capacity scarcity restaurant revenue management.pdf
TRANSCRIPT
-
8/16/2019 Capacity scarcity Restaurant revenue management.pdf
1/11
International Journal of Hospitality Management 35 (2013) 316–326
Contents lists available at ScienceDirect
International Journal of Hospitality Management
j ournal homepage : www.elsevier .com/ locate / i jhosman
Restaurant revenue management: Do perceived capacity scarcity and
price differences matter?
Cindy Yoonjoung Heo a,∗, Seoki Lee b, Anna Mattila c, Clark Hu d
a School of Hotel andTourism Management, TheHong Kong Polytechnic University, 17 Science MuseumRoad Tsim ShaTsui East, Kowloon, Hong Kong b Schoolof HospitalityManagement, ThePennsylvaniaStateUniversity, 201Mateer Building University Park, PA 16802, United Statesc School of Hospitality Management, ThePennsylvania State University, 201Mateer Building University Park, PA 16802,United Statesd Beijing UnionUniversity, Beijing, China
a r t i c l e i n f o
Keywords:
Restaurant revenue management
Fairness perception
Perceived scarcity
Capacity limitation
Commodity theory
a b s t r a c t
Revenue management (RM) has become an indispensable strategic tool in capacity-constrained service
industries whose total revenue often depends on the abilities of firms to use capacity efficiently. The
restaurant business is similar enough to traditional RM industries such as hotels and airlines, but restau-
rants also have unique characteristics that pose special challenges to restaurant operators. Among the
unique characteristics of restaurants are the relative flexibility of service capacity and the flexible dura-
tion of a meal, which are important subjects to be considered in the implementation of RM practices. In
addition, when a restaurant operator practices a demand-based variable pricing policy to adjust demand,
the magnitude of the price differences may influence fairness perceptions of the policy. Based on the com-
modity theory and the equity theory, this study hypothesizes that two main effects, namely, perceived
scarcity of capacity in a restaurant and price differences, influence the perceived value of a restaurant’s
offerings andthe fairness perceptions of a restaurant’s RM practices. As hypothesized, the negative effects
of price difference on fairness perceptions are supported by the results. However, findings suggest that
perceived scarcity of capacity influences neither the perceived value of a restaurant’s expected offering
nor the fairness perceptions for a restaurant’s RM practices.
© 2013 Elsevier Ltd. All rights reserved.
1. Introduction
Revenue management (RM) has become an indispensable
strategictool in capacity-constrained service industrieswhose total
revenues often depends on the abilities of firms to use capacity
efficiently. As the service provider reaches capacity, limitations
restrict ability to serve additional customers. A restaurant, for
example, may have insufficient seating capacity during the peak
midday period typically for serving lunches. Indeed, most service
providers face some type of capacity constraint (Desiraju and
Shugan, 1999) and the combination of perishability and capacity
constraints encourage service firms to focus on efficiently capital-
izing on existing capacity.
The capacity of a service firm is defined as “the highest quan-
tity of output possible in a given time period with a predefined
level of staffing, facilities, and equipment” (Lovelock, 1992, p. 26).
Measurement of capacity involves both physical and non-physical
aspects. For example, physical capacity may be the number of seats
(for airlines) and the number of rooms (for hotels). Non-physical
∗ Corresponding authors. Tel.: +1 852 9735 4004.
E-mail address: [email protected](C.Y. Heo).
capacity is usually time-based and reflects the notion of physical
capacity occupied during certain periods of time, such as seating
hours in restaurants and tee-times for golf courses. The capacity of
a service firm is usually fixed over the short term, and thus the
rationale for RM is the efficient use of fixed, perishable capaci-
ties by charging customers different prices for identical services
in an attempt to balance demand and revenues per capacity unit
(Kimes,1989a; McGilland Ryzin van, 1999). RM practitioners apply
concepts, such as market segmentation, demand forecasting, and
variable pricing, to ensure maximizing the service firm’s revenue
from limited capacity by adjusting prices to thehighest possible for
any given situation (Ng, 2007).
Kimes (1989a, 1989b) identified a number of preconditions
for the successful application and effective operation of RM. In
general, to gain additional revenues and profit, RM practices suit
service industries that have perishable inventory, fixed capacity,
high-fixed/low-variable cost structures, variable demand, and seg-
mentedmarkets (Kimes, 1989a, 1989b; Berman, 2005). In addition,
reservation systems can aid management of demand forecast-
ing by calculating inventory units prior to consumption. Airline,
hotel, and rental car industries represent traditional RM indus-
tries because they share those similar characteristics (Chiang et al.,
2007).
0278-4319/$ – see front matter © 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijhm.2013.05.007
http://localhost/var/www/apps/conversion/tmp/scratch_3/dx.doi.org/10.1016/j.ijhm.2013.05.007http://localhost/var/www/apps/conversion/tmp/scratch_3/dx.doi.org/10.1016/j.ijhm.2013.05.007http://www.sciencedirect.com/science/journal/02784319http://www.elsevier.com/locate/ijhosmanmailto:[email protected]://localhost/var/www/apps/conversion/tmp/scratch_3/dx.doi.org/10.1016/j.ijhm.2013.05.007http://localhost/var/www/apps/conversion/tmp/scratch_3/dx.doi.org/10.1016/j.ijhm.2013.05.007mailto:[email protected]://crossmark.crossref.org/dialog/?doi=10.1016/j.ijhm.2013.05.007&domain=pdfhttp://www.elsevier.com/locate/ijhosmanhttp://www.sciencedirect.com/science/journal/02784319http://localhost/var/www/apps/conversion/tmp/scratch_3/dx.doi.org/10.1016/j.ijhm.2013.05.007
-
8/16/2019 Capacity scarcity Restaurant revenue management.pdf
2/11
C.Y. Heo et al. / International Journal of Hospitality Management 35 (2013) 316–326 317
The restaurant sector is sufficiently similar to hotels’ and air-
lines’ operations that RM practices are applicable for strategic
planning. However, restaurants also have unique characteristics
that pose special challenges, requiring operators to be creative in
developing appropriate RM strategies. Among the unique charac-
teristics of restaurants are the relative flexibilities of capacities
and the flexible durations of meals, and these represent impor-
tant subjects for consideration when implementing RM practices.
