P R O D U C T A T T R I B U T E S 5 Part M O D E L Product Attributes Model Preferences for Attributes of Products Kelvin J. Lancaster (1966, 1971, 1979) developed a theory of consumer choice based on preferences for characteristics (or attributes) of products. Each product is a bundle of attributes. For example, laptop computers differ in terms of processor speed, RAM, storage capacity, weight, and screen size, among other characteristics. Rather than comparing the products themselves (as in traditional indifference curve and budget constraint analysis), attribute theory assumes that individuals choose among various bundles of attributes, as represented by specific models or varieties of goods and services. Understanding why a consumer chooses a product based upon its attributes helps us to understand why some consumers have preferences for specific models, specifications, and brands. This framework allows an analysis of product differentiation and brand competition. A simple example illustrates the general approach of the model. Suppose the only important attribute of digital camera is pixel resolution. We can locate three models of digital cameras in “resolution space:” Low Resolution High Resolution Largan JamC Polaroid 2300 In "resolution" attribute space, Largan is located to the left of JamC, which is to the left of Polaroid 2300. The higher the pixel resolution, the further to the right the camera is located. Here, there is a single dimension that shows the amount of the attribute (pixel resolution, in this case) and each product can be located in this space based upon its level of the attribute. Kent Jones and Lidija Polutnik 1 The Model The Product Attributes Model explains individual choice as a process of choosing bundles of product attributes inherent in goods and services. The model assumes that consumer choice is based on maximizing utility (or the level of satisfaction received) from the product attributes subject to a budget constraint. The model is particularly useful in analyzing differentiated product markets. These markets have brands or models that are substitutes for each other and are distinguished by their makeup of a specific set of characteristics. Understanding consumer preferences for attributes that distinguish among brands can help in defining the best marketing mix for a particular brand. To illustrate the model, we will use a simple example with two product attributes. In this case, a two-dimensional graph reveals the model’s main features and links it to the traditional indifference curve and budget constraint analysis of individual choice. Figure 1 shows three products (A, B, and C), each offering specific amounts of attribute X and attribute Y. Attribute Y Figure 1 YA A B Efficiency Frontier C Attribute X XA Product Rays Each product is depicted in attribute space as a ray from the origin. The slope of each ray is determined by the ratio of attribute Y to attribute X. The rays are linear if we assume each product offers a constant ratio of attribute Y to attribute X as quantity consumed increases. The highest ratio Y/X is offered by product A, the lowest by product C. Kent Jones and Lidija Polutnik 2 The Budget Constraint and Efficiency Frontier The quantity of a product that a consumer can afford to buy will depend on the price of the product and the budget of the consumer. For a given budget constraint and set of prices for the products, the end points A, B, and C represent the limits of consumption along each product ray. The end points (or total attainable attributes) are calculated by multiplying each attribute level for a single unit of the product by the quantity of the product that the consumer can afford to buy. The line segment ABC defines the budget (or efficiency) frontier for the consumer. The frontier is called efficient because a consumer maximizes utility by choosing combinations of attributes on this frontier. The efficiency frontier shows the total amounts of attribute X and attribute Y that the consumer can afford to buy. Product A offers XA of attribute X and YA of attribute Y. Similarly, products B and C offer the attribute bundles (XB, YB) and (XC, YC) respectively. While some attributes can be measured objectively (for example, processor speed, RAM and storage capacity), it may also be useful to consider more subjective attributes. Examples are “attractiveness in design” and “comfort in keyboard use” that represent distinguishing attributes among laptops. Subjective attributes imply that the attribute content of a particular product may be determined largely by the perceptions of the individual consumer. Subjective attributes are often measured by consumer surveys. Maximizing Utility from Attributes The consumer’s choice is made by maximizing her/his utility, as defined by the consumer’s set of indifference curves, subject to the budget constraint. In the Product Attributes Model, we interpret the slope of the indifference curve at a particular point (also called the marginal rate of substitution) as the rate at which the consumer is willing to trade off units of attribute Y for an additional unit of attribute X to maintain the same level of utility. Thus the consumer’s choice is influenced by her/his preference pattern in attribute space. Consumer Equilibrium As shown in Figure 2, the consumer shows a strong preference for attribute Y and therefore chooses product A. The consumer’s preference is represented by the relatively flat indifference curves (IC, IC*, IC'). The highest indifference curve that intersects the efficiency frontier is IC*. (We will always let IC* represent the optimal indifference curve throughout our discussion of this model.) This intersection is at the end point of the ray for product A. The consumer maximizes utility at this intersection. Kent Jones and Lidija Polutnik 