, and

Sirmans (1990, 1992), we posit that chain stores will be charged a higher premium than

their counterparts.

Last, ®ve other variables were used to capture the effects on rental rates of the different

terms of lease and management expertise. The dummy variable RENEW was used to

re¯ect any systematic differences between the rental rates charged to new tenants and

existing tenants at lease renewal. Fisher and Lentz (1990) argue that the difference in rent

found between these two parties re¯ects payments for business enterprise value that is

extracted from the tenants by the mall owner. The premiums in Hong Kong, however, are

more likely to be extracted from the tenants' savings in search and other transaction costs

associated with relocation such as removal and renovation costs.

The effect on rent of service charge levied on tenants by the management, however, is

ambiguous. Although service charges add to the total occupation cost, an increase in

service charges, is usually associated with better maintenance, management, and

promotion of the shopping center, which increase the demand for its retail space.

Although the term of lease may, in theory, be an important factor in determining retail

rent, Benjamin, Boyle, and Sirmans (1992), however, found only an insigni®cant negative

in¯uence. Since the average term of lease in our sample is shorter than that in Benjamin,

Boyle, and Sirmans (1992), we also do not expect it to be an signi®cant in¯uence.

Besides having a direct negative impact on the average rent, the number of rent-free

periods incorporated in the lease will also provide some indication of the bargaining power

of the tenant. Therefore, we expect it to have a negative in¯uence. Last, to examine if

tenants with turnover rent provision ultimately pay a higher or lower average rent than

those without, we include a dummy variable TORENT to capture any systematic difference

in rent between the two groups of tenants.

In addition to the six main in¯uences summarized in Table 1, we also incorporated

several other determinants. Six physical variables were used to differentiate the individual

retail units by their perceived desirability due to differences in location and their

associated levels of pedestrian ¯ow.22 Shops located on the ground ¯oor (GFLOOR) or

¯oors with either connecting footbridges or subways (FBSUB) to other buildings are

expected to command higher rents. In contrast, shops located on ¯oors further away from

the ground level (ALEVEL) are expected to command lower rents. FRONTAGE and

RETFRONT are the frontage and return frontage of the retail units and are expected to have

a positive in¯uence on rent because they provide greater area for displaying the retailers'

merchandise (Rhees, 1992). PROXACC is a dummy variable to indicate if the retail unit is

located next to an access point such as an entry or exit of the shopping center.

The correlation between the rental rates of retail spaces and the accessibility of


shopping centers, particularly by public transport, has received relatively little attention in

the literature. Accessibility by public transport, however, may be a major consideration for

shoppers in such highly urban and congested areas as Hong Kong. Shopping centers with

direct access to the Mass Transit Railway (MTR) and taxi stands (TAXIST) are expected to

be more attractive and thus will be able to command higher rents. In addition, by providing

more parking spaces for private cars in the shopping centers, even though most of them are

reserved for season-ticket holders, their accessibility by private vehicular travel may also

be increased. CPARK, which measures the number of parking spaces in a shopping center,

is thus expected to have a positive effect on LRENT.

The differential demands by tenants in different retail sectors have also received

relatively little attention in the literature. We hypothesize that stores dealing in luxury

goods (LUX) and fashion-related goods (CLOTH), as well as banks and ®nancial

institutions (BANK), are willing and able to pay higher rents for two reasons. First, these

stores may have higher demand and willingness to pay for any retail space because they

are more likely to have higher pro®ts (total revenue ÿ nonland cost). Second, they also

have a higher demand for prime space because prominent and convenient locations in the

shopping center are valued more by these tenants, as store image is relatively more crucial

to their businesses.

Finally, to control for the ¯uctuations in the market rental rates over the period, the

Jones Lang Wootton Retail Property Rental Index for the Central District (RINDEX) is

incorporated into the model.23

Data and statistical model

Our sample consisted of 405 retail shops in nine prime shopping centers located in the

central business district of Hong Kong, whose leases were entered between 1991 and

1994.24 Approximately 60 percent were renewal, and the mean rent during that period was

U.S. $19.55 per sq. ft. per month. The average tenancy ran for about thirty-four months,

with a rent-free period of 0.35 month. Only 8 percent of the leases are subjected to

turnover rent provision, and among those with such a provision, the average rate is 5.7

percent.

