ArticlePDF Available

A Review of Methods for Measuring Willingness-to-Pay

Authors:

Abstract and Figures

Knowledge about a product's willingness-to-pay on behalf of its (potential) customers plays a crucial role in many areas of marketing management like pricing decisions or new product development. Numerous approaches to measure willingness-to-pay with differential conceptual foundations and methodological implications have been presented in the relevant literature so far. This article provides the reader with a systematic overview of the relevant literature on these competing approaches and associated schools of thought, recognizes their respective merits and discusses obstacles and issues regarding their adoption to measuring willingness-to-pay. Because of its practical relevance, special focus will be put on indirect surveying techniques and, in particular, conjoint-based applications will be discussed in more detail. The strengths and limitations of the individual approaches are discussed and evaluated from a managerial point of view.
No caption available
… 
Content may be subject to copyright.
1 Preprint to appear in Innovative Marketing, 2006
A REVIEW OF METHODS FOR MEASURING
WILLINGNESS-TO-PAY
Christoph Breidert
1
, Michael Hahsler
2
, Thomas Reutterer
3
1
PONTIS Venture Partners, Austria
2
Institute for Information Business, Vienna University of Economics and Business Administration, Austria
3
Institute of Retailing and Marketing, Vienna University of Economics and Business Administration, Austria
Abstract
Knowledge about a product’s willingness-to-pay on behalf of its (potential) customers plays a
crucial role in many areas of marketing management like pricing decisions or new product
development. Numerous approaches to measure willingness-to-pay with differential conceptual
foundations and methodological implications have been presented in the relevant literature so far.
This article provides the reader with a systematic overview of the relevant literature on these
competing approaches and associated schools of thought, recognizes their respective merits and
discusses obstacles and issues regarding their adoption to measuring willingness-to-pay. Because
of its practical relevance, special focus will be put on indirect surveying techniques and, in
particular, conjoint-based applications will be discussed in more detail. The strengths and
limitations of the individual approaches are discussed and evaluated from a managerial point of
view.
Keywords: Willingness-to-pay, pricing, surveying techniques, conjoint measurement.
Introduction
Despite considerable advances in both academic and applied pricing research over the past
decades, many companies still make their pricing decisions without a profound understanding of
the likely response of (potential) buyers and competitors to alternative prices quotations. As a
result of missing adequate knowledge of the customer's willingness-to-pay (WTP) for their
products, these companies fail to pursue a pricing strategy that is suitably customized to their
marketing environment and thus also risk ignoring valuable sources for increasing profitability of
the products offered. Different practical studies have shown that minor variations of prices and
the corresponding consumer behavior can have notable effects on revenues and profits (Marn, et
al., 2003).
Companies often adopt some business rules and follow a strategy that could be denoted as
“intuitive” pricing. Remarkably, such a behavior is not limited to retailing or service industries
only, where mark-up-pricing is still representing the predominant practice (cf. Levy et al., 2004,
Berman and Evans, 2001, p. 555 ff., Monroe, 2003, p. 257). Several recent studies indicate that
only 8 to 15% of all companies develop pricing strategies based on likely buyer response
behavior (Monroe and Cox, 2001). In contrast to what seems to be common practice, managers
consider the knowledge of customers’ responses to different prices as a cornerstone of marketing
strategies, particularly in the areas of product development, brand management, value audits, and
competitive strategy (Anderson et al., 1993).
2 Preprint to appear in Innovative Marketing, 2006
Researchers agree with managers on the importance of valid WTP estimates. Balderjahn (2003,
p. 387) considers valid estimates of WTP essential for developing an optimal pricing strategy.
Similar arguments about the importance of WTP and perceptions of value by customers can also
be found by many other authors (cf. Monroe, 2003, pp. 11-12, Nagle and Holden, 2002, p. 7 and
pp. 104-105, and Simon, 1992, p. 365 ff.). Such estimates can be used to forecast market
response to price changes and for modeling demand functions. Furthermore, various approaches
to measure brand equity (cf. e.g., Farquhar, 1989, Srivastava and Shocker, 1991, Park and
Srinivasan, 1994) emphasize customers’ WTP in terms of the (monetary) added value endowed
by a brand to a specific product vis-à-vis its’ competitors or an unbranded baseline product. Good
overviews on contemporary approaches to brand equity measurement are presented by Sattler
(1995) or Ailawadi et al. (2003).
As our literature review will show, a huge variety of competing approaches and corresponding
analytical techniques for measuring WTP has been added to the realm of marketing literature in
the past. Equally, evidence on the relative empirical performance is scattered across a number of
comparative studies of selected measurement approaches. Despite that fact, today the research
area is rather fragmented and there has been relatively little attention paid to attempts towards an
in-depth and comprehensive synopsis of the various existing approaches as well as what is known
from empirical research regarding their effectiveness and efficiency to measure WTP. The
purpose of this paper is to provide some insights into this direction. Moreover, we conceptualize
a general evaluation framework that aims to be helpful for managers when it comes to select a
particular measurement approach in practical applications. Against this managerial background, it
is important to balance the specific merits and obstacles of the available approaches.
The remainder of this paper is organized as follows: First, we provide a review of methods for
measuring WTP and the corresponding data collection techniques. For the sake of clarity, we
start with a classification of the various methods, thereafter we provide references to related
substantial theoretical and empirical work, and discuss advantages and drawbacks for each
method. Finally, we consolidate the most important results to provide guidance for managers who
plan to measure WTP and provide ideas how to enhance indirect surveying techniques.
Classification of Methods
Several authors proposed different hierarchical classification frameworks to organize existing
methods to WTP estimation. Marbeau (1987) distinguishes the estimation methods on the highest
level, whether they are monadic tests or competitive tests. In the former, price information is
elicited without considering a competitive context. In the latter, a competitive context is present.
Balderjahn (2003) distinguishes estimation methods on the highest level, whether they elicit price
information at the individual level or at the aggregate level. Nagle and Holden (2002) classify
techniques for measuring price sensitivity at the highest level into uncontrolled and
experimentally controlled measurement of the variables. Furthermore, they classify the
techniques based on the variable measurement into measurement of purchase behavior and
measurement of purchase intention.
To guide the reader through this paper, we use the classification framework based on data
collection methods as presented in Figure 1. On the highest level, methods can be distinguished
whether they utilize surveying techniques or whether they are based on actual or simulated price-
response data. Taking a closer look at response data, market observations can be used or data can
be generated by performing experiments. Experiments can further be divided into field
experiments and laboratory experiments. Since auctions are a very important form of laboratory
3 Preprint to appear in Innovative Marketing, 2006
experiment, we included them as an extra method in the classification framework. Results
obtained from price-responses are often referred to as revealed preference data. Looking at
survey-based techniques for estimation of WTP, there exist direct and indirect surveys for
collecting the relevant data. In contrast to revealed preferences, preference data derived from
surveys are frequently referred to as stated preferences (cf. Louviere et al., 2000, p. 20 ff.).
Figure 1 Classification framework for methods to measure willingness-to-pay.
With direct surveys, respondents (e.g., selected customers) are asked to state how much they
would be willing to pay for some product. In indirect surveys some sort of rating or ranking
procedure for different products is applied in order to estimate a preference structure from which
WTP can be derived. Conjoint analysis and discrete choice analysis are examples of indirect
surveying methods.
In practice, selecting a feasible method for measuring WTP is often restricted, for example, by
time or monetary constraints. Data collection determines to a great extent the time and the costs
of the method. Therefore, this framework provides a useful guideline for choosing an appropriate
method and it was used to structure this paper.
Analysis of Market Data
Analyzing observed market data (i.e., sales data) is often used to estimate price-response
functions. Depending on their data sources, sales data suitable for WTP estimation can be
roughly subdivided into the following two subtypes: (1) Panel data individual purchase data
reported by members of a customer panel and (2) store scanner data sales records from retail
outlets. Using historical market data is based on the assumption that past demands can be used to
predict future market behavior. This implies that the product for which future demand is
estimated has only been exposed to minor variations in the product profile. A problem of the
method arises if the historical data do not contain the necessary price variations to cover the
desired spectrum of WTPs. Indeed, small ranges of observed price variations often appear to be a
4 Preprint to appear in Innovative Marketing, 2006
pitfall when analyzing historical sales data. For this reason Sattler and Nitschke (2003) even
classify WTP estimation based on market data as infeasible. Demand curves based upon sales
data are usually modeled with regression techniques. A restriction is that this is only possible if
the requirements of the independent variables are met (cf. Balderjahn, 2003, p. 399, Nessim and
Dodge, 1995, p. 72).