3 Attribute Y Figure 2 YA A B IC' IC* IC C Attribute X XA On the other hand, a strong preference for attribute X would lead her/him to choose product C. In this case, the indifference curves would be relatively steep. Exercise 1: Try drawing these indifference curves on Figure 2. The presence of differentiated products in a market can be explained as the result of the dispersion of tastes and preferences for various attributes among different segments in the population of consumers. Mixing Products Note that the consumer will spend her/his entire budget on a single product A, B, or C if the highest indifference curve just touches the respective end point. If the highest indifference curve touches a point on the line segment between two product ray end points, then the consumer would choose to split consumption between the two adjacent products. In Figure 3, the highest indifference curve that intersects the efficiency frontier is IC*. The intersection is represented on the graph as point M. However, there is no product that satisfies the attributes in the ratio represented by the point M. The consumer may be able to maximize utility by mixing products B and C. The consumer buys product B some times and product C other times. This can lead to brand switching by consumers for specific uses or occasions or just for variety. An example of mixing products would be dining out at various restaurants. Kent Jones and Lidija Polutnik 4 Attribute Y Figure 3 YA A B M IC* C Attribute X XA Product Indivisibility If the product’s consumption is indivisible (as in the case of an automobile or house), then consumption (or the consumer’s choice) is determined by the highest indifference curve that touches an end point of an attribute ray. This may occur if the price of the product is large relative to the consumer’s income and the consumer can only afford to purchase one of the products. As shown in Figure 4, the consumer cannot attain the highest indifference curve IC* if she/he can only afford to buy one of the products. The consumer will choose to buy product B (where the lower indifference curve IC intersects the efficiency frontier) and then utility will be less than optimal. Kent Jones and Lidija Polutnik 5 Attribute Y Figure 4 A YA B M C IC* IC Attribute X XA Strategic Implications The Product Attributes Model can be used to analyze a firm's strategy and its effect on consumer choice. The Product Attributes Model has important implications for the firm’s marketing mix. Market Segmentation/Target Market The Product Attributes Model can help the firm understand the market segments for a product. It can then enable the firm to identify the appropriate target market(s) for its specific brand. For example, in Figure 5, the consumer with indifference curve ICY has a relatively low marginal rate of substitution between attribute Y and attribute X. This consumer has a relatively flat indifference curve and will tend to purchase product A. Consumers with a low MRS can be called the “Attribute Y segment”. The consumer with indifference curve ICX has a relatively high marginal rate of substitution between attribute Y and attribute X. The consumer has a much steeper indifference curve. This consumer will tend to purchase product C and we can say this consumer belongs to the “Attribute X segment”. Kent Jones and Lidija Polutnik 6 Attribute Y Figure 5 YA A ICY B C ICX Attribute X XA We can use computer attributes to demonstrate differing consumer preferences and possible market segments. Using Figure 5, let “peripherals capability” be attribute X and “processing capability” be attribute Y. Additionally, let product A be a computer with rapid processing and math co-processing features; let product C be a computer designed for the easy hook-up and use of external projectors, printers, and communications; and let product B be a computer that strikes a balance between these two types of features. Suppose Jeffrey works for a biotechnology company and specializes in scientific research, with little need for presentation of his work on the computer (he therefore has a very low marginal rate of substitution of attribute Y for X). On the other hand, Samantha is a sales person who often takes part in presentations and teleconferences, but does not “crunch numbers.” (she therefore has a very high marginal rate of substitution of attribute Y for X). Exercise 2: How would you describe Jeffrey and Samantha’s indifference curves? Which computer would Jeffrey most likely buy? What about Samantha? What sort of consumer would purchase computer B? A firm that knows the tastes and preferences of consumers for the various attributes can incorporate the attributes into a product designed for a particular segment of the consumer base. This group of consumers becomes the firm’s target market for its differentiated product. Positioning and Opportunity Assessment The Product Attributes Model allows the firm to see how its product is perceived relative to its competition on critical attributes. The firm can assess whether it operates from a position of strength or if it is vulnerable to the competition. If it is vulnerable, it can determine the appropriate action to take: a new product, changes in the existing product, changes in price, or a new promotional strategy. Kent Jones and Lidija Polutnik 7 In addition the Product Attributes Model allows the firm to assess an opportunity for the new product introduction in the market. A firm may introduce a new product positioned to take advantage of an opportunity represented by a “gap” in the attribute space between rays. This “gap” indicates that existing products do not satisfy the ratio of attributes wanted by a particular segment of customers. A new product may be developed with the ratio of attributes that maximizes the utility of this newly “discovered” market