Measuring Price Elasticity

Measuring Price Elasticity Researchers today have a choice of methodologies. Here's how to make the right trade-offs. By Bashir A. Datoo For years, m...
Author: Felicia Tyler
1 downloads 3 Views 2MB Size
Measuring Price Elasticity Researchers today have a choice of methodologies. Here's how to make the right trade-offs. By Bashir A. Datoo

For years, marketing researchers have been trying to estimate market share for products and services by using a research design that permits calculation of price elasticities. Price elasticity methodologies based on experimental designs have evolved rapidly in the past couple of decades. The author focuses on the salient phases of that evolution and, by pointing out the key limitations of each generation, highlights the improvements that distinguish the next.


ne particular class of price elasticity measurement methodologies—distinguished by the use of experimental designs that require collection of primary data through survey research—has been used by researchers for decades to predict market share at different price points. Such designs involve exposing survey respondents to a series of future scenarios in which the priee of products or services is varied systematically, and (hen asking these respondents to project their behavior under eaeh scenario. On the basis of respondent.^' preferenees or ehoiees. researchers can estimate market share—or demand—for brands included in the design and calculate priee elasticities. However, it is important that researchers understand the evolution of this class of methodologies so they can determine which technique to employ under diiferent market conditions.


Vol. 6 No.2



Conjoint Scaling Prior to Ihe introduction of trade-off techniques, market researchers relied on the use of conventional rating scales to measure the role of different faetors. ineluding price, in brand selection decision making. The advent of full-profile conjoint scaling in the 1970s revolutionized the measurement procedure and, in the process, gave researchers the capability to do market simulations and estimate the market share of brands under different "what i f scenarios. Conjoint scaling offered two key benefits over the rating-seale methodology: • The measurement focus .shifted from individual factors lo product profiles; that is. instead of evaluating one factor at a time, respondents now evaluated one product profile at a time (a

profile being a "bundle" of factors, each with a set of descriptors). • Concurrently, the tneasurement proeess shifted from attitudes to preferences; that is. instead of indicating the extent of importance of each factor, respondents now indicated the degtee of their preference for eaeh product profile. Exhibit I shows a hypothetieal profile of an agricultural herbicide. The factors used to describe the herbicide remain the same from profile to profile, only the descriptors are systematically varied in accordance with an experimental design. Respondents are called upon to make trade-offs among the factors as described, for instance, the extent to which ''unacceptable" control of some is fine in return for "fair" to "good" control of some other grasses.

Exhibit 1 Sample conjoint product profile Profile #6 Mode of application:


Crop injury risk:


Soil residual activity;

Up to 3 weeks

Broadcast cost;

$14 per acre

Good control for:

Foxtail, quackgrass

Fair control for:

Shattercane, velvetleaf, pigweed

Unacceptable control for;

Volunteer corn, cocklebur, johnsongrass, morningglory

Exhibit 2

Two key outcomes of conjoint sealing both relate to the issue of pricing. First, in quantifying the contribution of each factor to brand selection decision making, marketers ean determine the overall importance of price. Exhibit 2 shows that price accounts for 15% of the decision proeess in this case, being about half as important as control of grasses (28%) but almost twice as important as crop injury risk (8%). Second, in t"neasuring the sensitivity to each descriptor of a factor, the price elasticity function of the product category emerges. Exhibit 3 shows that the same percentage price ehange results in tnuch greater loss in preference with a price increase (a difference of -.40 utile) than a gain in preference with a price deerease (a difference of -I-.25 utile). From the perspective of pricing research, although conjoint represents a major advance over previous methodologies, it suffers from two major limitations: • Cotijoint typically understates the itnportance of price in brand selection. This is particularly so when brand is not included as one of the trade-off faetors (as is often the ease), so respondents tend to use price as a proxy for quality, thereby dampening price elasticity. • The technique yields a generic priee elasticity function for the product category as a whole. The same curve is assumed to apply to all brands within the category, regardless of their pereeived performance and/or imagery. SECOND GENERATION Brand/Price Trade-off To overcome the limitations of conjoint scaling