Unlike airlines and hotels, restaurants have somewhat more flexi-
ble capacities, for example, a restaurantmay have availableoutdoor
patios for extra seating during pleasant weather and peak patron-
age periods. Moreover, the total available seating capacity per day
in a restaurant is not fixed since customers’ seating durations are
unpredictable. Restaurant operators may need to understand how
customers perceive the capacity limitations of restaurants. This
knowledge is important because customers are mostly familiar
with RM practices in traditional RM industries (e.g., airlines or
hotels) with fixed capacities; perceptions of the relatively flexible
capacityof restaurants may influence customers’perceptions of RM
practices.
In addition, when a restaurant operator practices a demand-
basedvariable pricing policy, the magnitudeof the pricedifferences
may influence fairness perceptions of the policy. The responses of
customers to restaurants’ RM practices are critical to successful
application of RM in restaurants because revenue maximization is
only attainable when customers accept the RM practices without
dissatisfaction. Previous literature suggested that perceptions of
value(e.g., Dodds,Monroe, andGrewal, 1991;Grewal, Monroe, and
Krishnan, 1998; Monroe, 1990; Rao and Monroe, 1989) and fairness
in service exchanges (e.g., Maxwell, 2002) are important factors
for sustaining customer satisfaction, positive behavioralintentions,
and consequently, long-term profitability. To sustain customer sat-
isfaction and to maintain positive relationships with customers,
andfor successfulimplementation of RM, customers’perceptionsof
RM, such as perceptions of value andfairness, arenecessaryconsid-
erations, coinciding with the characteristics of industries (Chiang
et al., 2007: Heo and Lee, 2009).
RM has had application to the restaurant industry, but priorresearch offered a limited number of specific strategies for imple-
mentation (Kimes et al., 1998; Susskind et al., 2004). Restaurants
offering promotions, such as “happy hour” and “early bird spe-
cials,” onlyfocus on discountsduring low-demandperiods. The core
element of RM is to charge premium prices during high-demand
periods based on capacity limitation. However, the potential for
customers’ dissatisfaction have discouraged restaurant operators
from applying various types of RM approaches, including demand-
based pricing (Kimes and Wirtz, 2003a). Therefore, the current
study focuses on customers’ perceptions of a scarcity of capacity
in restaurants and customers’ varying reactions to price different-
ials in terms of the perceived value of a restaurant’s offerings and
the perceptions of RM practices’ fairness.
2. Literature review
2.1. Restaurant revenue management
Capacity utilization is a major concern for restaurants when
attempting to maximize revenue because restaurants have, by
nature, relatively limited space, relatively high fixed costs, and no
inventory opportunities. However, restaurants unconcerned with
limited space, such as takeout restaurants that market the meal
without associated space and time, may have limited interest for
applying RM to maximize revenues. Kimes (1999) argued that the
principles of RM can apply to restaurants given that the unit of
sale in restaurants is the time required for service rather than just
the meal. RM is more applicable to restaurants enjoying demand
greater than seating capacity during peak times (e.g., Friday din-
ner). These types of restaurants can increase revenue by managing
demand and controlling the customers’ seating duration.
Kimes(1999) and Kimes et al.(1998, 1999)wereamongthe first
to address the issue of restaurant RM. They developed a strategic
framework for applying RM to restaurants to increase revenue by
effectively managing seating durations and demand-based pricing.
They proposed using therevenue peravailable seat hour (RevPASH:
revenue accrued in a given time interval divided by the number
of seats available during that time). RevPASH indicates the rate at
which capacity utilization generates revenue, and it increases as
the number of tables’ turnover increases and the length of sea-
ting durations decrease. Kimes et al. (1998) suggested managing
seating duration and customers’ arrival as methods for demand
management. Seating duration (subsequently, duration of a meal)
is the length of time that a customer occupies a seat in a restau-
rant; its importance arises from governing the availability of seats
(Kimes et al., 1998). By reducing variations in meals’ durations,
restaurant operators can manage reservations and seating deci-
sions more effectively (Kimes,1999;Kimesand Chase,1998;Kimes
et al., 1998).
Basedon theargument of Kimes et al. (1998), several researchers
expanded the discussion of the relevant issues for controlling the
duration of meals. For example, Kimes and Robson (2004) discov-
ered that seating durations in midscale restaurants vary based on
table characteristics such as table type and table location. Kimes
and Thompson (2005) reported the average duration of a meal
by party size and day of week for midscale restaurants. Bell and
Pliner (2003) showed that meal durations increase with party size
among several restaurant contexts. Kimes et al. (2002) discov-
ered that Europeans preferred a significantly longer dining time,
while North Americans and Asians share similar meal duration
expectations. Noone and Kimes (2005) examined the effects of
reduced durations on the satisfaction of customers and their inten-
tions to return to the restaurant, and found duration reduction
strategies can directly and negatively influence customers’ satis-
faction. Noone et al. (2007) f ound that an exceedingly fast paceduring a meal diminishes customers’ satisfaction. More recently,
Noone et al. (2009) f ound that the overall relationship between
a perceived service encounter’s pace and satisfaction follows an
inverted U-shape. Thompson (2009) conducted a simulation-based
study andfoundthat on average, therevenue bump experienced by
decreasing dining duration is less than one-quarter of the amount
predicted by the common assumption that a reduction in dining
duration yields a proportional increase in revenue.
Some researchers are more interested in maximizing rev-
enue by increasing efficiency of restaurants’ operations. Sill and
Decker (1999) proposed the use of capacity management science
(CMS) as a systematic approach to evaluate the capacity potential
and process efficiency of a restaurant. CMS includes monitoring
every element of service and of the production delivery processwith quantifiable measurements to enhance customer satisfac-
tion, improve employee satisfaction, and increase profitability.
Quain et al. (1998) and Muller (1999) identified managerial fac-
tors that may improve the efficiency of restaurants; those factors
include defining profit centers, dispersing demand, reducing oper-
ating hours,and decreasing service time by improvingefficiency, as
much as possible, of restaurants’ operations. Similarly, Kimes et al.
