CAN Low SIGNAL HIGH QUALITY? EXPERIMENTAL EVIDENCE

by

N. DAWAR and M. SARVARY

96/91/MKT

Assistant Professor of Marketing at INSEAD, Boulevard de Constance, 77305 Fontainebleau Cedex, France.

Assistant Professor of Marketing at Graduate School of Business, Stanford University, Stanford, CA 94305-5015, USA.

A working paper in the INSEAD Working Paper Series is intended as a means whereby a faculty researchers thoughts and findings may be communicated to interested readers. The paper should be considered preliminary in nature and may require revision.

Printed at INSEAD, Fontainebleau, France. Can Low Price Signal High Quality? Experimental Evidence

Niraj Dawar

and

Miklos Sarvary

Niraj Dawar is Assistant Professor of Marketing at 1NSEAD, Fontainebleau, France, currently on leave at the Richard Ivey School of Business, University of Western Ontario, London, Ontario, Canada, N6A 3K7. Miklos Sarvary is Assistant Professor of Marketing, Graduate School of Business, Stanford University, Stanford, CA 94305-5015. The authors thank John Hulland, Lydia Price, and Priya Raghubir for convents on earlier drafts. Can Low Price Signal High Quality? Experimental Evidence

Abstract

Economic signaling theory suggests that interpret the price signal within the context of conditions. Under specific conditions it predicts that low price may signal high quality. Results from an experiment designed to test the behavioral assumptions underlying this prediction indicate that consumers intentions to purchase conform to the predictions of economic signaling theory, but their judgments of product quality do not. The results suggest that consumers response to signals may be more complex than previously shown. 1. Introduction

Signaling games are often used to describe situations characterized by asymmetric information about product quality between firms and customers (see Chu 1993 for an example in marketing). On the assumption that consumers and firms are rational and capable of interpreting each others moves, signaling games specify market conditions under which firms can resolve information asymmetry and deliver product quality information to consumers through, for example, price or weight of advertising. Consumers update their beliefs about product quality by observing and interpreting these signals within the context of market conditions.

The objective of economic signaling studies is to determine equilibrium market outcomes.

However, a challenging and important task for research is to understand consumers ex post quality perceptions or, more generally, the behavioral outcome of signaling games. Previous research has examined and found support for signaling theorys assumptions about the effects of advertising expenditures and warranties on consumers perceptions of quality (Kirmani 1990;

Kirmani and Wright 1989; Boulding and Kirmani 1993).

The present paper contributes to this research by focusing on price as a signal.

Specifically we examine a set of market conditions in which signaling theory predicts that low price may signal high quality. We study consumers perceptions of product quality and their intentions to purchase based on different in the context of a specific price-signaling game proposed by Nelson (1970) and Schmalansee (1978). Our objective is to test the predicted behavioral outcomes of this game. Unlike previous consumer research, we do not find complete support for the predicted outcomes. A key result of our study is that while the relevant behavioral variable, "intention to purchase," is consistent with the predictions of the Nelson-Schmalansee game, quality perceptions are not. The findings suggest that the behavioral outcomes of economic signaling games may be more complex than previously expected.

1 In the next section we describe the specific Nelson-Sclunalansee game which we test.

Based on this game, we develop hypotheses regarding the relationships between price, perceived quality and intentions to purchase for the different market conditions. Section 4 describes the experiment and Section 5 discusses the implications of the results.

2. The Nelson-Schmalensee game: the case of introductory

Where consumers are faced with a new product for which quality is unobservable and can only be assessed through ("experience good"), Nelson (1970, 1974) suggested that

"wasteful" expenditures by firms (such as very low introductory prices or an expensive advertising campaign) may signal high product quality in a repeat purchase context. From such signals consumers infer that the firm is confident of repeat business for the product and is willing to forego current profits in exchange for future returns. A low quality producer on the other hand is unable to follow the same strategy because of the low likelihood of repeat purchases. For the high quality firm this "low- price" strategy results in current losses which can be viewed as the cost of signaling. If future gains for the high quality firm from repeat business are likely to compensate for and exceed the short term losses due to low introductory pricing, there exists a separating equilibrium in which a low price

2 signals high quality. Under such market conditions, from a low price consumers infer high product quality. In the opposite case, where long term gains cannot cover short term losses, both high and low quality firms follow the same pricing strategy and we have a pooling equilibrium. Under such market conditions, prices only reflect average quality on the market (consumers priors) and consumers do not learn about product quality at all. In a third market condition, unexplored by economic signaling theory, market information available to consumers may be insufficient to determine whether short term losses can be recovered by firms in the long run. In this case market conditions are silent on the relationship between price and quality and consumers remain uncertain of it.