Note that sales data are often available at the aggregate level only. The data are aggregated over
time and different stores are combined. In such cases derivation of individual customer level
estimates is infeasible. This is different with panel data where the actual prices paid for products
are observed at the individual level. The drawbacks are that running a customer panel entails high
operating costs. Furthermore, it is often questionable, whether the customer panel adequately
represents the market (Nagle and Holden, 2002, p. 335).
Conventional scanner data provided by commercial market research companies are usually not
aggregated over time but aggregated at the store level. Scanner data are useful for observing
response to short time price variations. But because of store level aggregation, individual level
repeated purchase behavior cannot be extracted from such sources of scanner data. Parallel to the
diffusion of point-of-sale (POS) scanning technologies combined with customer ID cards,
however, customer specific reactions to items’ price variations are accessible at least for purchase
transactions realized with a specific retailer (cf. Kahn and McAllister, 1997, p. 106 ff.). Inclusion
of price variations across competing stores is warranted by POS-ID scanner panels (cf. e.g.,
Simon et al,. 1982 and Erichson, 1992).
In general, using market data has the advantage that real purchases, which include information
about competing products, are used instead of, for example, only stated purchase intentions.
Limitations are that the price variations in the data are normally very limited and that it is not
possible to estimate WTP for entirely new products where no data are available (yet). Empirical
applications are provided by Kamakura and Russell (1993) or Leeflang and Wittink (1992).
Experiments
Generally, experiments can be divided into laboratory experiments and field experiments (e.g.,
Malhotra, 2004). Both types can be applied in pricing studies including WTP estimation.
Laboratory Experiments
In laboratory experiments, purchase behavior is typically simulated by giving the subjects an
amount of money and asking them to spend the money on a specific selection of goods. The
goods and prices are varied systematically. Methods for accessing price response of this form
have been applied by Silk and Urban (1978) in their well-know ASSESSOR procedure. In
laboratory experiments the results are obtained quickly. Due to the non-biotic context of
investigation, a major drawback is that the subjects are aware of the experimental situation. This
might lead to subjects becoming more rational of their purchase behavior compared to their
normal shopping behavior which can lead to low external validity (Nessim and Dodge, 1995,
p.74). Another source of bias might be the artificial setup as described above, in which the
subjects either do not take real possession of the “purchased” goods, or do not use their own
money (Nagle and Holden, 2002, p. 341).
5 Preprint to appear in Innovative Marketing, 2006
Field Experiments
Field experiments or in-store purchase experiments do not suffer from the problem of the
artificial setup because they are performed in a real-world shopping environment. Depending on
the experimental conditions, the respondents are aware of participating in an experiment or not.
Field experiments are often conducted in form of so-called test markets. In different test markets
the prices are systematically varied and the consumers’ responses are analyzed. A crucial issue in
test market analysis is to select small scaled market environments that are representative for the
target market under investigation. An overview of institutional test market simulations in
Germany is provided by Gaul et al. (1996). Compared to laboratory experiments, considerably
higher expenditures and the longer time intervals entailed by monitoring market responses to
price changes are reported as main drawbacks of test markets as well as other forms of field and
in-store experiments (Nagle and Holden 2002 p. 341, Sattler and Nitschke, 2003, and Urban and
Hauser, 1993, p. 495 ff.).
Auctions
A special application of experiments are auctions which can be carried out as laboratory or field
experiments. In laboratory settings auctions have been intensively employed for WTP elicitation.
If the true monetary evaluation of a product as perceived by the customer(s) were known, there
would be no need for an auction. The offering party would simply sell the good to the bidder with
the highest valuation at a price close or equal to that valuation. If the seller is uncertain about
customers’ valuations, however, an auction can provide valuable insights to sell the item at a fair
price. Therefore, an auction is useful to gain knowledge of consumers’ evaluations of a product
or brand and can therefore be used to reveal consumers’ valuations to facilitate future pricing
decisions.
According to Wertenbroch and Skiera incentives to reveal true WTP can be provided with
Vickrey auctions: “Vickrey suggests that incentive compatibility is ensured if a given bid
determines only whether the bidder has the right to buy the good that is auctioned off
Wertenbroch and Skiera (2002). The Vickery auction takes place in sealed form, and the
purchase price is determined by the second highest bid. A participant in the auction submits a bid
containing how much he or she would be willing to pay in sealed form, for example in a closed
envelope. If the participant has the highest bid, he or she wins the auction. However, the
participant only has to pay the price of the second highest bid. With this mechanism the
participants are provided an incentive to reveal their true valuation, because they must buy the
good if their bid wins the auction (Vickrey, 1961).
Skiera and Revenstorff (1999) investigate the ability of Vickrey auctions to reveal consumers’
WTP. The mechanism of the auction was described to the students, and the optimal bidding
strategy was explained (which is bidding the true valuation) and different phone contracts were
offered in a Vickrey auction. Based on a questionnaire the subjects seemed to have a good
understanding of the mechanism of the auction. But the optimal bidding strategy (bid the true
valuation) was less clear to the subjects which might be a problem for the method.
In a different experiment Sattler and Nitschke (2003) find that the Vickrey auctions in addition to
the first-price auctions (auction where the highest bid wins) both tend to overestimate consumers’
WTP. The authors suppose that this effect is due to the overbidding phenomenon. The
overbidding phenomenon occurs when bidders strategically place bids above their true WTP to
increase their chance of winning (Kagel et al., 1987).
6 Preprint to appear in Innovative Marketing, 2006
Another incentive compatible auction form is the well-known BDM procedure introduced by the
authors Becker, DeGroot and Marshak (cf. Becker et al., 1964). In BDM every participant
simultaneously submits an offer price to purchase a product. A sale price is randomly drawn from
a distribution of prices. The possible prices cover an interval from zero to a price greater than the
anticipated maximum price, which any bidder would submit. The bidders whose bids are greater
than the sale price receive a unit of the good and pay an amount equal to the sale price. The
mechanism is incentive compatible for the same reason as the Vickrey auction: A given bid
determines only whether the bidder has the right to buy the good that is auctioned off. The price
is set by some mechanism and is below the participants bid.
The BDM procedure was tested by a number of researchers for its ability to forecast WTP.
Wertenbroch and Skiera (2002) tested it together with a Vickrey auction in a field experiment
with a purchase obligation for the participants. The participants of the experiment easily
understood the BDM method and hardly any of the approached individuals refused to participate.
Validity was determined by relating the estimated WTPs to data from an additional questionnaire
asking the respondents to rate their desire of the tested products. After the experiment the
participants rated how satisfied they were with their purchase. The buyers as well as the non-
buyers were extremely satisfied with the outcome of the BDM experiment. This result indicates
that BDM does not suffer from the overbidding bias, which is found in some Vickrey auctions.
In a study published by Noussair et al. (2004) the Vickrey auction is compared with the BDM
mechanism, with the aim to test which method converges towards the optimal bidding strategy
(bidding the true valuation) more rapidly. The authors found that under the Vickrey auction the
bias from the true valuation is more rapidly reduced and the dispersion of bids is narrowed down
more rapidly. That is, the subjects learn the best bidding strategy more quickly. The authors argue
that the reason for this difference lies in the fact that a deviation from the optimal strategy is more
costly under the Vickrey auction than under the BDM mechanism. With respect to these results
Noussair et al. (2004) conclude that Vickrey auctions are superior to the BDM procedure for
elicitation of WTP towards private goods.
Applied research using experimental auctions in laboratory settings for estimating products’
WTP is rather limited. However, auctions are applied as sales mechanisms in practice as for
example by EBAY and can deliver useful information for understanding buyer's likely response
behavior to different prices.