(1999) suggested recommendations for the application of restau-
rant RM such as training employees,developing standard operating
procedures, and improving table management to increase the effi-
ciency of restaurant operation. Thompson (2002, 2003) f ocused on
restaurants that do not allow reservations and found that having
appropriately sizedtables in positions to allowrearrangementwith
other tables to serve larger parties can yield additional revenue at
-
8/16/2019 Capacity scarcity Restaurant revenue management.pdf
3/11
318 C.Y. Heo et al. / International Journal of Hospitality Management 35 (2013) 316–326
virtuallyno added cost.Bertsimasand Shioda (2003) developedtwo
classes of optimization models to maximize revenue for a restau-
rant. The two-step procedure proposed included using a stochastic
gradient algorithm to decide, a priori, how many future reser-
vations to accept, and using approximate dynamic programming
methods to decide, dynamically, when to seat an incoming party
during the operational day. Susskind et al. (2004) f ound that cus-
tomers are willing to shift their dining times to off-peak hours in
exchange for discounted pricing. Dickson et al. (2005) suggested
incentive strategies with a reservation system that can manage
shifting demand so that arrivals match capacities.
Other researchers focused on fairness perceptions of various RM
techniques in restaurants. For example, Kimcs and Wirtz (2002)
examined perceived fairness of five types of RM pricing in restau-
rants and found that framing demand-based pricing as discounts
rather thanas surcharges madeconsumers perceive RM practicesto
be fairer. However, only a fewstudies explored RM from customers’
perspectives. These studies focused on fairness perceptions for RM
techniques or the influence of the techniques on customers’ sat-
isfaction. Although demand-based variable pricing has proven to
be successful in other service businesses such as airlines and hotels
wherecustomersfind demand-based pricing to be moreacceptable
or fair, restaurants have greater constraints in use of demand-
based, variable pricing (Kimes and Chase, 1998). However, Kelly
et al. (1994) suggested demand-based menu pricing as a method of
managing revenue, by arguing that a 1% increase in price can yield
as much as a 20% improvement in profits. They also claimed that a
demand-based approach to menu pricing is predictable from cus-
tomers’ perceptions of a restaurant’s valuebecausecustomersfocus
more on value than on price. Kelly et al. suggested that for future
menu-pricing strategies, restaurants should consider the demand
fluctuation. Heide et al. (2008) also investigated different pricing
strategies for restaurants and concluded that a potential exists
for increased use of various revenue-enhancing strategies such as
pricediscrimination,peak-loadpricing,and bundling.However,the
depth of extant studies on RM practices applied to restaurants is
fairlyshallow (Bertsimas and Shioda, 2003). A theoretical approach
would identify the underlying factors that form perceptions of fair-ness among customers to advance restaurants’ RM practices.
2.2. Commodity theory and perceived value
Commodity theory explains how individuals respond to scarcity
by claiming thatvalues of commodities reflectthe extent of unavail-
ability (Brock, 1968). Hypothetically, a commodity represents
anything capable of being possessed, has use, and is transferable
from one person to another (Brock, 1968) and encompasses both
material goods and intangible services. Value refers to increases
in perceived utility or perceived desirability (Brock and Brannon,
1992). Unavailability refers to scarcity and other limits on avail-
ability. Typically, the concept has been interpreted as limits in
supplyor inthe numberof suppliers, costsin acquiringor providing,restrictions limiting possession, and delays in providinga commod-
ity (Lynn, 1991). In the literature, the terms “unavailability” and
“scarcity” are commonly interchangeable.
Based on the commodity theory, the influence of scarcity on
the valuation of goods has been studied extensively (Lynn, 1991,
1992; Mittone and Savadori, 2009; Verhallen, 1982; Vehallen
and Robben, 1994). Building on the commodity theory, several
researchers tested four propositions: A product or service will be
more attractive (1) when the numbers of suppliers are limited;
(2) when a supplier imposes a restriction on availability; (3)
when a consumer has to wait to gain the product, and (4) when
the consumer needs to exert extra effort to obtain the product
(Brock, 1968; Brock and Mazzocco, 2004). Researchers in psychol-
ogy and marketing found that knowledge of a product’s scarcity
affects consumers’ perceptions and evaluations of the attractive-
ness, desirability, expensiveness, quality, and tasteof a product. For
example, Verhallen (1982) f ound that people showed greater pref-
erence for a recipe book perceived to be scarce. Lynn (1989) also
found that artwork perceived as scarce was more desirable than
paintings perceived to be readily obtainable. Recently, Aggarwal
et al. (2011) tested the relative effect of two types of scarcity mes-
sages (limited-quantity and limited-time) on customers’ purchase
intentions and found that the limited-quantity messages are more
effective than limited-time messages in influencing the purchase
intentions of consumers.
Several researchers examined the influences of perceived
scarcity on customers’ perceptions of pricing. Lynn and Bogert
(1996) examined the effect of scarcity on an anticipated price
appreciation and foundthat scarcity increased the anticipated price
appreciation of collectible products. They argued that although
scarcity does not affect the actual potential of a product for price
appreciation, news reports of scarce collectibles that have appre-
ciated in value might lead people to develop naive economic
theories and associate scarcity with price appreciation. Suri et al.
(2007) examined how perceived scarcity influences consumers’
processing of pricing. During scarcity, consumers’ perceived qual-
ity and monetary sacrifice showed different patterns in responses
that depended on the relative price level and motivation of con-
sumers to process information (Suri et al., 2007). The study found
that a high price along with high-motivation, perceived quality
and value as well as purchase intentions increased during scarcity,
supporting the hypothesis that motivation to process information
moderates the effects of scarcity on information processing. The
researchers argued that an increase in perceived value of an offer
depends on whether or not application of pricing serves a greater
role to evaluate perceived sacrifice or perceived quality.
However, a few researchers argued that the appeal of scarcity
does not necessarily result in favorable perceptions for the scarce
product because potentialpurchasers scrutinizean offermore thor-
oughly(Brannon andBrock, 2001; Brock andBrannon,1992; Inman
et al., 1997). A liberalization of the commodity theory proposed by
Brock and Brannon (1992) included three modifications: (1) exten-sion ofthe domainfromany conveyable andpossibleobject totraits
and skills, (2) extension to negative objects, and (3) identification
of cognitive elaboration as a mediator between scarcity and eval-
uative polarization. Brock and Brannon (1992) argued that when
negatively valenced objects, for which an individual might have a
clear aversion, are scarce, the original notion of usefulness is dis-
carded. Moreover, a negatively valenced experience gains greater
aversive perceptions to the extent of its rarity.