2 It is important to note that the predictions of this specific signaling game (see Tirole, 1992 for a simple formulation) are based on the following assumptions: (1) the product is an "experience good" that is (2) repeatedly purchased; (3) low quality must not be profitable under full information, i.e. if consumers knew the quality of the product prior to purchase they would not buy the low quality product even if it were cheaper, and finally, (4) there is no consumer heterogeneity in preferences (i.e. consumers have roughly the same for the product). The third assumption is particularly important for the design of our empirical test. It ensures that in the context of this game purchase behavior is entirely driven by quality perceptions. Thus, in our experiment we expect, within an acceptable price range, the effect of price to be identical on intentions to purchase and perceptions of quality. Hypothesis for each of the market conditions outlined here are developed in the next section.

3. Hypotheses

The theory predicts an interaction of price with market condition on consumer behavior

(intentions to purchase and perceptions of quality). The specific effects predicted in each of the market conditions are described below.

1.) Consumers are able to assess that signaling high quality with a low introductory price is possible: under this condition, the market environment carries sufficient information for consumers to assess that a high quality producer may be able to signal quality with a low introductory price and make up for the current loss of by securing a high level of repeat purchase in the future. This condition is labeled the Separating condition. In this situation, economic signaling theory predicts that judgments of quality and intentions to purchase are negatively related to price.

Hi: When market conditions indicate that signaling quality with a low price is possible, judgments of quality and intentions to purchase are negatively related to price.

3 2.)Consumers are able to assess that signaling quality with price is not possible: If consumers have sufficient information to assess that future gains will not compensate good quality firms for current losses they incur from introductory low prices, they can be certain that the price signal is ineffective. Economic signaling theory predicts that both types of firms (low and high quality) will choose similar prices. This price reflects consumers (aggregate) prior expectation of quality on the market. Thus, a higher price indicates that the market evaluates that the expected quality is also higher. Therefore, we can expect a positive relationship of judgments of quality and intentions to purchase with price. This market condition is labeled the Pooling condition.

112: When market conditions indicate that price signaling is not possible, judgments of quality and intentions to purchase are positively related to price.

3.)Consumers are unable to assess signal effectiveness: If essential pieces of information required to interpret the price signal are unknown, the effectiveness of the signal is uncertain. For example, if the future prospects of a nascent industry (and consequently the ability of firms to benefit from repeat sales) are not known, consumers do not have sufficient information to assess whether price is an effective signal. Consistent with the logic of signaling theory, under these circumstances we would expect to observe no systematic relationship between price and perceived quality or intentions to purchase the product We label this condition the Uncertain condition.

H3: When there is insufficient information to assess the effectiveness of the price signal, judgments of quality and intentions to purchase are not related to price.

4. Experiment

The theory predicts that consumers interpret price based on market conditions, and that we should expect a negative, positive or no relationship between price and quality judgments depending on these market conditions. The following experiment is designed to test these predictions.

4 4.1 Design and Stimulus Materials

A 3x2 factorial between-subjects design was used to test the hypotheses. The independent variables were market condition (Separating, Pooling, and Uncertain), and price (Low and High).

The stimulus product used was a fictitious new type of optical computer diskette. The product was described as having a larger storage capacity than existing diskettes but being

compatible with existing PC disk drives. This product was chosen as a stimulus because: (1) of subject familiarity with the product category of diskettes; and (2) conformity with the assumptions of the price signaling game (it is a repeat purchase, experience good for which utility derived is homogeneous across consumers, and the down-side risk of low quality is considerable (the potential loss of data)); and (3) plausibility of an "introductory price" scenario.

4.2 Subjects and Procedure

Data consisted of responses from 161 subjects to a questionnaire administered in class.