segment. Exercise 3: Looking at Figure 4, what new product would you recommend? What additional information would you need to have before proceeding with a proposal for this new product introduction? Changes in Existing Product The Product Attributes Model indicates on what attribute(s) the existing product may be deficient. The firm can then make the product improvements needed to shift the product to a more favorable position in the market. The Product Attributes Model can also help predict increases in market share that will result from these product improvements. Changes in Price If the product’s price is reduced, the budget constraint will shift outward along the product ray (the consumer can afford more). If the product’s price is increased, the budget constraint will shift inward along the product ray (the consumer can afford less). Attribute Y Figure 6 A B C' IC* C IC Attribute X Look again at Figure 4. What can the firm that sells product C do to capture the consumer and make a sale? One option would be to cut price, shifting the budget constraint outward. A cut in price moves the Kent Jones and Lidija Polutnik 8 end point for Product C from C to C as shown in Figure 6. The consumer equilibrium will be at the intersection of IC* and the new efficiency frontier. The consumer now chooses to buy product C. Exercise 4: How would we expect the firm that sells product B to react to the price cut of product C? Pricing a Product Out of the Market Based upon a consumer’s perception of the attributes offered by a product, there is a maximum price he or she will pay for the product. In Figure 7, the consumer maximizes utility by mixing between products A and C. The price of product B at the point B on the product ray is too high for the consumer. The price of product B must be decreased to the point represented by B or further out on the product ray for B in order to capture the consumer. Attribute Y Figure 7 YA A B' B IC C Attribute X XA Promotion / Changes in Consumer Perceptions Advertising can affect the consumer’s choice in the context of product attributes analysis in two basic ways. First, advertising can change the consumer’s perception of the product in terms of attribute content and proportion (distance from the origin to the end point and/or slope of the product ray). In addition, advertising can also influence a consumer’s tastes for attributes (marginal rate of substitution, or shape of individual consumer’s indifference curves). Advertising may make a specific attribute important to consumers that might not have been considered previously. In this case, a new attribute dimension must be considered in the Product Attributes Model. Kent Jones and Lidija Polutnik 9 Price deals and coupons can be used to effectively lower the price for specific purchase occasions. These promotions adjust the efficiency frontier outward for the promoted product for those particular purchases. Summary The attribute approach to consumer behavior provides us with many valuable insights into choices that consumers make. The implications of this model for the “four P’s” (pricing, promotion, product, and distribution strategies) of the firm are far-reaching. The critical step in this approach is identification of key attributes on which the consumers base their purchase decisions.1 Delivering the preferred mix of attributes to consumers in critical market segments has profound implications for revenues, profits and general business decision making of the firm. An Example We continue the discussion of computer attributes by considering the purchasing decision of AnaLisa, a manager in a medium-sized company, who must outfit her sales force with laptop computers, subject to a budget constraint of $80,000. There are many different product attributes to consider, including processor speed, weight, screen size, peripherals compatibility, and CD vs. DVD player capability, to name a few. While allowing for some minimum standards on some of these features, however, Ana-Lisa decides to focus on two key attribute variables for her decision: RAM and hard disk storage capacity. She has identified six possible models, labeled A through F. The following table illustrates the price of each model, as well as the attributes ranking for these two features (shown in index measurements), the ratio of RAM-to-Storage, the number of units of each model that can be purchased with the budget of $80,000, and the total number of “points” for each attribute that can be purchased with this budget, model by model: . Table 1 Attributes and Prices of Six Laptop Computers Model Price A B C D E F $2,222 $2,500 $2,730 $2,647 $1,895 $1,974 Attribute Ratings* RAM Index 89 94 76 57 18 10 Hard Disk Storage Index 22 50 86 90 72 77 Ratio of RAM to Storage 4.05 1.88 0.88 0.63 0.25 0.13 Computers Total Attainable (within per $80,000 budget) RAM Storage 36.00 32.00 29.30 30.22 42.22 40.53 3,204 3,008 2,227 1,723 760 405 792 1,600 2,520 2,720 3,040 3,121 *Attribute ratings have been scaled to 100 points to allow an equivalent comparison of the attributes. Factor Analysis is frequently used to summarize a variety of factors (attributes) that influence the choices made by consumers. Conjoint Analysis technique is used to evaluate attribute trade-offs between different products in the market. 