Another approach to estimate WTP with an auction-like procedure is the so-called reverse-
pricing or name-your-own-price mechanism (e.g., Chernev, 2003 and Spann et al., 2004). In this
mechanism each buyer names a price he or she is willing to pay for a certain product. Based upon
a price threshold set by the seller but unkown to the buyers, the buyers who have the right to
purchase the product are determined. Each buyer pays the price which he or she named. Unlike
the Vickrey auction and the BDM mechanism, reverse-pricing is not incentive compatible. In the
latter each buyer has the incentive to name a lower price than the true valuation in order to get a
better deal. However, it is argued that the named prices will be determined by the WTP and if
each buyer can submit several bids, WTP can be estimated. In contrast to more conventional
auction procedures, numerous applications of reverse-pricing can be found in practice. A well-
known example is priceline.com, a US-based retailer operating since 1998, whose primarily focus
is on sales of flights, rental cars, and vacation packages.
7 Preprint to appear in Innovative Marketing, 2006
Direct Surveys
In the previous sections, we discussed methods for measuring WTP based on revealed preference
data. It is not always possible for a marketing analyst to obtain such data in order to estimate
price-response functions. For example, new or differentiated products would have to be designed
and manufactured before they can be tested experimentally. Typically the number of possible
differentiated products is large and not all candidates can be tested under justifiable budget and
time restrictions. In this case specific surveying techniques that render respondents’ statements
with respect to their price preferences are better suited.
In the next sections, an overview of various surveying and related data analytical techniques with
relevance to WTP estimation will be given. The most important methods, namely conjoint
analysis and discrete choice analysis, are discussed in more detail. According to the classification
provided above, direct surveys can be further divided into expert judgments and customer
surveys. Reflecting their relative suitability to deliver accurate WTP estimates, we only review
expert judgments briefly while focusing on customer surveys.
Expert Judgments
As a heuristic to assess customers’ WTP as well as to provide rough estimates of expected
demand in response to different price levels, expert judgments are quite popular in marketing
practice. They can be collected more time and cost efficient as compared to interviewing
customers.
Usually sales or marketing managers serve as experts in projecting customers’ WTP. Since sales
representatives work directly in the market and in close connection with the consumers, they are
aware of the competitive structure in the market and are sensitive to trends in consumer needs.
Therefore, interviewing sales people can provide an important source of information for demand
estimates. Nevertheless, the opinion of sales people might be biased because of conflicting
objectives of the marketer and the sales force. For example, if the sales force’s rewarding system
is tied to sales volume, intentionally or unintentionally overstated or understated expert
judgments might result in biased estimates (Nessim and Dodge, 1995, p. 70).
Conversely, marketing experts estimates of product demand under different price schedules
might suffer from the distance to the market and the consumers. In contrast to judgments
provided by sales people there are no incentives to over- or understate the true estimates. Usually,
there are less marketing executives in a company than sales force people and the opinions of few
marketing managers might lead to poor forecasts of future demand.
In general, expert judgments seem to be best applicable in a market environment with only small
numbers of customers. In such an environment the customers are known well enough to
adequately approximate their WTP. With a larger and more heterogeneous customer base, the
availability of this knowledge becomes a critical issue.
Authors are divided on the usefulness of using exclusively expert judgments. For example,
Nessim and Dodge (1995, p. 70) argue that expert judgments are an important source of
information because an educated guess is better than a random selection of a presumably
adequate price from a number of price possibilities. Other authors, for example Balderjahn (2003,
p. 391), label expert judgments as a poor measurement instrument with low validity and
discourage from its use.
8 Preprint to appear in Innovative Marketing, 2006
Customer Surveys
If one attempts to forecast consumer behavior in response to different prices, the evident way is
to directly ask the customers. One of the first applications of direct surveys was a psychologically
motivated method for estimating WTP developed by Stoetzel (1954). Stoetzel’s idea was that
there is a maximum and minimum price for each product which can be elicited by directly asking
the customers. Studies based on this idea consist of the following two questions formulated by
Marbeau (1987).
1. “Above which price would you definitely not buy the product, because you can’t
afford it or because you didn’t think it was worth the money?
2. Below which price would you say you would not buy the product because you would
start to suspect the quality?”
Directly asking respondents to indicate acceptable prices is referred to as a direct approach to
measure WTP. Other researchers continued to build upon this idea and research in this area
became quite popular (e.g., Abrams, 1964, Gabor et al,. 1970, and Stout, 1969). Van Westendorp
(1976) introduced a price sensitivity meter which included two additional questions on a
reasonable cheap price and a reasonable expensive price of the product under consideration.
Applications of this approach can be still found in commercial applications (for example, the pan-
european market research company GfK utilizes the procedure to attain critical price ranges for
new or re-launched products).
Recently, several other procedures based on direct customer surveying have been established. An
example is the commercial tool BASES Price Advisor by ACNielsen. The tool’s procedure
presents the subjects with several typical product profiles. The products can be in an early
conceptual phase or already marketable. The subjects are then asked to name prices at which they
consider a product to have a very good value, an average value, and a somewhat poor value.
From the responses, purchase probabilities for different prices are derived. According to
Balderjahn (2003, p. 392) the price for “a somewhat poor” value could be interpreted as
reflecting a respondent’s WTP.
Quite obviously, directly surveying customers has some flaws:
1. By directly asking the customers for a price, there is an unnatural focus on price which
can displace the importance of a product’s other attributes.
2. Customers do not necessarily have an incentive to reveal their true WTP. They might
overstate prices because of prestige effects or understate prices because of consumer
collaboration effects. Nessim and Dodge (1995, p. 72) suppose that buyers in direct
responding may also attempt to quote artificially lower prices, since many of them
perceive their role as conscientious buyers as that of helping to keep prices down”. Nagle
and Holden (2002, p. 344) observe the opposite behavior. To not appear stingy to the
researcher respondents could also overstate their WTP.
3. Even if customers reveal their true valuations of a good, this valuation does not
necessarily translate intro real purchasing behavior (Nessim and Dodge, 1995, p. 72).
4. Directly asking for WTPs especially for complex and unfamiliar goods is a cognitively
challenging task for respondents (Brown et al., 1996). While it remains unclear whether
this leads to over- or understating of true valuations a bias is likely to occur. Note, that
9 Preprint to appear in Innovative Marketing, 2006
this effect also occurs in the Vickrey auction and the BDM mechanism which were
introduced in the previous section about experiments.
5. The perceived valuation of a product is not necessarily stable. Buyers often misjudge the
price of a product, especially if it is not a high frequency purchase or an indispensable
good (Marbeau, 1987). This can lead to an abrupt WTP change once the customer learns
the market price of the product.
An empirical comparison between a field experiment, a laboratory experiment, and a personal
interview was carried out by Stout (1969). In this experiment the prices for different products
were varied and the changes in demand were measured. The results showed significant quantity
changes on systematic price changes in the field experiment. As expected, the demand for the
products decreased as the prices were raised and vice versa. For the other two methods, no
significant changes in demand for the products could be measured caused by raised and lowered
prices. The personal interview even contained reversals. For some respondents the demand
increased when the prices were raised.
Overall, directly asking customers’ WTP for different products seems not to be a reliable method.
Balderjahn (2003, p. 402) explicitly alludes to the distortional effects of direct surveys and pleads
against its use. Nagle and Holden (2002, p. 345) even state that “the results of such studies are at
best useless and are potentially highly misleading.”
Indirect Surveys
Brown et al. (1996) argue that for a respondent it is cognitively easier to decide whether a
specific price for a product is acceptable than to directly assign a price. When the respondent is
presented competing product alternatives and their prices, he or she can be asked to apply a
preference rating, preference ordering or select his or her most preferred choice.
In contrast to directly asking respondents for their WTP, customers are presented product profiles
with systematically varied prices and are asked to indicate whether they would purchase the good
at that price or not. This measurement approach is denoted to as indirect survey (cf. Marbeau,
1987).
