Brand as a reliability reference point: a test of prospect theory in the used car market.

ABSTRACT

Prospect theory (Kahneman & Tversky, 1979) suggests that consumers are risk seeking for losses (below the reference point) and risk-averse for gains (above the reference point). In this paper we investigate whether consumers behave according to prospect theory in the used car market. We develop hypotheses and test the idea that consumers compare reliability of automobiles using reference points based on brand and that this in turn affects the price. We find empirical support for the prospect theory predictions that consumers are risk seeking when the reliability of a specific brand/model of a car is below the average for the brand and risk averse when the reliability is above average. The paper concludes with suggestions for alternative reference categories and a discussion of implications for academics and practitioners.

Keywords: Prospect Theory, Pricing, Reference Pricing, Decision Making, Brands

1. INTRODUCTION

Prospect theory (Kahneman and Tversky, 1979) explains how people make choices among alternatives (prospects) by comparing them to reference points. Daniel Kahneman was awarded the Nobel Prize in 2002 "for having integrated insights from psychological research into economic science, especially concerning human judgment and decision-making under uncertainty" (The Nobel foundation, 2002). Before prospect theory, individual decision behavior was explained and predicted by expected utility theory (Von Neumann & Morgenstern, 1944). Expected utility theory posits that rational choices are made based on the levels of relevant decision criteria (Friedman & Savage, 1948). However, the rationality assumptions of expected utility theory have been questioned and have not found empirical support (Slovic & Tversky, 1974). Upon further examination, human choices seem to frequently depend on the framing of a problem (Tversky & Kahneman, 1986; Hogarth & Reder, 1986), which follows the propositions of prospect theory but cannot be explained by expected utility theory (Kahneman & Tversky, 1979; Shafir, 1999).

Prospect theory has been widely applied to decisions regarding financial outcomes; however, it has not found much application in relation to choices regarding product quality (Ong, 1994). Quality warrants study because it is an important characteristic used by consumers in their evaluations. However empirical investigations into decision-making under uncertainty using quality have been hindered by difficulties in operationalizing and measuring quality (Ong, 1994).

In this paper we empirically test the non-linear effect of quality on price for durable high-ticket goods with as existing resale market as predicted by prospect theory. We use the new car market because prices are established primarily though a 'competitive market' dynamic and there is a relative abundance of information related to car quality.

The paper begins with a general discussion of prospect theory, followed by an examination of prospect theory in application to quality of products. This is followed by an exploration of the used car market and specific hypotheses. In the next section the sample and methods are addressed, followed by the analysis and results. The paper concludes with implications for practitioners and suggestions for further research.

2. PROSPECT THEORY

Nobel-winning prospect theory (Kahneman and Tversky, 1979) is one of the most influential theories of decision-making. It was developed to explain instances where the traditional expected utility theory failed to explain people's choices. Under uncertainty, expected utility theory suggests that people weigh the outcomes by the probabilities of their occurrence. They then choose the one that provides the highest absolute expected value. However actual individual judgment is often heavily affected by the framing of the situation and violates the rules of expected utility estimation (Tversky & Kahneman, 1986).

Framing of alternatives (Thaler & Kahneman 1981) is one of the major drivers of choices according to the prospect theory. The process can be briefly described as follows: people 'code' the outcomes of various prospects (alternative outcomes) as either gains or losses relative to some reference point. Then they weigh the resulting gains and/or losses by their subjective probabilities. When weighing the outcomes, they weight gains differently from losses, acting risk-seeking for losses, and risk-averse for gains.

2.1 Coding and Reference Points

According to the prospect theory, Individuals 'code' each alternative as a gain or a loss in utility relative to a reference point rather than using the absolute value of the outcome. The resulting losses or gains are then weighed by their perceived probabilities of occurrence, forming a non-linear value function. Because people typically approach gains and losses differently, generally acting risk-averse on gains and risk-seeking on losses (Kahneman and Tversky, 1979), the result is a non-linear utility function.

An integral part of prospect theory is the reference point relative to which the outcomes are coded (Kahneman & Tversky, 1979) as gains or losses. Two kinds of reference pricing are recognized: internal (stored in consumer's memory) and external (provided by the environment). For example, "manufacturer suggested retail price $55.00" on the price tag of a $39.00 item is an attempt to establish an external reference price and subsequently invoke coding of the decision making problem as a gain. In fact, reference prices ("any price in relation to which other prices are seen"--Biswas & Blair, 1991, Briesch, Krishnamurthi, & S P Raj, 1997) are where prospect theory has found the greatest application (Erdem, Mayhew, & Sun, 2001; Niedrich, Sharma, & Wedell, 2001).

Tversky and Kahneman (1986) also observed that people derive more disutility from a loss than they derive utility from an equivalent amount of gain. Thus, the 'pain' from losing $100 exceeds the 'pleasure' of gaining $100. This "strong distaste for losses" has been evolving over the years to also account for multiple alternatives and pessimism and optimism.