Contribution of price to brand selection Broadcast cost per acre Crop injury risk Soil residual activity Mode of application Control for 4 broadleaves Control for 5 grasses

Exhibit 3 Product category price elasticity 1.0 (+.25)

0.8 0.6 0.4 0.2 0.0





(Current) Broadcast cost per acfe


Vol. 6 No. 2 3 1

Exhibit 5

Exhibit 4

Brand's self-elasticities: typical pricing function

Sample brand/price trade-ojf scenario Scenario #9

Brand A Broadcast cost per acre

50% - Market share

$15.00 $15.00 $16,25 $15,00 $17,50 $17.50 $ 9.00 $11.25 $33.75 $ 6,00 $ 9,00

Basagran Blazer Dual Fusilade Lasso Lorox Poast Prowl Roundup Sencor Treflan



38,1 30% 20%


10% 0



$20 (Current) Broadcast cost per acre

Exhibit 6 Brand's self-elasticities: other pricing functions Brand D


$20 (C)


Broadcast cost per acre

Brand C

Brand B Market share

Market s nare 50%

50% —^^(+1C 5) 47,1



30% -







^ "^

30% -


20% -

20% (-4,1)


10% — 1


$10 (C)


$10 (C) Broadcast cost per acre

C = Current price

Vol. 6 No. 2




Broadcast cost per acre


(-2.9) 13.3


• 10.4


for pricing research, a few market research companies and/or practitioners developed an alternative approach that can be characterized as a direct brand/price trade-off. Total Research Corp. was one of them, introducing its Price Elasticity Measurement System (PEMS*) in 1982. The underpinning of this methodology is the same experimental design u.sed in conjoint scaling. However, in the scenarios presented to respondents, brands now replace factors, and price points replace descriptors, as ean be seen in Exhibit 4. This represents a shift from profiling a product to profiling a market, in which respondents* task is no longer to indicate which product they would prefer but. rather, whieh one they would buy. Hence, there is a shift in the method of evaluation from preference to choice. The brand/price trade-off technique offers two additional benefits over eonjoint sealing: • It permits the development of brand specific self-elasticity effects, not just one generic function for the entire product category. The effect of changes in the price of a eompany^ brands can have a differential impact on the demand of those brands. • It measures brands" cross-elasticity effects directly, rather than just inferring them from differences in the end results of market simulations. The beneficiaries of a given brand's price increase may be somewhat different from the losers of that brand's priee decrease. Self- and cross-effects are developed for every brand whose price is manipulated in the design. Exhibit 5 shows a typical pricing function; it is elbowshaped beeause brands typieally gain more share (+15.7 share points) than they lose (-5.2 share points) with the .same propoitional change in price (25%). Exhibit 6 shows several other functions that may be observed within the same produet category: Brand D shows a reversed elbow curve characterizing a brand that basically has saturated the market; Brand B shows a linear function that may be applicable to eommodity-type products; and Brand C shows a counter-intuitive curve typifying high-status brands. Exhibit 7 illustrates the effects of one price change, a 25% inerease. Base shares at the bottom of the exhibit represent market shares of the brands at their current prices (e.g.. Brand A has a share of 22.4%). The numbers in the diagonal of the matrix indicate the impact on the share of a brand in the wake of a priee inerease (Brand A loses 5.2 share points, resulting in a new market share of 17.2%). The figures off the diagonal show the beneficiaries of a price increase (for example.