In an approach proposed by Camron and James (1987), the authors suggest to present to a
random sample of respondents product profiles with randomly assigned prices. Camron and
James (1987) explain their technique as follows: “Across the selection of product scenarios, the
investigator is free to vary not only the proposed price, but also the levels of all other product
attributes. Each consumer’s willingness or unwillingness to purchase the specific product at the
designated price is recorded. If the experimental design includes variability in the levels of prices,
product attributes, and consumer characteristics, the researcher will be able to use the statistical
techniques [...] to calibrate the demand function.”
Methods based on this idea are discussed in the following sections.
Conjoint Analysis
Generally speaking, conjoint analysis is a technique for measuring individuals’ preference
structures via systematical variations of product attributes in an experimental design. A product’s
attribute is considered as a set of possible realizations, which are referred to as the attribute’s
levels. The respondent is presented a number of product profiles consisting of realizations of the
10 Preprint to appear in Innovative Marketing, 2006
products attributes and arranges them according to her or his perceived preference, e.g., by
indicating a rank order with respect to the degree of preference. These overall preference
evaluations are used to make inferences of the relative contributions of the different attribute
levels. The latter are called part-worths and the evaluation of a full product stimulus is referred to
as the product’s utility. In a conjoint study part-worths are estimated for all attribute levels. That
is, each level is assigned a number, such that the respondents’ preference structure based on the
attributes and levels is represented. The measurement focusing on the different attributes is called
importance. The importance of one attribute is based on the level’s part-worths and simply
describes the range of the part-worths from the least preferred to the most preferred level.
Overview articles on the development of conjoint analysis can be found in Carroll and Green
(1995), Green and Srinivasan (1978, 1990), Gustafsson et al. (2000), Louviere (1994), and Rao
and Hauser (2004).
The conjoint methods considered here follow the assumption of an additive compensatory
decision rule that governs respondents’ information processing (cf. Young, 1973, Lilien et al.,
1992, p. 93 ff., and Johnson, 2001). This assumption on utility formation is quite common in the
marketing literature (cf. Lilien et al., 1992, p. 93 ff.). Hence, the utility of product c is calculated
as the sum of the part-worths of the levels of all attributes as follows:
==
=
a
L
l
alal
A
a
c
xy
11
β
with
c
y
: Rank of product card c
al
β
: Unknown part-worth of level l and attribute a
al
x
= 1 if product card c has level l of attribute a and 0 otherwise.
Green and Rao (1971) were among the first who introduced conjoint measurement into the
marketing literature. Since then, the methodology experienced many extensions and refinements
and today represents an important technique of the modern marketing analyst’s toolbox (see e.g.,
Wittink and Cattin, 1989, Wittink and Burhenne, 1994, Voeth, 1999, Hartmann and Sattler,
2002a, 2002b).
The classic approach to conjoint measurement is full profile conjoint analysis (Green and Rao,
1971). In full profile conjoint analysis the subjects are presented sequences of product profiles
that are described as a combination of the levels of all attributes. These full profile stimuli are
presented on cards with textual and/or graphical descriptions of the product profiles. The
respondent’s task is to evaluate each of these profiles. In the classical case, preference rankings of
the product cards are provided and the parameters al of the respondent’s preference structure
need to be fitted. If the rankings of the respondent are treated using an ordinal scale, typically
MONANOVA (Kruskal, 1965) is applied. If the rankings are assumed to be equidistant, they can
be treated using an interval scale and OLS regression or ANOVA can be applied. However,
11 Preprint to appear in Innovative Marketing, 2006
various studies have shown that different estimation procedures do not lead to significantly
different results (e.g., Cattin and Wittink, 1976, Carmone et al., 1978, and Wittink and Cattin,
1981).
Once the part-worths are fitted, a utility score for any stimulus composed from the attributes and
levels can be predicted using the additive composition rule. Note that the utility can also be
calculated for products that were not actually presented to the respondent during the study as long
as they are constructed only from attribute level which were part of the analysis.
Naturally, apart from rankings other answer formats for measuring preferences in conjoint
experiments can be employed as well (see, e.g., Otter, 2001, p. 64 ff. for an overview and a
comparative empirical evaluation). For example, the respondents can be instructed to rate the
product profiles on a 1-7 attractiveness scale (Huber, 1997) or a 0-100 purchase likelihood scale
(Mahajan et al., 1982). Such rating-based approaches are also frequently denoted as metric
conjoint analysis (Louviere, 1988). The recovery of part-worths is accomplished in the same way
as the estimation procedure for ranking data as described above. Furthermore, the use of choices
as an answer format in conjoint studies converges to the methodology of discrete choice analysis,
which is further discussed below.
Apart from full profile there exist other conjoint approaches. So-called trade-off methods
confront the respondent with only two attributes at a time (Johnson, 1974). This is done for all
attribute pairs. The intention of using partial profiles instead of full profiles is to avoid
information overload tendencies that are likely to occur when all possible attributes are present in
a product stimulus (Green and Srinivasan, 1978). However, the drawback with trade-off matrices
is that by decomposing the factors into two-at-a-time partial profiles, there is a sacrifice of
realism to the respondent who is confronted with incomplete products (Green and Srinivasan,
1978). In this sense full profiles can be considered as more realistic and trade-off method is rarely
used in practice (Wittink and Cattin, 1989).
There exist conjoint methods that use self-explicated data to elaborate the estimation of each
respondent’s utility function. Self-explicated data are elicited directly by asking the respondent to
rate the different attributes levels. Such a procedure typically entails the following steps (cf.
Fishbein, 1967 and Green and Kriege,r 1996).
1. The decision maker rates the desirability of each of a set of possible levels of each of a set
of attributes on, e.g., a 0-10 scale.
2. Following this, the decision maker rates the importance of each attribute on, e.g., a 0-10
scale. In this model, a part-worth is defined as the product of the attribute’s importance
times the associated level’s desirability.
In hybrid conjoint analysis self-explicated data are used to obtained preliminary part-worths for
each respondent. After this, a limited number of full-profile cards is presented for evaluation.
Thus, the number of profiles and, as a consequence, information overload tendencies can be
reduced. The profiles are chosen and presented to all respondents in such a way that each profile
is at least rated once. The evaluations are pooled and group level part-worths are estimated by
dummy variable regression (Green et al., 1981). The part-worths at the individual level are then
fitted with data from the group level estimates of the full profile task by multiple regression
analysis. Several empirical studies have shown that cross-validity of hybrid models is better than
self-explicated data alone. However, full-profile models appear to be superior of hybrid models
(Green and Krieger, 1996). In newer hybrid models, part-worths from a self-explicated task are
fitted with individual level estimates from full-profile conjoint. If the number of full profiles is
12 Preprint to appear in Innovative Marketing, 2006
reasonable and an orthogonal design can be applied, individual level part-worth estimates can be
calculated for all attributes and levels. This can increase the internal validity measured on holdout
scores of hybrid models (Green and Krieger, 1996).
Adaptive conjoint analysis (ACA) also uses self-explicated data as preliminary input. Instead of
combining the data with full product profiles partial profiles are used. ACA was developed in the
beginning of the 1980’s when the technological possibility arose to perform computer-
administered interviews (Johnson, 1987). Based upon the self-explicated data, the respondent has
to evaluate a sequence of pairs of partial profiles. The method is called adaptive because the
selection of the subsequent partial profiles is adapted based on the respondent’s preference
indications. Using ACA studies with more attributes and levels can be performed than with full
profile conjoint analysis. ACA has successfully been implemented in many conjoint studies and
was the most frequently chosen method in Europe (Wittink and Burhenne, 1994) and the USA in
the 1990s (Orme, 2003). The advantage of ACA over other conjoint methods is that it is a
combined presentation and estimation computer package. With a web-front end it can be also
used for online surveying.