The questionnaire was initially administered to 220 respondents. In order to ensure a rigorous test of the economic signaling hypotheses, respondents were screened so as to meet the assumptions of the signaling game on the basis of three questions: (1) whether they saw the diskettes as a repeat purchase product (yes/no); (2) whether they would purchase a product at a low price, if they knew it was of a low quality (yes/no); and (3) what they thought was the storage capacity of the diskettes relative to those they were currently using (1=much smaller, 7=much greater). Subjects who did not see the diskette as a repeat purchase product, who said they would purchase the product at a low price even if they knew it to be of a low quality and/or thought this product had a lower storage capacity than their current diskettes (<4 on the seven-point scale) were withdrawn from analysis.

Subjects were instructed to carefully read an excerpt from an article about the new optical diskette from a fictitious magazine called Computing Memmy Monthly. The article excerpt

5 described the new diskette, and manipulated the market conditions. It indicated a price range per box of ten diskettes ($19 to $69)3, and an average price per box (around $44). Market condition was manipulated by including information about (1) the differences in the costs of of good and poor quality diskettes; and (2) long term prospects of the optical diskette in terms of whether it is likely to establish itself as a standard or be overtaken by new technology before it becomes a mass market item. The excerpts used for the three market conditions are reproduced in

Figure 1. In the Separating condition, the excerpt indicated that cost differences between producing high and low quality diskettes were small, and that industry specialists expected the diskette to become the new industry standard. 4 In the Pooling condition, the article indicated that cost differences for producing high and low quality diskettes were very high (with the implication that it would be expensive for high quality producers to signal quality through a low price). The key manipulation information suggested that firms would not have time to benefit from repeat business because industry specialists forecast that optical diskettes would be overtaken by new technology before they could become mass market items. The Uncertain condition provided subjects with no indication of differences in costs of production, and suggested uncertain future prospects for the diskette by stating that industry specialists were divided in their forecast of the optical diskettes future.

Subjects were then informed that they had a special offer from Typhon Tech Inc. (a fictitious firm), to purchase a box of ten diskettes for the special introductory price of $19 or $62

(crossed with the market information condition and manipulated between subjects). They were then asked to rate the diskettes on five quality dimensions (reliability, durability, likelihood of malfunction, relative quality, and overall quality), and indicate their intention to purchase. All variables were measured on seven point scales (1=low, 7=high).

6 4.3 Analysis and Results

Under the hypotheses, we expected a negative relationship between price and judgments of quality as well as price and intentions to purchase in the Separating condition; a positive relationship in the Pooling condition; and no relationship in the Uncertain condition. Cell means are graphically presented in Figure 2, along with cell sizes and standard deviations.

A factor analysis of the five quality dimensions confirmed that they loaded on a single factor which accounted for 71% of the variance. The linear combination represented by this factor was used as the judgment of quality" dependent variable.

4.3.1 Effects ofPrice on Intentions to Purchase

An omnibus F-test across the three market conditions crossed with the two levels of price reveals a significant interaction of information condition with price (F 2, 154=8.33, p<0.001) indicating that the relationship of intentions to purchase to price differs depending on the market condition. Planned contrasts compared cell means of intentions to purchase at the two price levels to determine the source of the interaction. The Separating condition shows a significant negative relationship of intentions to purchase with price (F 1, 55=4.19, p<0.05, means 4.50 vs. 3.66). In the

Pooling condition, there is a significant positive relationship of intentions to purchase with price

(F1,53=14.18, p<0.001, means 2.91 vs. 4.22). Finally, in the Uncertain condition, intentions to purchase indicate only a marginally significant positive relationship to price (F 1,46=2.96, p<0.10, means 3.80 vs. 4.54).

4.3.2 Effects ofPrice on Judgments of Quality

An omnibus F-test across the three market conditions crossed with the two price levels shows that price and market condition do not interact (F2, 151 .79, >0.45). A strong main effect of Price exists (F1,151=58.66, p<0.001) and there is no main effect of market condition (F2,

151=3.57, p>0.55). 5 Analysis using planned contrasts compared cell means of judgments of

7 quality for the two price levels in each market condition. The results indicate a clear positive relationship between price and judgments of quality in each of the three market conditions (F1,

54=15.68, p<0.01 with means -0.58 vs. 0.40 in the Separating condition; F 1, 52=14.6, p<0.01 with means -0.48 vs. 0.40 in the Pooling condition; and F 1, 45=34.61, p<0.01 with means -0.64 vs.

0.67, in the Uncertain condition).