An average shopping center in the sample was twenty-one years old and had a retail

space of about 140,000 sq. ft. An average retail shop, however, had a lettable area of about

1,364 sq. ft. with a frontage of 27 ft. A quarter of the shops were located on the ground

¯oor, and over half of the units were located on ¯oors with either connecting footbridges

or subways.

Most of the tenants were private companies (84 percent), based in Hong Kong (89

percent) and more than half (70 percent) were chain operations. The most common types

of business were stores selling clothing, shoes, and other fashion-related products (45

percent), followed by luxury-goods stores selling antiques, jewelery, and leather goods (16

percent).

This database will be used to estimate an ordinary least squares regression model.

Table 2 provides a summary and description of the variables used in this analysis. Since

some of the leases in the sample have turnover or average rent provisions while others do

not, the ``total'' rent is used as the dependent variable. Consistent with Gatzlaff, Sirmans,

and Diskin (1994), Sirmans and Guidry (1993), and Benjamin, Boyle, and Sirmans (1990,

1992), LRENT is de®ned as the logarithm of the average monthly rent for the retail shop,

measured in dollars per square foot ($psf ), over the duration of the lease term. For leases

that are subject to turnover rent provisions, LRENT is the logarithm of the average rental

rates paid by the tenants in 1994.


Empirical results

The estimation results are presented in Table 3. Overall, the model ®tted the data quite

well, with an R-square of 0.7845 and an F-statistic that strongly rejected the null

hypothesis of no explanatory power. In addition, all explanatory variables that are

signi®cant, at either the 1 percent, 5 percent, or 10 percent levels, have their expected

signs. Furthermore, 94.8 percent of the predicted rents fell within + 15 percent of the

actual values, and 79.8 percent were within + 5 percent of the actual rents.25

Consistent with previous ®ndings, the rental rate of a retail unit is positively related to

the size of the shopping center but is negatively related to its own size. Also consistent is

the ®nding that stores with customer generating power pay lower premiums than their

counterparts. However, in contrast to previous ®ndings, the age of the shopping center and

the term of lease were found to be positive, but the latter was insigni®cant in determining

retail rent. In addition, contrary to Benjamin, Boyle, and Sirmans (1990, 1992) chain

stores in Hong Kong pay higher premiums than single stores.

Table 2. Summary and description of variables

Variable De®nition Minimum Maximum Mean

Standard

Deviation

Dependent

LRENT Logarithm of monthly rent in HK$ per sq. ft. 2.773 6.215 4.80 0.70

Independent

LRSIZE Logarithm of the size of the shopping center

in sq. ft.

8.655 12.268 11.69 0.65

AGE Age of shopping center in years 6 33 21.21 8.99

LSIZE Logarithm of the size of the retail shop in sq. ft. 4.205 9.558 6.68 0.95

ALEVEL Number of levels away from the ground ¯oor 0 4 1.29 1.11

FBSUB Footbridge or subway on same ¯oor� 1,

otherwise� 0

0 1 0.54 0.50

GFLOOR Shop on ground ¯oor� 1, otherwise� 0 0 1 0.25 0.43

FRONT Frontage of retail shop in ft. 0 160 27.47 21.08

RETFRONT Return frontage of retail shop in ft. 0 109 7.73 14.52

PROXACC Shop located next to access point � 1,

otherwise� 0

0 1 0.50 0.50

CPARK Number of car parking spaces in the shopping

center

0 177 44.58 55.47

MTR MTR entrance at shopping center� 1, otherwise� 0 0 1 0.88 0.33

TAXIST Taxi stand in the shopping center� 1, otherwise� 0 0 1 0.12 0.32

BANK Banks and ®nancial institutions� 1, otherwise� 0 0 1 0.03 0.18

CLOTH Boutiques and sports stores� 1, otherwise� 0 0 1 0.45 0.50

LUX Antiques, jewelery, and leather goods� 1,

otherwise� 0

0 1 0.16 0.36

REST Restaurants� 1, otherwise� 0 0 1 0.07 0.26

NCHAIN National chain stores� 1, otherwise� 0 0 1 0.51 0.50

ICHAIN International chain stores� 1, otherwise� 0 0 1 0.18 0.39

RENEW Renewal of existing lease� 1, otherwise� 0 0 1 0.59 0.49

TERM Term of lease in months 2 49 34.27 6.29

TORENT Lease with turnover rent� 1, otherwise� 0 0 1 0.08 0.27

RENTFREE Rent-free period in months 0 5 0.35 0.65

UNIT Number of retail shops incorporated in a single

lease

1 4 1.13 0.50

SERVC Monthly service charge in HK$ per sq. ft. 0 7.2 5.86 1.11

RINDEX Jones Lang Wootton Central Retail Property

Rental Index

268 345 310.25 23.73


As expected, retail space in shopping centers that are more accessible by transport and

shops with better locations within the shopping centers commanded higher premiums. The

latter results are inconsistent with the assertions made by Benjamin, Boyle, and Sirmans