Using Conjoint Measurement in Pricing Studies
According to the literature, pricing studies are one of the most important applications of conjoint
analysis (e.g., Gustafsson et al., 2000, pp. 6-7). In a study on conjoint applications in the US in
the years 1981-1985 Wittink and Cattin (1989) surveyed 59 companies who carried out 1062
conjoint studies. 38% of the identified studies were pricing studies. In a similar study on the
application of conjoint analysis in the European market in the years 1986-1991 Wittink and
Burhenne (1994) surveyed 66 companies and reported a total of 956 conjoint studies. Out of
these 46% were pricing studies. Baier (1999) carried out a smaller study in the German market. 8
companies were interviewed and 382 conjoint studies were identified, of which 62% were pricing
studies. Hartmann and Sattler (2002a,b) surveyed 54 marketing research institutes in Germany,
Austria, and Switzerland in the year 2001. These institutes performed a total of 304 studies
regarding preference measurement. 121 studies were documented in greater detail by the
marketing research institutes, showing that 48% were pricing studies.
Not only overview articles on the usage of conjoint analysis show its importance in pricing
research. Also, publications of the application of conjoint analysis in scientific journals illustrate
its importance. In a broad review Voeth (1999) summarizes the publications on conjoint analysis
in German between the years 1976-1998. Most of the identified 150 studies were published in the
1990s of which 31 studies explicitly focused on pricing.
Some of the best examples from the literature regarding pricing studies performed by conjoint
analysis in important German and English scientific journals are Currim et al. (1981), Mahajan et
al. (1982), Goldberg et al. (1984), Green and Krieger (1990), Hanson and Martin (1990),
Balderjahn (1991), Green and Krieger (1992), Eppen et al. (1991), Venkatesh and Mahajan
(1993), Balderjahn (1994), Eggenberger and Christof (1996), and Green et al. (1997). As can be
seen from practical applications and journal publications, pricing studies are an important field of
conjoint analysis. Apparently, conjoint analysis is a method which is well suited for pricing
studies (cf. Diller, 2000, p. 202).
The commonly used approach in pricing studies by conjoint analysis is incorporating the price in
the study as an additional attribute (e.g., Green and Srinivasan, 1990, Orme, 2001). The levels of
the attribute price are then assigned part-worths like the other attributes. For different part-worths
13 Preprint to appear in Innovative Marketing, 2006
of price points, interpolation heuristics are applied to obtain the part-worth for any intermediate
level which was not part of the conjoint analysis.
Kohli and Mahajan (1991) published the first article explicitly focusing on WTP estimation
within a conjoint analytical framework. The authors define WTP as follows: “We assume that a
consumer’s reservation price for a new product is determined by his or her (estimated) utility for
the product in relationship to the price and utility for his or her most preferred product among all
product offerings in his or her evoked set. An evoked set consists of all products that are
accessible to an individual consumer and perceived as potential consumption alternatives, of
which one, and only one, can and will be purchased. Formally, Kohli and Mahajan model WTP
estimation based on conjoint data as follows:
.)(
*
|~
ε
++
iipit
upuu
In this notation individual i prefers product t over some status quo product with utility u
i
. The
status quo product has the highest estimated utility of any currently available product in
consumer i’s evoked set. Product t is preferred if the sum of the part-worths of the non-price
attributes u
it|~p
and the part-worth due to price u
i
(p) is higher than the utility of the status quo
product plus some arbitrarily small number .
In the remainder of their work the authors assume that the WTP observations are drawn from a
normal distribution. They estimate this distribution and describe shares of preference for different
products at different prices based upon the distribution’s density function.
In an empirical application study, Kohli and Mahajan tested different apartment concepts among
MBA students. The preference structure for the concepts is estimated for each individual via
conjoint analysis. Price is included as an attribute and modelled as a continuous linear variable in
the multi-attribute preference function. A status quo apartment is assumed to be given at a fixed
price. This status quo apartment is the same for every respondent. Against this apartment the
prices for all other concepts at which the respondents would switch away are calculated. These
prices are representing the participants’ WTPs.
Note that the prerequisite for a correct forecast of a respondents WTP is that he or she perceives
the status quo apartment the best alternative in his or her evoked set. Furthermore, every
respondent must be willing to purchase the status quo product at the current price. Notice this
approach rests on the critical assumption that all respondents are willing to purchase the status
quo product, otherwise, conjoint data only poorly reflect realistic market behaviour (cf.
Balderjahn, 1993, Weiber and Rosendahl, 1997).
Another problem is respondents’ heterogeneity with respect to status quo products: Different
participants might consider different products their best alternative. Hence, using the same status
quo product for all participants might not yield correct WTP predictions. Researchers reacted to
this problem by letting every respondent indicate his or her own status quo product. Voeth and
Hahn (1998) let the subjects arrange the product profiles in a rank order and then insert a so-
called limit-card. Up to the position of the card, the respondent would be willing to purchase the
product profile at the indicated price, below the card the respondent would not purchase the
product. The product at the position of the limit-card is used as the status quo product. This
approach was picked up by other researchers. A somewhat different approach was used by Sattler
and Nitschke (2003) where the subjects order different product profiles with prices and then
indicate those which they would actually purchase. The stimulus with the lowest overall part-
worth is then used as the status quo product.
14 Preprint to appear in Innovative Marketing, 2006
A crucial design question for conjoint interviews with price is to set appropriate price levels.
Usually the attribute price is set so that it covers the range of usual market prices. This is
problematic for respondents whose WTP is far above or below the average market price. These
respondents rate or rank a large number of profiles with prices assigned that are far displaced
from their WTP. It can happen that the relevant product stimuli are only presented in very few
conjoint questions and therefore only few relevant data-points are elicited. Furthermore, using a
fixed range of market prices does not allow the estimation of market expansion or contraction
effects, if the prices for the products were set outside of the range of usual market prices.
Jedidi and Zhang (2002) question that the assumption of unconditional category purchase holds
when new products are introduced that attract consumers, who did not buy in that category
before. Therefore, the authors depart from the approach by Kohli and Mahajan (1991) by
estimating an origin of zero utility for each respondent individually. With this, the authors
dismiss the assumption that every respondent would accept a status quo product and allow the
estimation of market expansion and contraction effects caused by consumers who switch to and
from the category. In their approach a consumer’s WTP is the price at which the consumer is
indifferent between buying and not buying the product, given all consumption alternatives
available to the consumer (e.g., products in other categories). Formally, Jedidi and Zhang present
the condition for WTP r
i
(P) that some individual i has for some product P as:
.0,0
)(
,
y
i
i
i
y
i
ii
i
p
m
U
p
Prm
PU
As in economic theory each individual has a utility function Ui(P, y
i
) for the consumption of the
product P and the consumption of some amount of the composite product y
i
. The amount of the
composite product consumed by the individual is expressed in terms of a budget constraint
pypm
i
y
ii
+=
and the price p for product P. Based upon the assumption that WTP solely
depends on the alternative purchase opportunities, Jedidi and Zhang interpolate between the
levels of the attributes in the conjoint study and extrapolate to zero. The derived utility for the
absence of the attribute is assigned zero monetary value. This is assumed to be the origin of
utility for that attribute. Offset to this part-worth, the exchange rate between utility and price is
used to calculate the WTPs.
Hard Drive Part-Worth
100 GB 10
200 GB 15
Price Part-Worth
500 10
200 50
Table 1 Example of attributes price and hard drive with assigned levels and estimated part-worths.
15 Preprint to appear in Innovative Marketing, 2006
To illustrate the approach, we give a brief example for the two attributes price and size of hard
drive shown in Table 1. The part-worths of the attribute levels are estimated by conjoint analysis.
The linear extrapolation of the size of hard drive to 0 GB leads to 5 utility units (calculated from
the numbers in the upper part of Table 1). Since 5 utility units is equivalent to the absence of hard
drive, the authors derive that the respondent would pay no money for 5 utility units. The
exchange rate between utility and price in this example calculates to 10 € per utility unit ((500-
100 ) / (50 -10) = 10 ). Using the exchange rate between utility and price and subtracting the
utility for which the respondent would pay no money the authors calculate an individual’s WTP
for the levels of hard disc. For 100 GB hard drive the WTP of the individual would be 50
(calculated by (10 -5)·10 € = 50 ).