5. Discussion and Conclusion

As predicted by the theory, the relationship of intentions to purchase with price appears to vary depending on market condition. The specific effects are consistent with predictions from the

Nelson-Schmalensee game which suggests that in the Separating condition a negative relationship between price and intentions to purchase should hold; in the Pooling condition, a positive relationship; and no relationship in the Uncertain condition. However, in contrast to intentions to purchase, consumers judgments of quality appear to be positively related to price regardless of market condition. This pattern of results suggests that behavioral outcomes of the market conditions specified by the signaling game may be more complex than previous research has shown. In particular, the two dependent variables may be affected by different aspects of the market conditions which make up the game. However, since Economic Signaling Theory is an

"as-if theory whose purpose is to develop equilibrium predictions, it is silent on the mechanisms by which different behavioral outcomes may be reached. The results of the present experiment suggest that investigating such mechanisms may be a fruitful area for future research..

An implication of these results is that behavior (intentions to purchase) may be inconsistent with judgments of quality under certain market conditions (e.g. Separating condition).

These results support Monroe and Krishnans (1985) conclusion that "perception of quality and willingness to buy are separate behavioral responses." Indeed, the two responses may be subject to different antecedents. Quality judgments may be formed on the basis of simple heuristics, using

8 price as a direct positive indicator of quality. As Lichtenstein and Burton (1989) suggest,

consumers appear to operate on the basis of a "general schema" of a positive price-quality

relationship across product categories. The results reported here suggest that the use of such

schemas may also generalize across market conditions. Intentions to purchase, on the other hand, involve greater commitment and may be based on a more elaborate processing of available information in the market.

A possible mechanism by which price information combines with market conditions and leads to behavioral outcomes is suggested by previous research and argues that consumers response may be driven by perceptions of "fairness" of price given cost information (Kahneman,

Knetsch and Thaler 1986; Okun 1981; Urbany, Madden and Dickson 1989). Our stimuli for the different market conditions included information on the variations in cost of production. In the

Separating condition it was stated that "Large quality differences are surprising in view of the fact that the cost differences in the production of the high versus low quality diskettes are small and insignificant." In this condition, .consumers may not have been willing to pay a higher price given that it did not reflect cost differences. Subjects in the low price condition had higher intentions to purchase than subjects in the high price condition. In the Pooling condition it was stated that

"...cost differences between producing a high quality and low quality diskette are in fact yew high." When costs drive differences in quality, we observe a positive relationship between price and intentions to purchase. Finally, in the Uncertain condition, there was no information about costs and we observed no relationship between price and intentions to purchase. This explanation for the inconsistency of judgments of quality and intentions to purchase is based on the premise that the two dependent measures are affected by different antecedents. Understanding these different antecedents is a promising area for future research.

9 NOTES

There need not be two firms. Consumers may be uncertain about product quality even if there were only one firm. Assuming that lower quality is cheaper to produce and that quality is not observable, any firm has an incentive to sell low quality products at a high price. Therefore, even a monopolist must convince consumers that its product is of a good quality. Sending a costly signal is one means of doing so.

2 Schmalensee pointed out that consumers need to also know that the price charged by the high quality producer is below the of the low-quality producer. While in reality it is unlikely that consumers have such detailed cost information, we do examine situations where consumers are given the information and where they are not.

3 These prices were pretested to be within an acceptable price range for this product for these subjects.

4 In one sub-condition subjects were additionally given the marginal cost of producing a diskette. s The Separating condition had been split into two sub-conditions where one set of subjects received marginal cost information and the other set did not. The results show no interaction of sub-condition with price on judgments of quality (F1, 52=2.17, p0.14), and no main effect of sub-condition on judgments of quality (F1,p>0.85). Similarly, no interaction of the sub-condition with price was observed for intentions to purchase (F1, 53=0.26, p>0.61) and there was no main effect of sub-condition (F1,53=0.66, p0.42). The two sub-conditions were collapsed into a single Separating condition for further analysis.