(1992) that shopping spaces are mostly homogeneous in that they are typically

subdivisions of a larger structure. In particular, retail units located on the ground ¯oor

or ¯oors connected by either footbridges or subways command higher rents, whereas units

located on ¯oors farther away from the ground level experience discounts. Also, units with

larger return frontage command a higher premium because they are generally located at

the intersection of two passageways that increase the customer traf®c. On the other hand,

although frontage is positively related to rent, its effect is not signi®cant. Even though

larger frontage is desired by many retailers, its impact on rent may be negated by units

with very large frontage because these units are associated with large retail units and, as

discussed earlier, these units are usually situated at less desirable locations. The value of

the space within a shopping center is also dependent on the types of tenants bidding for the

space, with ®nancial institutions and retailers of luxury goods and fashion-related wares

paying higher premiums.

Finally, many of the lease provisions were found to have little in¯uence on the retail

rent. Since the average term of lease was relatively short, it was not surprising that the

duration of the lease and whether the lease was a renewal or new lease were found to be

insigni®cant determinants. The number of rent-free periods and turnover rent provisions

were also found to have minimal impact on the average rent. This implied that landlords

may have adequately incorporated their effects into the expected rents.

Table 3. Estimation results

Variable Coef®cients Standard Error T-Statistics Prob4 |T|

INTERCEP 0.4079 0.7569 0.539 0.5902

LRSIZE 0.1518 0.0515 2.948 0.0034

AGE 0.0226 0.0068 3.254 0.0012

LSIZE ÿ 0.1663 0.0270 ÿ 6.157 0.0001

GFLOOR 0.8223 0.0823 9.991 0.0001

ALEVEL ÿ 0.0687 0.0265 ÿ 2.598 0.0098

FRONT 0.0002 0.0010 0.181 0.8567

RETFRONT 0.0021 0.0013 1.696 0.0908

PROXACC 0.0370 0.0397 0.934 0.3511

FBSUB 0.3756 0.0538 6.984 0.0001

MTR 0.5585 0.1085 5.150 0.0001

CPARK 0.0044 0.0009 4.586 0.0001

TAXIST ÿ 0.0652 0.1147 ÿ 0.568 0.5701

BANK 0.2676 0.1108 2.416 0.0162

CLOTH 0.1426 0.0481 2.970 0.0032

LUX 0.1655 0.0574 2.882 0.0042

REST ÿ 0.2645 0.1069 ÿ 2.475 0.0137

NCHAIN 0.1611 0.0489 3.296 0.0011

ICHAIN 0.2554 0.0624 4.095 0.0001

RENEW 0.0285 0.0418 0.682 0.4957

TERM 0.0043 0.0029 1.501 0.1342

RENTFREE ÿ 0.0454 0.0317 ÿ 1.433 0.1526

TORENT ÿ 0.0888 0.9561 ÿ 0.929 0.3536

UNIT ÿ 0.0954 0.0382 ÿ 2.500 0.0128

SERVC 0.2031 0.0436 4.654 0.0001

RINDEX 0.0025 0.0007 3.392 0.0008

Note: Number of observations is 405, R-square is 0.7845, F-statistics is 55.18, Prob4 F is 0.0001.


Concluding remarks

Despite the increasing importance of shopping centers, empirical analysis on the

determinants of shopping center rents has been rather scarce, especially from outside the

United States. This article estimated a model of shopping center rents using a sample of

data from Hong Kong and found that the results obtained were mixed relative to those

found in the United States.

Consistent with previous ®ndings, the rental rate of a retail unit is positively related to

the size of the shopping center but is negatively related to its own size, and stores with

customer generating power pay lower premiums than their counterparts. However, in

contrast to previous ®ndings, the age of the shopping center was found to be positively

related to the retail rent, several provisions in the lease were found to be insigni®cant in

determining retail rent, and chain stores in Hong Kong paid higher premiums than single

stores.