Jedidi and Zhang present an empirical study carried out amongst MBA students at a major U.S.
university. The students participated in a conjoint analysis for notebooks consisting of the
attributes price, brand, memory, speed, and hard drive with either two or three levels. The data
were collected with a traditional conjoint measurement method. The authors simulate market
shares and profits at different prices for the new notebook and show that their estimation yield
different results compared to a conventional approach with a status quo product. In the
conventional approach the market is simply divided between the offered products. In order to test
the validity of their model Jedidi and Zhang compute the correlation between self-stated WTPs
and estimated WTPs for the new product and find a positive correlation between the two values
across the consumer population.
Jediddi and Zhang’s approach does not require status quo products and is able to estimate the
share of non-buyers which reflects real market behavior with market expansion and contraction
phenomena more adequately. This is accomplished by using an augmented set of conjoint data
with assumed choices via interpolation heuristics. However, since actual choice data are not
explicitly collected, validation of the interpolation results remains an open issue.
Limitations of Existing Conjoint-Based Approaches
In all approaches presented in the previous section price was incorporated into the conjoint
designs as an additional attribute in order to provide WTP estimates. This practice, however, has
some severe shortcomings which will be discussed in the remainder of this chapter. By doing so,
we distinguish the following three types of problems encountered with the inclusion of price
attributes in conjoint experiments:
1. Theoretical problem: By treating price as an attribute in a conjoint study, part-worth
utilities are estimated for the presented price levels. By definition price does not have
a utility, rather it reflects an exchange rate between different utility scales, implying,
the price of goods do not influence the goods’ utility. Rather, it denotes how much of
alternative consumption (with the associated utility) has to be given up to consume the
good.
2. Practical problem: The occurrence of interactions between price and other attributes
are likely to occur in a conjoint study. When this happens the additive-compensatory
model is violated. Furthermore, crucial effects are the price effect, the range effect,
and the number-of-levels effect that occur when price is included as an attribute. Last
but not least, problems calculating the interpolation heuristics between utility and
price can occur, when more than two price levels are used.
16 Preprint to appear in Innovative Marketing, 2006
3. Estimation problem: Traditional conjoint analysis does not incorporate a decision
rule. That is, only preference structure is estimated and not choices for or against
different products. When the objective is to estimate WTP, researchers need choice
information. This information is usually added to the data by assuming or explicitly
asking the respondents for a status quo product with an associated WTP. Estimating
the WTPs for all products of the conjoint study based on one data-point (the status
quo product) only might not be sufficient for accurate estimations.
Let us first address the theoretical problems raised: In neoclassical economic theory of
consumer behavior price is treated as an exogenous variable. It bears no more information to a
customer besides how many units of different goods he or she may consume before the budget is
exhausted. This treatment of price permits the construction of choice behavior and indifference
curves as well as demand functions. In these neoclassical approaches price enters any model
solely through a budget constraint (cf. Varian 2003, chap. 5-6).
1
By assigning part-worth utilities to the price levels of the study, price is treated fundamentally
different than in neoclassical economic theory. As emphasized by Rao and Gautschi (1982) this
approach is data-based rather than theory-based. The conjoint analyst would simply treat price as
another attribute in the multi-attribute utility function because it makes estimation of response
behavior simple. Srinivasan (1982) responded to the argumentation of Rao and Gautschi (1982)
that the distinct treatment of price is not so great after all and should not cause theoretical
problems. If a consumer’s WTP for a product solely depends on the consumer’s budget constraint
and the consumption alternatives given elsewhere, the utility function of two goods, the good of
interest and the composite good, can be re-arranged into a utility function of only the good of
interest with price included as an attribute that attaches value to the product (cf. Ratchford, 1979
and Srinivasan, 1982).
However, we believe that in real purchase situations WTP does not only depend on the composite
product and a budget constraint but also on alternative product offerings, so-called reference
products. Therefore, the theoretical problem of including price as an attribute in conjoint analysis
still remains unresolved.
Practical problems: When the participant of a conjoint analysis with randomly assigned profiles
is presented a ranking or rating task, it is possible that some of the presented profiles have an
unfairly high price to some respondents or appear to be an extremely good deal. When this
happens, the respondents fail to compare the current profile with other profiles and this would
lead to a non consistent ranking or rating. These profiles would be rated artificially low or high. If
this happens, the additive-compensatory model is violated and interactions between certain price
levels with other attribute levels have occurred. This is likely to happen in a conjoint study with
1
Nevertheless, besides an allocative function of the budget constraint, price can bear information for the customer.
Price can provide evidence of quality, as was first documented by Scitovsky (1945). On this topic a great number of
empirical studies have been carried out, an overview is given in Rao (1993) and recent publications include Sattler
and Rao (1997) and Brooks et al. (2000). These studies indicate a mixed relationship between quality and price. A
very low price might be perceived as an indicator of low quality, and vice versa a very high price might be perceived
as an indicator for high quality. There also exist goods for which the allocative effect of price is reversed. For these
products preference for buying increases as a direct function of price. They are called Veblen goods, examples are
expensive wines and perfumes. Decreasing their prices decreases consumers’ preferences for buying them because
they are no longer perceived as exclusive or high status products (Leibenstein, 1950). However, in this work we
focus on “normal” products and do not wish to separate the allocational and informational effect of price. Instead, we
are generally interested in WTP estimation which contains both effects.
17 Preprint to appear in Innovative Marketing, 2006
price as an attribute (Weiber and Rosendahl, 1997). Attempts to relax the restrictive assumptions
of compensatory decision models are numerous in the psychometric literaterature (cf., e.g.,
Tversky, 1972, Tversky, et al. 1988) as well as in marketing. A general discussion as well as
advanced more flexible modeling approaches towards this direction are provided by Gilbride and
Allenby (2004), Jedidi and Kohli (2004), Yee et al. (2005).
The price effect occurs when the number of attributes becomes large. In conjoint studies with
many attributes of which one is price the importance of price tends to be artificially understated,
and the degree of understatement increases as the number of attributes increases (Orme 2003).
Practitioners attempt to overcome this problem by calibrating the importance of price in a post
process by re-scaling the importance of the attribute price.
The range effect is a well studied effect in psychophysics (e.g., Parducci, 1974). If the physical
range of attribute levels in an experiment is altered, the range of the stimuli responses is also
altered (Verlegh et al., 2002). This is important for price, because price does not have a natural
upper or lower limit. In a traditional price study using conjoint analysis determining the range of
acceptable prices is crucial. Choosing a price range that is very wide, covering all possible prices
will result in a larger importance of the attribute price, than if a narrower range was chosen.
Another important effect in conjoint analysis is the number-of-levels effect. This effect in has
been studied by many authors (e.g., Wittink et al,. 1989, Steenkamp and Wittnik, 1994).
Increasing the number of levels of an attribute increases the attribute’s importance significantly
(Green and Srinivasan, 1990). The number-of-levels effect is even stronger than the range effect
(Verlegh et al., 2002). Again, since price does not have a natural number of levels, the conjoint
analyst must decide how many levels to use. In many cases, certain price levels are of special
interest to a researcher which confronts the researcher with the dilemma that inserting the
intermediate levels of interest, artificially increases the importance of price.
The price effect, the range effect, and the number-of-levels effect cannot be avoided for price,
when it is included as an attribute in a conjoint study.
Another problem occurs by using more than two price levels. The estimation procedure for
attribute level parameters is usually not constrained to support a natural ordering of the levels.
But price has a natural ordering: A higher price level should have a lower part-worth than a lower
price level. In unconstrained estimation it is possible and sometimes expected that the natural
ordering of part-worths of price levels contains reversals (Orme, 2001). Practitioners get around
these problems by using as few levels as possible (Orme, 2002). Another problem that occurs if
than two price levels are used is to decide how interpolation heuristics should be applied to
estimate an exchange rate between utility and price. Possible heuristics are piecewise linear
interpolation between the price levels, least squares fitting, or using the highest and lowest price
level for linear interpolation only. Depending on the interpolation heuristics different WTPs are
estimated for the products of the conjoint study.