1 Kent Jones and Lidija Polutnik 10 Figure 8 shows the “raw” attribute ranking per unit for each model. Note that such an illustration shows how the models compare directly in terms of the key features, but does not put the manager’s decision into the context of multiple purchases within a budget constraint. In order to present this perspective, we refer to the total attribute “points” of each feature attainable with a budget of $80,000 (shown in the last two columns of table 1), and plot the attribute rays A through F in figure 9. Joining these end-points creates the efficiency frontier. Figure 9 illustrates Ana-Lisa’s purchasing decision. Her choice will depend on the relative preference for RAM vs. storage, as determined by the particular needs of the sales force, in conjunction with the goals of the firm. For example, a high relative preference for RAM, based perhaps on the need for running multiple applications simultaneously, would mean the applicable indifference curves would be relatively flat. In contrast, a strong relative preference for storage capacity, based on the size of data bases, spread sheets or other information files, implies relatively steep indifference curves. Suppose that the firm’s preferences, in the context of this decision and based on Ana-Lisa’s judgment, favor RAM, relative to Storage. Thus the marginal rate of substitution shows a strong (but not absolute) willingness to sacrifice the storage attribute for the RAM attribute, and the indifference curves will be relatively gently sloped. Figure 9 illustrates the case where the preference pattern implies that the firm cannot maximize utility with either model A or B alone, but rather with a mixed combination of the two. This situation was described earlier in the “Mixing Products” section. The optimal choice appears to be that Ana-Lisa will purchase a combination of model A and model B. Geometrically, this would be shown as a linear combination of vectors along OA and OB that sum to the resultant vector OM. RAM Index Figure 8 100 90 80 70 60 50 40 30 20 10 0 B A C D E 0 10 20 30 40 50 60 70 F 80 90 100 Hard Disk Storage Index In practical terms, the situation described in figure 9 may indicate that some sales persons benefit more with the stronger RAM model, and some with the stronger Storage model, based perhaps on Kent Jones and Lidija Polutnik 11 differences in the technical needs of maintaining various accounts. If the firm can smoothly trade off the RAM vs. Storage configurations, it would indeed be best to purchase a mix of the two models. We would assume that compatibility of the models (for networking or file sharing, for example) or other common features would facilitate this choice. If on the other hand the firm is constrained to stick with “one model,” for whatever reasons, then the second-best solution would be to maximize utility at either point A or point B, whichever reaches the highest indifference curve. As an exercise, draw an additional indifference curve in figure 9 (remember the rules!) that would show the best choice for purchasing just one model. Consider another qualification. The budget constraint of $80,000 resulted in purchases of various numbers of laptop units at each of the points A through F. The purchasing decision, set up in this manner, presumes some flexibility in allocating the laptops to larger or smaller numbers of sales persons. Perhaps the purchase of a smaller number of the more expensive laptops (C or D for example) would result in their being allocated on a hierarchical basis to the more productive sales persons, while others receive or keep older models. If one added the constraint that a minimum of, say, 32 units would need to be purchased, then one of two adjustments must be made in the analysis. Either models C and D must be eliminated from consideration, based on the $80,000 budget, or the budget must be expandable to cover 32 units on the contingency that C or D is the “best” model. Figure 9 Finally, this example contains the elements of a “principal-agent” problem. We assume in the example that Ana-Lisa is acting as a reliable agent of the firm, and that her decision is thus designed to maximize the welfare of the firm. In other words, the assumption is that the purchase decision and the allocation of computers will maximize the firm’s efficiency, as measured (typically) by the Kent Jones and Lidija Polutnik 12 contribution to profits over the relevant time horizon. One can imagine, for example, how the purchase decision might be different if the sales force collectively decided on the laptop model. It may be that some sales persons, with long hours on the road, might prefer laptops with DVD players, or enhanced game-playing capabilities, or with faster processors (for “bragging rights” or other reasons). This potential conflict does not arise in a typical end-user consumption decision, in which the decision maker presumably maximizes his or her own utility directly. In any case, it is reasonable to ask, in the context of an industrial purchasing decision, just how “utility” is defined, and how the incentive structures are set up to maximize the utility function. Exercise 5. In the preceding example, suppose now that the price of laptop model A rose to $3000. How would this change affect the efficiency frontier? How would it affect Ana-Lisa’s choice? Note that the increase in A’s price makes A inferior to B in both total attainable RAM and Storage attribute points (calculate these new values). A therefore is “priced out of the market.” Assuming the preference pattern describe in the example, Ana-Lisa’s choice would switch to B alone. Exercise 6. Suppose a new model “G” appeared on the market, with attribute index ratings of RAM = 90 and Storage = 40. At what price for G would Ana-Lisa consider purchasing this model over A, B, or a combination of those two models, based on the information from the example? References Douglas, Evan J. 1992. Managerial Economics: Analysis and Strategy. 4th Edition. Englewood Cliffs, New Jersey: Prentice Hall. Lancaster, Kelvin J. 1966. "A New Approach to Consumer Theory." Journal of Political Economy. 74:13257. ______. 1971. Variety, Equity and Efficiency. New York: Columbia University Press. ______. 1979. Consumer Demand: A New Approach. New York: Columbia University Press. Kent Jones and Lidija Polutnik 13