Our results also showed that tenants, particularly those who were more concerned about

the location of stores within the shopping centers, were willing to pay higher premiums for

the units that were more prominently and conveniently located. Speci®cally, retail units

located on the ground ¯oor or ¯oors with connecting footbridges or subways, and units

with larger return frontage were found to command higher rents. Furthermore, ®nancial

institutions, fashion boutiques, and luxury-goods dealers were found to pay relatively

higher rents.

Landlords in Hong Kong, therefore, should continue to renovate their shopping centers,

differentiate the retail space available by their desirability and segment the market

appropriately to better match the corresponding demands so as to maximize rents. Retail

units with larger return frontage or located on either the ground ¯oor or ¯oors that are

connected by subways or footbridges to other buildings should be leased at a relatively

higher rate to chain stores, ®nancial institutions, fashion boutiques, and luxury goods

dealers.

The mixed results, relatively to previous research, found in our study were not

surprising because the landlord is faced with many concerns when selecting his tenants

and the price to charge each tenant. Many of these concerns are interrelated, either

positively or negatively, and depending on their relative impacts, may produce different

results depending on the business environment and setting. Therefore, more empirical

research, using data from diverse sources, should be conducted to provide a better

understanding on the determination of rent in shopping centers. Landlords should

therefore exercise caution in interpreting and utilizing the ®ndings from studies conducted

in a different business environment and setting.

Notes

1. Hong Kong Monthly Digest of Statistics, various issues. US$1�HK$7.78.

2. Brooke Hillier Parker Research. Places with less than 100,000 sq. ft. of retail space are too numerous and not

considered as shopping centers.

3. Most of the differences discussed in this section are also applicable to other Asian cities such as Singapore

and Kuala Lumpur, Malaysia.

4. The photograph in Figure 1 shows an example of a shopping center, in our sample, with its of®ce complex

above and pedestrian walkways connecting it to other buildings. Figure 2 shows an example of the

interconnections between buildings via subways and pedestrian walkways.

5. Figure 2 shows the locations of several shopping centers and public transit stations in the central district.

6. This is also true in other highly urbanized city states such as Singapore.

7. Eating and shopping are considered the favorite national pasttimes of people in both Hong Kong and

Singapore. Relative to other places, customers in Hong Kong and Singapore will go to great lengths to seek

out good eating places.


8. Besides being located in less desirable ¯oor levels, restaurants are usually located at more remote areas where

there is ``lower'' pedestrian traf®c and the corridors can be used as waiting areas for customers.

9. This difference is not unique to Hong Kong alone but is common in many Asian cities. For examples, Funan

Center in Singapore and IMBI plaza in Kuala Lumpur, Malaysia, also specialize in offering small personal

computer stores. Unfortunately, none of the shopping centers in our sample is of the specialized type.

10. Consistent with most studies of rent determination, we assume that the decision to build has already been

made and landlords are simply deciding on the rents to charge subject to their supply of retail space. Our

assumption that supply is ®xed in the short run is in contrast to Brueckner (1993) and Benjamin, Boyle, and

Sirmans (1992), which did not impose a constraint on total quantity even though they were analyzing only

space allocation in shopping centers and not the decision to build.

11. Although the law of demand is well accepted in economics as the basic explanation for the inverse

relationship between price and quantity, it has often been ignored in the real estate literature, which usually

attributes the inverse relationship to economies of scale in leasing contracts (Benjamin, Boyle, and Sirmans

1990, 1992).

12. For simplicity, we have ignored marginal cost. When marginal cost is not negligible, then pro®t is given by

the area between the demand and marginal cost curves.

13. For the linear demand shown, the per unit rent are given by the demand curves at aQ/2 for A and �1ÿ a�Q=2

for B.

14. A ``large'' tenant in our model refers to one with higher demand for space, which includes a tenant who

occupies a large retail unit and also a tenant who has a higher willingness to pay for a smaller space.

15. It should be pointed out that although the landlord's reservation price is lower, it does not imply that he will

automatically accept a lower price if the tenant has no bargaining power; this action is contrary to the notion

of pro®t maximization. If producers always accept the reservation price, there will be no producer surplus in

any market.

16. It should be noted that Harvey's discussion refers to location choice within the central business district and

not the shopping mall. His arguments, however, are applicable to location choice within a shopping center.

17. It is easier for retailers to ®nd an alternative site if his landlord controls only a small portion of the market

supply but ®nding an alternative supplier may be fairly dif®cult if his landlord controls a larger share of the

market supply.