Based on the commodity theory, the findings of the majority of
the research suggest that when individuals perceive a scarce prod-
uctas uniqueor valuable,scarcitywill elicitpositivefeelings forthe
product. From the positive perspective of scarcity, the commodity
theory should predict an increase in attractiveness of a restau-
rant’sscarce capacity andperceivedvalue of the dining experience.Therefore, this study proposes the hypothesis:
H1: The perceived scarcity of capacity in a restaurant will pos-
itively influence the perceived value of that restaurant’s expected
offering.
2.3. Equity theory and fairness perception
The equity theory, proposed by Adams (1965), focuses on an
individual’s perception of fairness with respect to a relationship.
Thetheory postulates that individuals consider what they putintoa
given situation relative to what they gain from it andthen compare
the evaluation with the inputs and outcomes of others. If the rela-
tionships are inferentially inequitable or unfair, individuals may
-
8/16/2019 Capacity scarcity Restaurant revenue management.pdf
4/11
-
8/16/2019 Capacity scarcity Restaurant revenue management.pdf
5/11
320 C.Y. Heo et al. / International Journal of Hospitality Management 35 (2013) 316–326
andWiner, 1995). According to the adaptation-level theory, assess-
ments of stimuli differences depend on the magnitude of the
standard against which the assessments are made, and behavioral
responses of individuals to stimuli represent modes of adapta-
tion to environmental and organismal forces (Helson, 1964). In
a pricing context, the adaptation-level theory suggests that price
perceptions rely on the actual price and on the reference price or
adaptation level of the customer. Furthermore, the adaptation level
is a function of the magnitude of a series of stimuli, the range of
stimuli, and the dispersion of stimuli from the mean.
Second, the assimilation-contrast theory (Sherif and Hovland,
1958) states the existence of regions that are internal to peo-
ple’sperceptual judgments,namely,latitudeof acceptance,latitude
of rejection, and latitude of non-commitment. A new stimulus
that falls within the latitude of acceptance may be acceptable,
and consequently assimilate, while the stimulus that falls within
the latitude of rejection is unacceptable or objectionable. In the
context of price evaluations, latitude of acceptance consists of
an acceptable price range encompassing a reference point; lat-
itude of rejection translates into an unacceptable price range,
and latitude of non-commitment constitutes a range of neither
acceptable nor unacceptable prices. Thus, the assimilation-contrast
theory suggests that the price differences falling within con-
sumers’ acceptable price ranges are either accepted or assimilated
(Blair and Landon, 1981; Liefeld and Heslop, 1985). In other
cases, when price differences fall outside the acceptable price
range, they are contrasted or rejected (Monroe and Petroshius,
1981).
When a restaurant charges different prices for the same menu
on different days of the week, consumers may perceive the qual-
ity of food and service to be the same, but the prices are different.
As a result, as the price difference increases, the perceived value
will decrease. The price difference will also have a negative influ-
ence on fairness perceptions toward RM practices in restaurants.
The assimilation-contrast theory suggests that price differences
outside the acceptable range are contrasted or rejected (Monroe
and Petroshius, 1981). Therefore, if the price during high-demand
periods fallsoutside the acceptable range, thatconditions may neg-atively affect the perception of fairness toward the restaurant’s RM
practices. Therefore, this study proposes the hypotheses:
H3: The price difference between low-demand and high-
demand periods willnegativelyinfluencethe perceived valueof the
restaurant’s expected offerings. In otherwords, the higherthe price
difference is, the lower is the perceived value of the restaurant’s
expected offerings.
H4: The price difference between low-demand and high-
demand periods will negatively influence perceptions of fairness of
the restaurant’s RM practices. In other words, the higher the price
differenceis, thelowerthe perception of fairness perceptions of the
restaurant’s RM practices.
In addition, Brock and Brannon (1992) argued that consumers
tend to assume scarce things are more expensive than availableones, and that expensive things are often assumed to be of bet-
ter quality and reflect higher status than inexpensive commodities.
Although Brock and Brannon did not empirically test the relation-
ships, the study proposed perceived expensiveness as a moderator
of scarcity effect. Therefore, the positive relationship between
perceived scarcity and perceived value of the restaurant’s expected
offerings may be stronger when price differences between low-
demand and high-demand periods are large than when price
differencesare small. In addition,the positive relationship between
perceived scarcity and perceptions of fairness of the restaurant’s
RM practices may be stronger when the price differences are large.
Therefore, the current study proposes hypotheses:
H5: The price difference between low-demand and high-
demand periods will moderate the relationship between perceived
Table 1
Experimental design.
Constructs Levels
PS (perceived scarcity) No Low High
PD (price difference) 10% 20% 30% 40%
scarcity of capacity in the restaurant and the perceived value of the
restaurant’s expected offerings.H6: The price difference between low-demand and high-
demand periods will moderate the relationship between perceived
scarcity of capacity in the restaurant and perceptions of fairness for
the restaurant’s RM practices.
The basic assumption of these hypotheses, however, is that
a restaurant in this study faces capacity limitation during high-
demand periods. Restaurants unconcerned with limited dining
space (for example, takeout restaurants that market the meal itself
rather than space and time) may not be appropriate considerations
for this study.
3. Methodology
3.1. Design of the study
This study used three (scarcity of capacity: No, Low, or High) by
four (price difference: 10%, 20%, 30%, or 40%) factorial, between-
subjects design to test the hypotheses (Table 1). More specifically,
the study adopted factorial analysis of covariance (ANCOVA) to
examine not only main effects but also moderating effects.
This study used written scenarios to manipulate the perceived
scarcity of capacityin restaurants and theprice differencesbetween
high- and low-demand periods. A sample scenario appears in
Appendix A. Fromthe literature, unavailabilityrefers to scarcityand
other limits on availability (Lynn, 1991); thus, these terms provide
the basis for the concept of perceived scarcity of a restaurant’s
capacity. The survey used in the study includes two question-
naireitems, limitedness and availability, to perform a manipulationcheck for the perceived scarcity construct.