Estimation problems: Traditional conjoint analysis does not include a way to estimate choice
behavior. In particular, the respondent cannot indicate that he or she would refuse to accept a
product at a certain price level, even though he or she would prefer that product offering over
others which are even less desirable. Indication of refusal to accept is only explicitly present in
discrete choice analysis which is discussed in the next section. Forecasting choice behavior based
on conjoint data can only hypothesize that a respondent would actually purchase some of the
product stimuli at certain price levels. Calculating the WTPs of an individual is usually done by
offsetting the products prices to a status quo product. However, the estimation of the exchange
18 Preprint to appear in Innovative Marketing, 2006
rate between utility and price for all product stimuli and the assumption that all possible prices
only rely on just one indicated status quo product and its price seems not very robust. We believe
that it would be better to elicit more data points in order to fit the exchange rate between utility
and price.
Discrete Choice Analysis
In discrete choice analysis the respondents choose between alternative product profiles (Ben-
Akiva and Lerman, 1985 and McFadden, 1980, 1986). In a conjoint measurement context this is
also referred to as choice-based conjoint analysis (cf. Louviere and Woodworth, 1983). The
connection to conjoint analysis lies in the ability of both methods to decompose products into
attribute levels and estimate part-worths for these levels. The difference lies in the underlying
estimation methods (for a detailed discussion see, e.g. Louviere et al., 2000). Albeit, as further
outlined below, recent developments allow individual level estimates as well, conventional
discrete choice analysis tries to estimate a latent utility structure on the aggregate or segment
level (see, e.g. mixture models as outlined in Wedel and Kamakura, 2000).
The utility structure is estimated based on a choice set, which is typically (but not necessarily)
fixed across all respondents. Every choice can be fully described in terms of its attributes. The
respondents are presented different alternatives and indicate which one they would actually
choose. Often the respondents are provided a no-choice alternative, to indicate that they would
not choose any of the presented product profiles (cf. DeSarbo et al., 1995 and Haaijer et al.,
2001).
A latent preference for every choice in the evoked set is assumed to exist at the aggregate (or
segment) level. The evoked set refers to the set of possible products or brands the respondent is
currently considering in the decision process. This latent preference is estimated based on choices
between different product profiles the participants make during the analysis. For every participant
the utility value for a choice is modeled consisting of a deterministic component, that represents
the latent preference structure at the aggregate level, and a random component. The random
component is due to fluctuations in perceptions, attitudes, or other unmeasured factors
(McFadden, 1986). Depending on whether the random component is normally or logarithmically
distributed, the model is referred to as a probit-model or a logit-model. The more common model
is the logit model.
Based upon random utility theory, the utility that an individual i assigns to some alternative can
be described as
iii
VU
ε
+=
.
In this notation U
i
is the unobservable, but true utility of alternative i. V
i
is the observable or
systematic component of utility and
i
is the random component.
Price is included as an attribute of the product profiles and the levels cover the range of the
possible and meaningful prices. The probability for the choice for a specific alternative i from a
specific choice set can be described by the multinomial logit model
=
Cj
jiC
VEXPVEXPiP ).(/)()(
19 Preprint to appear in Innovative Marketing, 2006
“In this model, C = {1, 2, ..., M} denotes a set of available alternatives, indexed from 1 to M, and
P is the probability that an individual when presented with this set will choose alternative i
(McFadden, 1986). Note that V does not depend on the individual. This parameter describes the
latent preference structure of the population. The unknown parameters V
j
for all alternatives j
C are typically estimated from the data by the maximum likelihood procedure. The probability
for a product P
C
is used as a market-forecast and can be viewed as the potential market share.
Since part-worths for different prices are estimated, a change in price can be expressed in terms
of change in utility and exchange rate between utility and price can be calculated. Given this
exchange rate, the WTP for any product profile relative to the most preferred choice in an
individual’s evoked set can be calculated. An example of an empirical study using this approach
can be found in Balderjahn (1991). A comparison between ratings-based conjoint analysis and
discrete choice analysis can be found in Elrod and Louviere (1992). The authors find little
difference between the two methods with respect to predictive validity on holdouts.
From the estimation procedure described above we see that discrete choice modeling aims at
estimating preference structure at the aggregate level. In this approach it is not possible to
directly estimate part-worths at the individual level because usually too few data points are
elicited for each respondent. This is due to the fact that the observation of a choice out of an
evoked set only contains information about the chosen product and not about the remaining
products. This is different in conjoint analysis in which a ranking or rating of all products is
provided by the respondent.
However, recent improvements of powerful Markov Chain Monte Carlo simulation
methodologies have been shown to successfully alleviate the estimation problems of individual
level part-worths (and hence also WTP) in a number of discrete choice type studies (cf., e.g.,
Allenby and Ginter, 1995; Lenk et al., 1996).
Comparison of Methods
Many authors have compared competing approaches to WTP measurement. A brief summary of
their findings is reported in the following. An early comparison between direct surveys eliciting
WTP, conjoint analysis using ranking, and conjoint analysis using rating was performed by
Kalish and Nelson (1991). The experiment was conducted among undergraduate and first year
graduate students of different business classes. The authors tested the three approaches in terms
of their predictive validity on holdout products. The products of the experiment were airline
tickets described by the non-price attributes service level, seating room, and non-stop. In the
direct survey the respondents were asked to state their WTP for different product configurations.
WTP was explained to the students as the amount of money that would make them indifferent
between purchasing the ticket and keeping the money. For the two conjoint approaches prices
covering the usual range of typical prices in the market were used. For the ranking the students
were asked to bring the products into a preference order, for the rating the students were asked to
distribute a number of rating points over the presented products. The main goal of Kalish and
Nelson’s experiment was to test for internal validity by predicting holdout products. At the end of
every survey the participants were presented four product profiles (so-called holdouts) with
assigned prices and were asked to indicate their preferred choice. The predicted choices derived
from the data of the three surveys were compared to the actual choices of the respondents. The
predictive validity of the conjoint models based on rankings as well as the model based on ratings
clearly outperform the model fit from the directly elicited WTPs. 62% of the first choices were
correctly predicted in the two conjoint approaches compared to only 46% for the direct survey.
20 Preprint to appear in Innovative Marketing, 2006
The authors find that directly surveying WTP “is not as robust to respondent involvement as are
ranks or ratings (Kalish and Nelson 1991).
More recently, researchers tested different approaches to WTP estimation for external validity.
Sattler and Nitschke (2003) performed an empirical comparison of the methods direct survey,
conjoint analysis, first-price auction, and Vickrey auction. The authors elicited WTP for different
prepaid telephone cards among students. Each of the students was exposed to all four instruments
in random order. Based upon the WTP estimates derived from the four instruments, Sattler and
Nitschke systematically tested for differences. Furthermore, they tested for external validity by
requiring a sub-sample of respondents for each instrument to actually purchase the telephone card
at the indicated WTP. All approaches except the two auction mechanisms show significant
pairwise differences in estimated WTPs. The results of the study indicate that WTP is
systematically higher in hypothetical settings where the subjects do not have to make a purchase
at the end. In real settings, with a purchase at the end, the estimated WTPs are systematically
lower. These findings are consistent with other studies, for example by Harrison and Rutström
(2004) and Wertenbroch and Skiera (2002). Sattler and Nitschke discover this bias for the
methods conjoint analysis, ascending auction, and Vickrey auction. Also in the setting with the
real purchase at the end, the estimated WTPs exhibit significant pairwise differences between the
four methods. The authors draw the conclusion that one cannot decide which method mimics real
market best and thus should be advised for use.
In a different study Backhaus and Brzoska (2004) used a Vickrey auction to test external validity
of WTPs estimated by a conjoint procedure and by discrete choice analysis. The authors assume
that the Vickrey auction is feasible to elicit true product valuations and therefore can be used to
test hypothetical procedures for external validity. For the conjoint procedure a Limit-card (as
described above) was used. The object of their study is a selection of four different DVD players
for which the subjects could place bids in a Vickrey auction after completing one of the two
interviews. Backhaus and Brzoska constructed a price-response curve for each player from the
observed bidding data as well as from the data elicited by conjoint analysis and by discrete choice
analysis. A comparison showed that the two hypothetical procedures substantially overestimated
the WTP for the participants of the experiment. At the aggregate level the overestimation by the
conjoint approach was smaller than the overestimation by discrete choice analysis. However, at
the individual level underestimations of WTPs also occurred which lessens the overestimation at
the aggregate level. In a recent comparative study of a broad range of alternative approaches to
WTP estimation conducted by Völckner (2005) significant and substantial differences between
the derived WTP are reported depending on whether respondents had to pay the stated prices or
not (revealed versus stated preferences in our classification system).