18. Empirically, we are not able to hold demand constant but are able to account for several factors, such as

location within the center and the business types of the retailers, that are likely to shift the demand curve.

19. For simplicity, the transaction cost is assumed to be distributed equally or converted into a constant average

cost. A decreasing average cost curve, however, does not affect the results.

20. Consider the simple case where a tenant defaults if it suffers an economic loss but will not default if it makes

an economic pro®t. A lower probability of defaults implies a lower probability of making a loss and a higher

probability of making a pro®t, which should, ceteris paribus, result in a higher expected pro®t.

21. Relative to the average American, a typical Hong Kong shopper is very ``label conscious,'' and the premiums

on products sold by international chains is much higher.

22. A very important aspect of the retail units is their Feng-Shui or geomancy, for which data are not available.

23. This index is the most widely used index in the retail property market in Hong Kong. Consistent with Sirmans

and Guidry (1993), the index was used as an independent variable rather than a conversion factor for the rent.

Subsequent estimates using log(rent/rindex) as the dependent variable yielded results that are qualitatively

and quantitatively very similar.

24. Since all the shopping centers in the sample are in the central business district, this would effectively control

for the effect of distance from the CBD on rents. The authors would like to thank the management of Hong

Kong Land Holdings Ltd. for allowing us to use the data from ®ve of its subsidiaries in the real estate market.

25. According to D.R. Stevenson, rating advisor, Rating and Valuation Department of Hong Kong, an estimate

that falls within + 15 percent of the actual value is considered as a very good estimate. In the 1994 to 1995

General Revaluation exercise conducted by the department, about 50 percent of the estimates produced by the

departments' computer models were revaluated manually.

References

Benjamin, J.D., G.W. Boyle, and C.F. Sirmans. (1990). ``Retail Leasing: The Determinants of Shopping Center

Rents.'' AREUEA Journal 18(3), 302±312.

Benjamin, J.D., G.W. Boyle, and C.F. Sirmans. (1992). ``Price Discrimination in Shopping Center Leases.''

Journal of Urban Economics 32, 299±317.

Brooke Hillier Parker Research. (1996). ``Survey of Shopping Centers in Hong Kong.'' Research report submitted

to Hong Kong Land Limited, Hong Kong.


Brueckner, J. (1993). `Ìnter-Store Externalities and Space Allocation in Shopping Centers.'' Journal of RealEstate Finance and Economics 7, 51±56.

Census and Statistics Department. (various dates). Hong Kong Monthly Digest of Statistics.

Cheung, Y., S. Tsang, and S. Mak. (1995). ``The Casual Relationship Between Residential Property Prices and

Rentals in Hong Kong: 1982±1991.'' Journal of Real Estate Finance and Economics 10, 23±35.

De Palma, A., V. Ginsburg, Y. Papageorgiou, and J. Thisse. (1985). ``The Principle of Minimum Differentiation

Holds Under Suf®cient Heterogeneity.'' Econometrica 53, 767±781.

Eppli, M.J., and J.D. Benjamin. (1994). ``The Evolution of Shopping Center Research: A Review and Analysis.''

Journal of Real Estate Research 9(1), 5±32.

Fisher, J.D., and G.H. Lentz. (1990). ``Business Enterprise Value in Shopping Malls: An Empirical Test.''

Journal of Real Estate Research 5(1), 167±175.

Gatzlaff, D.H., G.S. Sirmans, and B.A. Diskin. (1994). ``The Effect of Anchor Tenant Loss on Shopping Center

Rents.'' Journal of Real Estate Research 9(1), 99±110.

Harvey, J. (1987). Urban Land Economics: The Economics of Real Property. London: MacMillan Education.

Jones Lang Wotton Research & Consultancy. (1996). JLW Property Index, Hong Kong: The Market as at April1996.

Mok, H., P. Chan, and Y. Cho. (1995). `À Hedonic Price Model for Private Properties in Hong Kong.'' Journal ofReal Estate Finance and Economics 10, 37±48.

Rhees, W. (1992). Valuation: Principles into Practice. London: Estates Gazette.

Sirmans, C.F., and K.A. Guidry. (1993). ``The Determinants of Shopping Center Rents.'' Journal of Real EstateResearch 8(1), 107±115.


Documents

The Determination of Rent in Shopping Centers: Some Evidence from Hong Kong