This study developed a questionnaire for the context of dining
in a casual restaurant and uses a design with relevant constructs
based on scales adopted from previous research. Some adjust-
ments to the questionnaire were necessary to reflect the specific
characteristics of restaurants, and. employs a seven-point Likert
scale to measure each item in the questionnaire. Perceived value
represents “consumers’ overall assessment of the utility of a prod-
uct based on perceptions of what is received and what is given”
(Zeithaml, 1988, p.14). The survey includes three direct measures
to capture customers’ perceived value, adopting measures from
Cronin, Brady, and Hult (2000) with modifications for the context
of restaurants. Consistent with the studies by Kukar-Kinney, Xia,
and Monroe (2007), measures for procedural fairness and distribu-tive fairness use a set of four items: fair, acceptable, unfair, and
satisfactory. Kukar-Kinney et al. (2007) adapted this measure from
the study of Campbell (1999) and added additional items to mea-
sure pricing policy, fairness (procedural fairness), and pricefairness
(distributive fairness).
In addition to the main factors, the current study includes three
covariates in factorial ANCOVA to enhance the internal validity:
tolerance-of-crowding, familiarity with RM, and gender. Perceived
scarcity of capacity in a restaurant may relate to customers’ per-
ceptions of crowding. Therefore, a tolerance-of-crowding measure,
adopted from Machleit et al. (1994), was included as a covariate in
the factorial ANCOVA. Moreover, previous research found that the
familiarity of customers with RM practices has an impact on fair-
ness perceptions toward RM (e.g., Taylor and Kimes, 2010; Wirtz
-
8/16/2019 Capacity scarcity Restaurant revenue management.pdf
6/11
C.Y. Heo et al. / International Journal of Hospitality Management 35 (2013) 316–326 321
and Kimes, 2007). Wirtz and Kimes (2007) suggested that as cus-
tomers become more familiar with RM practices, the unfairness
perceptions decline, over time, because familiarity tends to adjust
customers’ reference-points for transactions and prices, resulting
in lower or even no impact on perceptions of pricing fairness. Thus,
the familiarity of customers with RM practices (that is, a demand-
based pricing policy: weekdays-versus-weekends) in restaurants
was included as another covariate in the factorial ANCOVA. Last,
Beldona and Namasivayam (2006) examined gender differences
in relation to perceived price fairness and subsequent repurchase
intentions. Beldona and Namasivayam found that females’ percep-
tions of fairness were significantly lower throughout all pricing
scenarios which framed both discount and surplus situations.
Accordingly, the model in this study incorporates gender as the
third covariate.
The survey, initiated with a brief introduction to the nature of
the study, proceededwith participants reading scenarios.The ques-
tionnaire includes two questions for perceived scarcity of capacity
and one question for perceived price difference. Second, the par-
ticipants answered questions regarding their perceived value of
the restaurant’s expected offerings and their fairness perceptions
for the variable representing demand-based pricing policy (proce-
dural fairness) and the price difference (distributive fairness). For
the final portion, questions measured respondents’ tolerance for
crowding and familiarity with RM practices in restaurants. Demo-
graphic questions such as gender, age, ethnicity, and education
concluded the survey.
3.2. Pretest
The questionnaire underwent a pilot test, which involved 66
undergraduates in a hospitality program at a university on the east
coastof the United States. Afterexcluding unqualifiedand/orincon-
sistent responses, 55responses(83.3%)remainedfor analysis of the
pilot test. Among the respondents, 18.2% were males, and 81.8%
were females. Assessment of the scales included checks for inter-
nal consistency and unidimensionality (Traub, 1994). The average
variance extracted (AVE) for each construct was above the recom-mended value of .50 (Fornell and Larcker, 1981). For all constructs,
reliability was above the suggested cut-off point of .70 (Nunnally,
1978). The t-test revealed that respondents perceived price differ-
ences between weekdays and weekends for “10% vs. 40%” to be
different (t -value: −5.09, p-value: .00). However, post hoc anal-
ysis shows that manipulations for perceived scarcity of capacity
did not elicit responses consistent with the hypothesis. A revised
description, “The tables are always unavailable on Friday and Sat-
urday” clarified the notion of perceived scarcity of capacity in the
“High” scarcity situation. The description of a restaurant in the
“Low” scarcity situation had similar revision from “The tables are
sometimes unavailable on Friday and Saturday” to “The tables are
normallybut not always availableon Friday and Saturday.” Another
pre-test, using the edited questions, evaluated the revised manip-ulations for perceived scarcity of capacity, resulting in appearance
of significant differences in perceived scarcity of capacity among
the three scenarios.
3.3. Data collection
The final data, collected during May 2010, represent subjects
from a general population, who requested tourism information for
Arizona, Florida, and Texas. Twelve thousand emails were divided
into twelve groups and allocated to each of the twelve scenarios.
Each respondent was asked to participate in one scenario (scarcity:
No, Low, or High; price difference: 10%, 20%, 30%, or 40%). Two
reminder emails, after 7 days and15 days from theinitialinvitation,
attemptedto enhance therate of response.From the 12,000 emails,
Table 2
Sample’s demographics.
Frequency Percentage
Gender Male 119 24.7
Female 362 75.3
Age 20–29 27 5.6
30–39 75 15.6
40–49 119 24.7
50–59 148 30.8
60 and over 112 23.3Income US$ 20,000 or under 31 6.7
US$ 20,001–35,000 42 9.1
US$ 35,001–50,000 77 16.6
US$ 50,001–75,000 102 22.0
US$ 75,001–100,000 107 23.1
US$ 100,001 or more 104 22.5
Education High school or Less 116 24.1
Associate 105 21.8
Bachelors degree 158 32.8
Masters 84 17.5
Doctorate degree 18 3.7
Ethnicity Caucasian 419 86.7
African American 19 3.9
Hispanic 8 1.7
Asian 17 3.5
Other 20 4.1
549 respondents participated in the survey (4.6% response rate),
and the response rate for each scenario varied from 3.5% to 5.5%.