10 REFERENCES Boulding, William and Amna Kirmani (1993), "A Consumer-Side Experimental Examination of Signaling Theory: Do Consumers Perceive Warranties as Signals of Quality?" Journal of Consumer Research, 20 (June), 111-123. Chu, Wujin (1993), "Deand Signalling and Screening in Channels of ," Marketing Science, 11, 4, 327-347. Kahneman, Daniel, Jack L. Knetsch and Richard Thaler (1986), "Fairness as a Constraint on Profit Seeking: Entitlements in the Market," The American Economic Review, 76 (September), 728-741. Kinnani, Amna (1990), "The Effect of Perceived Advertising Costs on Brand Perceptions," Journal of Consumer Research, 17 (September), 160-171. Kinnani, Amna and Peter Wright (1989), "Money Talks: Perceived Advertising Expense and Expected Product Quality," Journal of Consumer Research, 16 (December), 344-353. Lichtenstein, Donald R. and Scot Burton (1989), "The Relationship Between Perceived and Objective Price- Quality", Journal ofMarketing Research, (Nov.), 429-443. Monroe, Kent B. and R. Krishnan (1985), "The Effect of Price on Subjective Product Evaluations," in Perceived Quality: How Consumers View Stores and Merchandise, Jacob Jacoby and Jerry C. Olson eds., Lexington, MA: Lexington Books, pp 209-232. Nelson, Philip (1970), "Information and Consumer Behavior," Journal of , 78 (2), 311- 329. Nelson, Philip (1974), "Advertising as Information," Journal ofPolitical Economy, 81 (July-August), 729- 754. Okun, Arthur (1981), Prices and Quantities: A Macroeconomic Analysis, Washington: The Brookings Institution. Schmalensee, Richard (1978), "A Model of Advertising and Product Quality," Journal ofPolitical Economy, 86 (31), 485-503. Tirole, Jean (1992), The Theory of , Second Edition, Cambridge, MA: MIT Press. Urbany, Joel E., Thomas J. Madden, and Peter R. Dickson (1989), "Alls Not Fair in Pricing: An Initial Look at the Dual Entitlement Principle," Marketing Letters, 1(1), December, 17-26.

11 Fi re l

Stimulus Article Excerpt

Optical Diskettes: Convenient, But Hard to Choose the Right One? There has been a lot of debate about the new optical diskette recently launched by various companies. The new diskettes resemble traditional floppies in size and form but have 10 times the storage capacity. The optical diskettes big advantage over compact disks is that users need no new hardware: one can read and write on the diskette using a standard 3.5" floppy drive available on all personal computers. The diskette itself contains the device that converts magnetic signals to optical ones.

A recent consumer survey revealed that there are large quality differences among different optical diskettes currently on the market. Quality ratings of various brands of diskettes by a panel of experts ranged between 1 and 9.5 on a ten point scale, with average quality being 4.7. Satisfied customers say the new diskettes are "much better" than traditional ones. However, over 94% of the buyers who claimed not to be satisfied with the diskettes they purchased said they would never buy the same brand again.

Large quality differences are surprising in view of the fact that the cost differences in the production of the high versus low quality are small and insignificant.

The survey also shows that cost differences between producing a high quality and low quality diskette are in fact very high.

Prices currently on the market show considerable variation. A box of ten diskettes can cost anywhere between $19 and $69, with average price around $44

Industry specialists forecast that optical diskettes will become a new standard and expect that they will eventually replace standard magnetic diskettes as a primary storage device.

Industry specialists forecast, however, that optical diskettes will be overtaken by new technology before they become mass market items.

Industry specialists are divided in their forecast of the o ptical diskettes future. While some believe that optical diskettes will become a new standard and exoect that they will eventually replace standard magnetic diskettes as a orimary storage device, others are more pessimistic and exoect optical diskettes will be overtaken by new technology before they become mass market items.

Note: Text in Bold appeared in the Separating condition; Text in Italics appeared in the Pooling condition; Underlined text appeared in the Uncertain condition. All other text appeared in all three conditions.

12 Figure 2

Cell Means (Cell sizes; Standard Deviations) of Judgments of Quality and Intentions to Purchase under three Market Conditions

SEPARATING POOLING UNCERTAIN

7 6 6 4.50 0.40 0.40 (28; 1.75) (29; 0.91) 5 (32; 0.79) 5

•••••••••••••••••■••••••••••••■■••••••••••••••■••••••••••••■•••••• 4 --- -4.22 4 2.91 ....- 3.66 0 (32; • 1,18. 0 (29; 1.34) 3 n; 1-g)----- 3

2 2 -0.48 -0.59 1 23; 0.91) 1 (28; 0.96) -1 0 -1 0 -1

Low High Low High Low High PRICE PRICE PRICE

- Judgment of Quality; left-hand scale Intention to Purchase; right-hand scale