Exhibit 7 Brands' cross elasticities Effect of 25%

Brand A

Effect on: Brand C Brand B

Brand D

price increase for:

Brand A



+ 1.6


Brand B





Brand C





Brand D









Basic share

Brand D gains the most, 3.2 share points, resulting in a new share of 30.9%). Notiee that the relationship between Brands A and C is asymmetrical: Brand A loses to Brand C proportionately more than Brand C loses to Brand A when they respectively raise their prices. Armed with self- and cross-effeets, it now becomes possible to simulate how market shares of the brands would be reconfigured in the event that several brands simultaneously or sequentially alter their prices. Sueh ''what it^' seenarios can be run for any conceivable combination of prices that fall within the price range tested for each brand. The brand/priee trade-off technique has become extremely popular in price elasticity research, in part because it has been successfully tailored to address a wide variety of market conditions. This flexibility notwithstanding, the technique does have some constraints: • Brands are presented to respondents as '•gestalts," as an assemblage of whatever brand perceptions they bring to the task or whatever brand descriptions are provided with the task. Thus, the technique eannot directly measure the added value of a new or improved feature. • Brand set is fixed across all the scenarios, with no brand being added or deleted. Hence, the leehnique cannot readily test the effect of sequential new product introductions. THIRD GENERATION Discrete Choice Modeling The constraints mentioned above mean that the brand/price trade-off technique cannot be used in a few situations. Companies that pioneered the use of discrete ehoice modeling have pressed this technique into use in the "90s for such special situations. MARKETING RESEARCH:

Vol, 6Nc. 2 3 3

Exhibit 9

Exhibit 8

Elasticity of current vs. improved product

Sample DCM choice set Set #12 Broadcast cost per acre

Market share

30% r

Basagran New indication for control of (broadleaf)


Blazer New/improved formulation




Fusilade Current packaging





Lorox New indication for control of (grass)







(Current) Broadcast cost per acre



Discrete choice modeling involves the development of choice sets that show product choices available in the market, together with their descriptions (in terms of factors/descriptors). Thus, they represent a combination of conjoint scaling and brand/pi'ice trade-off scenarios. However, the method of evaluation remains the same as in brand/price trade-off, namely choice of, not preference for brands. Exhibit 8 illustrates one of the special situations that calls for discrete choice modeling by showing a choice set that focuses on two product improvements: Blazer is assumed to come out with a new formulation, whereas Fusilade is assumed to introduce a cuiTent packaging. To disguise our interest in these two specific changes, a couple of other features also are altered from one scenario to the next. The outeome from such an approach is shown in Exhibit 9, with findings indicating that the improved feature adds significant value to the brand. First, the function for the "improved" product is higher on the graph than that of the "current" product (a difference of 4 share points at the proposed/current price of the products). Second, with a price decrease, the curve of the "improved"' product is steeper than that of the "current"' product (gains more share): with a price increase, (he curve is gentler (loses less share). CURRENT STATUS Bashir A. Datoo is Senior Vice President and Head of ihe Strategic Research Support and Developnieni Group (SRSAD) at Total Research Corp.. Princclon. N,J,


Vol, (i No, 2

The move toward newer generation methodologies has ihu.s provided greater realism and Hexibility—from conjoint scaling (CS) through brand/price trade-off (BPT) to discrete choice MARKETING RESEARCH:

modeling (DCM). Specifically, the evolution into second generation methodology has added realism to the task that is set up for respondcnt.s, whereas the development of third generation methodology has offered greater flexibility in addressing the needs of pricing research: • In terms of realism, respondents are shown a profile of the market (BPT/DCM), as opposed to separate profiles of products (CS): they make a choice of products or services from among alternatives presented (BPT/DCM), as opposed to giving a preference for each alternative (CS). • In terms of flexibility, changes in product profiles can be described (DCM). as opposed to staying with fixed profiles (BPT): changes in choice alternatives can be introduced (DCM), as opposed to assuming a stable market (BPT). Even though conjoint scaling is still an important tool in the understanding of product and service feature trade-off.s. it is an outmoded technique for the measurement of brand-specific price elasticity. Brand/price trade-off is currently the most prefeiTed method for determining elasticities associated with current products and limited market changes: discrete choice modeling may be the preferred technique for changing products and changing markets. In other words, brand/price trade-off is applicable in a wide variety of situations and discrete choice modeling jumps in where the former is found to be wanting.EDI

Suggest Documents