Of the 549 participants, 44 incomplete questionnaires disqualified
those responses, resulting in, 505 responses remaining for analysis
(4.2% valid response rate).
Asseenin Table 2, the sample(n= 505) in theanalysisrepresents
75.3% females (n= 362). The ages of respondents ranged from 20 to
85, with a median age of 49.3 years. Respondents aged 50–59years
old (30.8%) and 20–29 years old (5.6%) accounted for the largest
and smallest proportions of respondents, respectively. In terms of
income, the respondents were fairly evenly distributed, with the
largest group (23.1%) reporting incomes between US$ 75,000 and
US$ 100,000, and with the smallest group (6.7%) reporting incomes
of US$ 20,000 or under. In terms of education, the largest category
was holders of bachelor’s degrees (32.8%), followed by high school
diplomasor less(24.1%).Last,86.6% of respondents wereCaucasian;
the remainder were African-American (3.9%) and of other ethnici-
ties (3.9%).
4. Analysis and results
Manipulation checks included a series of one-way ANOVAs to
determine whether or not the experimental treatments led to the
desired significant differences in the various conditions (results
appear in Tables 3 and 4). The study used two questions (PS1 and
PS2) to check thedifference in perceivedscarcitybetween High and
Low, and Low and No, and results of the two questions show sig-nificant differences between both High and Low, and Low and No
(Table 3). The perceived price difference among 10% (mean: 4.10),
20% (mean: 5.04), and 30% (mean: 5.75) was also significant, but
Table 3
Manipulation checks for perceived scarcity.
Level N Subset for alpha = .05
1 2 3
PS1 High 173 2.20 – –
Low 167 – 2.66 –
No 165 – – 5.83
PS2 No 165 2.33 – –
Low 167 – 4.57 –
High 173 – – 5.60
-
8/16/2019 Capacity scarcity Restaurant revenue management.pdf
7/11
-
8/16/2019 Capacity scarcity Restaurant revenue management.pdf
8/11
C.Y. Heo et al. / International Journal of Hospitality Management 35 (2013) 316–326 323
Table 8
ANOVA analysisof theeffects of perceived scarcity on perceivedvalue.
Price difference
10% 20% 30% 40%*
F -value 1.31 .20 .55 3.11
P .27 .82 .58 .05
* It indicates two-tailed significance at 0.05 level.
Table 9
Post hoc analysisof the effects of perceived scarcity on perceived valuefor the40%
price difference scenario.
Subsetfor alpha= 0.05
N 1 2
Low 35 3.31 –
No 35 3.74 3.74
High 35 – 3.94
Table 10
ANOVA analysis of the effects of perceived scarcity on fairness perceptions.
Price difference
10% 20% 30%* 40%
F -value .21 .67 3.53 .02
P .89 .52 .03 .98
* It indicates two-tailed significance at 0.05 level.
capacity does not influence customers’fairness perceptions toward
the restaurant’s RM practices, and that the price difference influ-
ences customers’ fairness perceptions. Two covariates, familiarity
with RM practices (F -value: 108.18, p-value: .00) and gender (F -
value: 14.07, p-value: .00), were found to be significant, suggesting
that familiarity with RM practices and gender affect customers’
fairness perceptions of a restaurant’s RM practices. No difference
appears for perceptions of difference between the 30% and 40%.
Hence, the data from the 40% scenarios were omitted from thefactorial ANCOVA. However, additional analysis including the 40%
of the price difference scenarios provided the qualitatively same
results.
This study performedone-wayANOVA to investigatewhetheror
not the effects of perceived scarcity of capacity on perceived value
and fairness perceptions differ for each price difference scenario.
As presented in Table 8, no significant effects for perceived scarcity
of capacity appeared for the 10%, 20%, and 30% price differences,
but for the 40% price difference situation, the effect of perceived
scarcity of capacity on theperceived value was significant (F -value:
3.11, p-value .05). Post hoc analysis revealed that the difference
in perceived value appeared to be between Low and High scarcity
situations (Table 9).
Table 10 shows that no significant effects from perceived
scarcity of capacity on fairness perceptions appear for the 10%,
20%, and 40% of price differences; the effect of perceived scarcity
of capacity was significant only for the 30% of the price difference
situation (F -value: 3.53, p-value: .03). Consistent with the resultfor
Table 11
Post hoc analysis of the effects of perceived scarcity on fairness perceptionsfor the
30% price difference scenario.
Subsetfor alpha= 0.05
N 1 2
Low 35 2.62 –
No 35 2.72 2.72
High 35 – 3.33
perceived value, the differential for perceived scarcity of capacity
was present between Low and High scarcity situations, as shown
in Table 11.
5. Discussion
The main purpose of this study is to examine the effects of
perceived scarcity of capacity and the price difference on cus-
tomers’ perceived value for a restaurant’s offerings and their
fairness perceptions regarding a restaurant’s RM practices. In addi-
tion, an empirical investigation considers the interaction effect of
the scarcity of capacity andprice difference, proposed by Brock and
Brannon (1992). Factorial ANCOVA tested six hypotheses.
According to the commodity theory, knowledge of a product’s
scarcity affects consumers’ perceptions and evaluations of attrac-
tiveness, desirability,expensiveness, quality, and tasteof a product.
However, the results of this study do not support the hypothesized
positive effect of customers’ perceived scarcity of capacity for a
restaurant on the perceived value of the restaurant’s offerings. In
addition, the equity theory proposed by Adams (1965), suggests a
positive effectof scarcity of capacity on customers’ fairness percep-
tions in the context of a restaurant’s RM (H2). However, this study
finds an insignificant effect.
Although thecommodity theoryhas thesupportof many empir-
icalstudies,several researchersarguedthat the scarcity of a product
does not have a positive effect on consumers’ evaluations of the
product in allcases.Some studies argued that the appealof scarcity
led consumers to scrutinize an offer more thoroughly and did not
necessarily result in favorable perceptions for the scarce product
(Brannon and Brock, 2001; Brock and Brannon, 1992; Inman et al.,
1997). Brock and Brannon (1992) argued that for scarce, negatively
valenced objects, for which an individual might have a clear aver-
sion, the original notion of usefulness is discarded. Moreover, a
negatively valenced experience will be regarded as more aversive
to the extent that it is considered rare, because people confronting
scarcity have motivation to think about the message. Thus, scarcity
can also render negative evaluations more extreme (Brannon andBrock, 2001).