Conclusions and Managerial Implications
In the present paper different methods to estimate consumers’ WTP have been discussed. The
methods were classified into four groups: Analysis of market data, experiments, direct surveys,
and indirect surveys. As discussed, all methods have specific theoretical as well as practical
advantages and drawbacks, which we summarize in Table 2.
Market data represent customers’ purchase behavior. Depending on whether the data are already
available and on the size of the data set this can be a cost effective and time efficient method to
estimate consumer’s WTP. Since WTP estimates are derived from actual demand data, they are
generally very reliable and reflect highly external valid results. In many practical situations,
however, usage of market data is not appropriate for WTP estimation. In particular, for new or
21 Preprint to appear in Innovative Marketing, 2006
hypothetical products that are not yet available in the market there is no market data available.
Equally, market prices frequently do not contain sufficient price variations to estimate demand at
different price levels.
Indirect Surveys
Market
data
Experiments
Direct
Surveys
Conjoint
Analysis
Discrete
Choice
Analysis
(CBC)
Cost effective
+/- -- ++ + +
Time efficient +/- -- ++ + +
Flexibility to include new
price/product combinations
-- ++
+/-
++ ++
Validity of estimations
++ +/- --
+ +
Real purchase behavior ++ +/- -- -- --
Observed choice behavior
++ + --
-- +
Individual level estimations
+/- +/- ++
++ +
+ (++) = (strong) advantage
- (--) = (strong) disadvantage
+/- = no clear advantage or disadvantage
(depending on data-collection and/or estimation method)
Table 2 Comparative evaluation of competing methods for measuring willingness-to-pay.
By using experiments the above mentioned problems encountered with new products or
insufficient price variations can be overcome. In this approach, WTP for different products is also
estimated by observing purchase behavior, but in an experiment the products and prices are easily
adaptable such that the participants are opposed with the necessary price variations. Depending
on the setup, the participants are more or less aware that they are participating in an experiment
which might result in biased estimates and loss of external validity (cf. Völckner, 2006). Practical
disadvantages related to experiments are the associated expenditures and time needed which
makes them less suitable for many practical application contexts.
In general, surveying techniques will be the preferred methodological approach when the
manager is facing monetary and/or time constraints. This can be the case when consumer
reactions need to be collected as repeated measurements or the results are required to be available
quickly. Surveys are also very flexible when product features need to be varied and when a larger
set of possible prices need to be tested. In surveys the respondents state their choice or desire for
22 Preprint to appear in Innovative Marketing, 2006
a number of products. Since, in general, real purchase behavior remains unobserved WTP
estimates derived from survey data can be typically considered to exhibit a lower degree of
validity as compared to, e.g., WTP estimated from market data. Direct surveys require the
respondents to (directly) state how much he or she is willing to pay for a specific product or
bundle of attributes. Naturally, this approach has a number of possible biases. For example, it is
often difficult to state a WTP for an unfamiliar product. WTP also tend to be sometimes
overstated because of prestige effects or understated due to consumer collaborations effects to
keep the prices low.
In many real-world applications, indirect survey approaches turn out to be the method of choice
for WTP estimation, because they usually exhibit both higher internal and external validity.
Using indirect approaches the respondents are confronted with a number of different products (or
attribute combinations) with assigned prices and have to choose the most preferred one or are
involved in a task to rank or rate the offered combinations of products and prices. Based upon the
choices the respondents make (or the applied rank order) WTPs for the different products can be
estimated by statistical techniques.
Among indirect surveys, two groups of approaches, namely conjoint analysis and discrete choice
analysis (also referred to as choice based conjoint), are available for marketing analysts to
estimate WTP. Most of the available methods for conjoint analysis typically are capable to
estimate WTP for the respondents at the individual level based on every respondents data.
Generally, WTP estimation at the individual level is particularly important if the price-sensitivity
in the market under study is assumed to be heterogeneous. In discrete choice analysis WTP is
traditionally estimated at the segment or sample level. Notice, however, that with the diffusion of
advanced empirical Bayesian estimation techniques, individual level estimates becomes feasible
also in a choice-based conjoint context. Using commonly available statistical software, WTP
estimation based on data collected within a discrete choice task currently remains to be more
expertise demanding and therefore also more time-consuming than estimations based on ordinary
conjoint data.
In conjoint analysis, individual level WTP is estimated based on each respondent’s data only. But
in contrast to the no-purchase option in discrete choice analysis, in the classical conjoint analysis
approach the respondent typically is not asked whether he or she would actually buy a product.
With respect to the presentation technique to the respondent, this is regarded as the main
disadvantage of conjoint analysis compared to discrete choice analysis. Letting the respondent
choose rather than rate or rank mimics real purchase behavior more closely.
To estimate product choice probabilities at different prices based on conjoint data marketers
usually assume the existence of a (preferred) status quo product. Furthermore, the respondents of
the interview are a priori assumed to buy this product. The WTP for a competing product is then
estimated as the price at which the respondent would switch away from the status quo product.
With this set of assumptions, WTP cannot be estimated for customers who would actually not
buy the status quo product in the first place or have a different (unknown) status quo product. To
circumvent this problem, respondents can be allowed to select their individual status quo product
themselves.
The gray area in Table 2 (above) shows that the two indirect surveying methods have
complementary strengths. In discrete choice analysis, product choice probabilities are estimated
the aggregate level and with HB individual level choices can be regained. In conjoint analysis,
the preference structure is estimated for each respondent individually but choice behavior is not
elicited and price only enters as an additional attribute.
23 Preprint to appear in Innovative Marketing, 2006
A modified approach that combines the relative strengths of both methods in a two-step interview
approach can proceed as follows:
1. A regular conjoint analysis using non-price attributes only is performed to estimate the
respondent's individual utility structure. This avoids the above discussed issues with using
price as an attribute while exploiting the methodological strength of easily estimating
individual utility structures.
2. WTP for product profiles is estimated in a choice-based interview scene (including a no-
purchase option). The respondent is presented a sequence of dynamically selected product
profiles with associated prices based on the previously determined utility structure. With a
suitable search algorithm, one can find several (at least 2) points in the utility-price space
where the respondent is indifferent between buying and not buying the presented product.
Based upon these data-points, a model can be estimated (for example a simple linear
model by least-squares fitting) that maps the utility of each product profile on a price
scale, which represents the individual’s WTP for the product.
This new hybrid approach eliminates the shortcomings while combining the strengths of the
currently mostly used conjoint and choice based methods. A first implementation of such a
hybrid approach is discussed in Breidert et al. (2005).
Our previous discussion revealed that, in general, each method has its specific merits and
limitations. Choosing a suitable method depends on the managerial task underlying the
estimation of WTP and is influenced by both conceptual considerations (e.g., if individual
estimates are required or not) and practical restrictions (e.g., time and budget availability). With a
particular emphasis on conjoint-based methods this article provides a thorough review of
available approaches for measuring WTP, evaluates their strengths and limitations and therefore
provides marketing managers with a basis for selecting an appropriate method.
References
1. Abrams J. A New Method for Testing Pricing Decisions // Journal of Marketing, 1964.
– N
o
28(1). – 6-9.
2. Addelmann S. Orthogonal Main-Effect Plans for Asymmetrical Factorial Experiments
// Technometrics, 1962. – N
o
4(1). – pp. 21-46.
3. Agarwal M.K. How Many Pairs Should We Use in Adaptive Conjoint Analysis? An
Empirical Analysis // AMA Winter Educators’ Conference Proceedings, 1989. pp. 7-
11.
4. Ailawadi K.L., Lehmann D.R., Neslin S.A. Revenue Premium as an Outcome Measure
of Brand Equity // Journal of Marketing, 2003. – N
o
67 (October). – pp. 1-17.
5. Allenby G. M., Ginter J. L. Using Extremes