In addition, Suri et al. (2007) examined the influence of
perceived scarcity on consumers’ processing of price information
in the context of product and service purchases. The results of that
studyshowed that during scarcity, consumers’ perceptions of qual-
ity and monetary sacrifice exhibit different patterns of responses
depending on the relative price level and on consumers’ motiva-
tions to process information. According to Suri et al., motivation
to process information moderates the effect of scarcity. During
scarcity, a high price in a high- motivation condition, increase the
perceptions of quality and value, as well as purchase intentions. In
the current study, the respondents read a scenario and imagined
participating in the situation. However, their motivation to process
given information from the scenario is unknown, perhaps explain-ing the results of the insignificant effects of perceived scarcity of
capacity in this study.
The unexpected insignificant effects of perceived scarcity may
also associate with thetypeof scarcityand thetypeof product.Gierl
and Huettl (2010) argued that positive effects from evaluations of
products depend on the type of scarcity (i.e., scarcity due to sup-
ply vs. scarcity due to demand) and the type of products (i.e., high
vs. low suitability for conspicuous consumption). Gierl and Huettl
(2010) combined the types of scarcity and the types of products
and found that scarcity due to supplywould resultin more positive
attitudes toward products than scarcity due to demand for high-
conspicuous consumption conditions. The findings of the current
research support the notion that scarcity due to supply enhances
evaluations of products, but positive evaluations of the product
-
8/16/2019 Capacity scarcity Restaurant revenue management.pdf
9/11
324 C.Y. Heo et al. / International Journal of Hospitality Management 35 (2013) 316–326
decline when attributing scarcity to high demand. Thus, in the cur-
rentstudy, customers’perceptions of a dining experienceat a casual
dining restaurant (whether conspicuous consumption or not) may
affect the results. If the respondents regarded having dinner at the
casual dining restaurant as conspicuous consumption, scarcity due
to high demand may not affect attitude.
Regarding the price difference factor, the current results sup-
port the negative effect of the price difference on customers’
fairness perceptions regarding a restaurant’s RM practices. These
results suggest that as theperceived price differencebetweenhigh-
demand and low-demandperiods increases, and customers tend to
perceive a restaurant’s RM practices to be unfair. However, incon-
sistent with the hypothesis, the results of the present study do not
support the negative effect of the price difference on customers’
perceived value for a restaurant’s offerings.
The findings of this study may provide restaurant managers
withsome practicalimplications.Based on the commodity theory, a
restaurant’s manager may prefera marketingmessage thatempha-
sizes the restaurant’s full capacity during high-demand periods.
However, based on this study’s findings, such information may
not positively affect customers’ perceptions regarding the value
of the restaurant’s offerings or the fairness of the restaurant’s RM
practices. In such a case, the manager may instead focus on other
features or information in advertising. Based on the finding of this
study, a manager may need to avoid sharing information of avail-
able discounts that differentiate high- and low-demand periods
with customerswho plan to dine or aredining duringhigh-demand
periods because such information may discourage patronage; cus-
tomers may likely forma negative perceptionof the restaurant’s RM
practices. An important issue for future research in this regard is to
discover the optimal level for price differences at which customers
will not consider the price as unfair, thus allowing restaurants to
maximize revenues.
This study performed several additional analyses and found
that for the 40% price difference scenario, the perceived scarcity
of capacity (between high and low) positively affects the perceived
value of a restaurant’s offerings. In other words, customers who
perceive a restaurant’scapacityas very scarce consider that restau-rant’s offerings more valuable compared with the customers who
perceive a restaurant’s capacity as less scarce when the restau-
rant provides a 40% discount during weekdays. This finding may
suggest that the scarcity factor plays a significant role only when
substantial andnot marginal price differences arepresent between
low- and high-demand periods. A restaurant manager whose
restaurant enjoys full capacity during high-demand periods may
want to consider heavily discounting menu offerings only dur-
ing low-demand periods to enhance customers’ perceived value of
the restaurant’s offerings. This suggestion arises from the notion
that when the restaurant provides a marginal discount during
low-demand periods, the restaurant’s high occupancy during high-
demand periods (i.e., high scarcity of capacity) does not positively
affect customers’ perceptions of value.From exhaustive research, this study is the first to apply com-
modity and equity theories to the RM context, more specifically
for the context of restaurants. Although the results do not support
the hypotheses regarding perceived scarcity of capacity, this study
raises an important issue of consumers’ perceptions of scarcity of
service resourcesin thecontext of restaurants’ RM.In addition,pre-
vious RM research has provided little investigation of the effect of
the price differences between low- and high-demand periods on
perceived value and fairness perceptions, and the current study
fills this gap in the literature. Previous RM literature also found
that familiarity with RM practice is a considerable factor for fair-
ness perceptions of RM, and this study confirms the importance
of a respondent’s familiarity with RM practices in the restaurant
context.
6. Limitations and suggestions for future research
Several limitations ofthis studyshould benoted. First,giventhat
the respondents represent only one geographic region (i.e., U.S.),
findings ofthe studymay not generalizefor otherregions.However,
considering the representative sample of 505 respondents with a
wide range of backgrounds, as shown in profile of the sample (e.g.,
age, education, and income level) discussed in 3.3. Data Collection,
the sample may be an adequate proxy for the general restaurant
goers in the U.S. In addition, the nature of the scenarios, including
scarcity, might limit the generalizability of the results of the study,
which only analyzes the effects of scarcity in cases of no provided
information regardingother restaurant’s practicesand alternatives.
In a scenario of choosing a restaurant, information ofnearby restau-
rants may be a critical factor influencing customers’ interpretation
of signals of scarcity.
Second, the experiment in this study is scenario-based and did
not occur in a field setting. Online-based scenarios may not fully
represent a restaurant’s on-going situations in the field. Capturing
all of the nuances of the actual situation in a scenario is difficult,
and consequently, respondents may have difficultypredicting their
feelings (e.g., perceived scarcity of capacity and attitude toward
the restaurant’s RM practices) in a hypothetical situation. Specifi-
cally, the description of the high-scarcity situation (tables always
unavailable) may not be realistic. However, after the manipula-
tion check through the pre-test, the original manipulation for the
scarcity measure is, apparently insignificantly different among dif-
ferentsituations.Thus, thecase is intentionally extreme to ensurea
significant difference among the scenarios. Furthermore, the main
findings of theinfluenceof scarcityof capacity on value andpercep-
tions of fairness do not appear significant even with this extreme
case.
Third, regarding the measures for perceived value, different
aspects of value (e.g., hedonic vs. utilitarian values) may represent
extensions of this research in the future. Another constraint in this
study is the sole focus on customers’ perceptions of disadvanta-
geous situations, because the hypothetical plans of respondents
were for a visit to a restaurant on a Friday. Future research mayconsider examining the reactions of customers in advantageous
situations (e.g., obtaining discounts) to a restaurant’s RM practices
using scenarios of scarcity.
This study is experimental in nature and one of the first few
studies to explore the effects of perceived scarcity of capacity in
the context of restaurants’ RM practices. Although this study did
not find a significant effect from scarcity on perceived value and
fairness perceptions, additional research is necessary to investi-
gate the effects, using different settings and other factors. Future
research are encouraged to identify other factors that could influ-
ence consumers’ evaluations of price information to understand
further the impact of scarcity on the valuation of service offerings
from a restaurant. In addition, future research also can examine
cognitive procession as a mediator to understand better the under-lying mechanism for the effects of scarcity.
Acknowledgement
The authors would like to thank the anonymous reviewers for
their constructive comments on improving an early version of this
paper. This project is partly supported by a research grant funded
by the Hong Kong Polytechnic University.
Appendix A. APPENDIX. Sample Scenario
Imagine that you are planning to have a dinner with your
family on Friday. Your friend recommended several casual dining
-
8/16/2019 Capacity scarcity Restaurant revenue management.pdf
10/11
http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0240http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0235http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0230http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0230http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0230http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0230http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0230http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0230http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0230http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0230http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0230http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0230http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0230http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0230http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0230http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0230http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0230http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0230http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0230http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0230http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0225http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0225http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0225http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0225http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0225http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0225http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0225http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0225http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0225http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0225http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0225http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0225http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0225http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0225http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0225http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0225http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0225http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0220http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0220http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0220http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0220http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0220http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0220http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0220http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0220http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0220http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0220http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0220http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0220http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0220http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0220http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0220http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0220http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0220http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0220http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0215http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0215http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0215http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0215http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0215http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0215http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0215http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0215http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0215http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0215http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0215http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0215http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0215http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0215http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0215http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0215http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0215http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0215http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0215http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0210http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0210http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0210http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0210http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0210http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0210http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0210http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0210http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0210http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0210http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0210http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0210http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0210http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0210http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0210http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0205http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0205http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0205http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0205http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0205http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0205http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0205http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0205http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0205http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0205http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0205http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0205http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0205http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0205http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0205http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0205http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0205http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0205http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0205http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0205http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0200http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0200http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0200http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0200http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0200http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0200http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0200http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0200http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0200http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0200http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0200http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0200http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0200http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0200http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0200http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0195http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0195http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0195http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0195http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0195http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0195http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0195http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0195http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0195http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0195http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0195http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0195http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0195http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0195http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0195http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0195http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0195http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0195http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0190http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0190http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0190http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0190http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0190http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0190http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0190http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0190http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0190http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0190http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0190http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0190http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0190http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0190http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0190http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0190http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0185http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0185http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0185http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0185http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0185http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0185http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0185http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0185http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0185http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0185http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0185http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0185http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0185http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0185http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0185http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0185http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0185http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0180http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0180http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0180http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0180http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0180http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0180http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0180http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0180http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0180http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0180http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0180http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0180http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0180http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0175http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0175http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0175http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0175http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0175http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0175http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0175http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0175http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0175http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0175http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0175http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0175http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0175http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0170http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0170http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0170http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0170http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0170http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0170http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0170http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0170http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0170http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0170http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0170http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0170http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0165http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0165http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0165http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0165http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0165http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0165http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0165http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0165http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0165http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0165http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0165http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0165http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0165http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0165http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0165http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0165http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0165http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0165http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0165http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0165http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0160http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0160http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0160http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0160http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0160http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0160http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0160http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0160http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0160http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0160http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0160http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0160http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0160http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0160http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0160http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0160http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0160http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0160http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0155http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0155http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0155http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0155http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0155http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0155http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0155http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0155http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0155http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0155http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0155http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0155http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0155http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0155http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0155http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0155http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0155http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0155http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0155http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0150http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0150http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0150http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0150http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0150http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0150http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0150http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0150http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0150http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0150http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0150http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0150http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0150http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0150http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0150http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0150http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0150http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0150http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0150http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0145http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0145http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0145http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0145http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0145http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0145http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0145http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0145http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0145http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0145http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0145http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0145http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0145http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0145http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0145http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0145http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0145http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0145http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0145http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0140http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0140http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0140http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0140http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0140http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0140http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0140http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0135http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0135http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0135http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0135http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0135http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0135http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0135http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0135http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0135http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0135http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0135http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0135http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0135http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0135http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0135http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0135http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0135http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0130http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0125http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0120http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0120http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0120http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0120http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0120http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0120http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0120http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0120http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0120http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0120http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0120http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0120http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0120http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0120http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0115http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0115http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0115http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0115http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0115http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0115http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0115http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0115http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0115http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0115http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0115http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0115http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0115http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0115http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0115http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0115http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0115http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0115http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0115http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0110http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0105http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0100http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0095http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0095http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0095http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0095http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0095http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0095http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0095http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0095http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0095http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0095http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0095http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0095http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0095http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0095http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0090http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0090http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0090http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0090http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0090http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0090http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0090http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0090http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0090http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0090http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0090http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0090http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0090http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0090http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0090http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0090http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0090http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0090http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0090http://refhub.elsevier.com/S0278-4319(13)00077-7/sbref0090http://refhub.elsevier.com/S02