2.2 Prior Research Utilizing Prospect Theory

Prospect theory is utilized in a great many academic disciplines, such as marketing, finance, economics, management, decision theory and political science. In marketing, prospect theory has frequently been applied to decisions involving money. Prices, discounts, coupon promotions, advertising, monetary incentives including sales force compensation, product bundling (Stremersch & Tellis, 2002: Johnson, Herrmann & Bauer, 1999) have all been investigated using prospect theory. Occasionally prospect theory has been applied to decisions regarding time (Mowen & Mowen 1991; Leclerc, Shmidt & Dube', 1995) and much less frequently to quality Ong (1994).

There have been a great many designed laboratory experiments that have shown strong support for prospect theory (Kahneman & Tversky, 1979; Salminen & Wallenius, 1993), and some studies that have used panel data (Mayhew & Winer, 1992). Despite strong findings, reliance on experimental data, amplified by the almost exclusive use of college students as the subjects of such experimentation, raises concerns related to the generalizibility (external validity) of the findings.

2.3 Quality and the Used Car Market

For our current research, we narrow our focus to durable goods that is such goods that last throughout many uses and maintain value over a period of time. Among the durable goods, there are expensive items, such as cars and major appliances. Purchasing such an item is usually a high-involvement, information-processing intensive process. For these goods the high initial cost and retained value allows for a resale market (Ocada, 2001) that is close to pure competition. Despite the support for applying prospect theory to the effects of quality, it remains an underdeveloped area (Ong, 1994; Liu, 1998) due to the general difficulty of operationalizing and quantifying quality (Ong, 1994). In the used car market however, there is a well-developed set of quality information that allows for the effects of quality to be analyzed.

3. HYPOTHESES

The basic dynamics underlying prospect theory to the relationship between quality and prices can be tested by examining the used car market. In this case different cars are the prospects which get evaluated relative to quality reference points. Generally we expect reliability (quality rating) to be related to price. Additionally, in line with prospect theory, we expect the effect of the relative quality to be nonlinear, people will act as risk-averse for gains, and risk seeking for losses, and the effect will be is steeper for losses than for gains. These relationships are illustrated in Figure 1.

[FIGURE 1 OMITTED]

We assume that the used car buyer compares the reliability rating of a specific car with a reference reliability rating. In our analysis we use a reference reliability level as the reliability level associated with the brand name. Research has shown that the brand average reliability influences the evaluation of specific cars by what is referred to as a 'brand halo' effect (Betts & Taran, 2002). The baseline model includes new price and wheelbase, as more expensive new cars tend to become more expensive used cars, and size of the vehicle is an important factor affecting their price (Betts & Taran, 2002).

3.1 Utility Reference Point

In situations where the reliability rating of a specific car is higher than the reference rating, since people are risk-averse for gains, the incremental utility of the gain due to higher reliability decreases the further from the reference point. On the contrary, in a situation where the reliability rating of a specific car is lower than the reference rating, since people are risk-seeking for losses, the incremental (dis)utility of the loss due to lower reliability increases the further from the reference point. This utility should manifest itself in the price that the buyer is willing to pay, therefore:

Hypothesis 1--reliability will have a decreasingly positive effect on price for cars with reliability rating above average for its brand and an increasingly positive effect on price for cars with reliability rating below average for its brand

3.2 Strong Distaste for Losses

The second hypothesis reflects the "strong distaste for losses" observed by Tversky and Kahneman (1986). The absolute value of the utility of a higher reliability than the reference would be lower than that of an equally lower reliability. This utility should also manifest itself in the price, therefore:

Hypothesis 2--The increase in the relationship between reliability and price when reliability rating is below average for the car's brand is greater than the decrease in the same relationship when reliability rating is above average for its brand.

3.3 General Reliability Effect

Finally, the reliability of a car should directly impact the used car's price, as the more reliable the car, the more of utility it can offer. This utility should manifest itself in the price that the buyer is willing to pay, therefore:

Hypothesis 3--The price of a used car will be positively related to the car's reliability.

4. Methods

4.1 Measures

Information on passenger cars of 17 makes (parental brands) from 1993 to 1999 were obtained from Kelley Blue Book (Kelley, 2002), Consumer Reports (2001), and Edmunds (2002). The dependent variable, price of used cars, was "private-party price" as reported by Kelley Blue Book (Kelley, 2002). Two independent variables, wheelbase and engine size, are also as reported in Kelley. The price of a new vehicle produced in 2001 comparable to the used vehicle was as reported by Edmunds (2002). For each car, reliability rating on 14 "trouble spots" was obtained from the Consumer Reports (2001). These 14 ratings were combined to create a reliability scale with an alpha reliability of 0.927 (Chronbach, 1951). Brand reliability was computed as the average reliability of all the passenger cars for a specific brand, across all the models and years of cars within that brand. The particular brands chosen represent a variety of countries of origin and price ranges.

4.2 Modeling and Analysis

To analyze the data, we used regression analysis methods. To approximate the nonlinear dependency between quality and price depicted in Figure 1, functional transformation using cubic root of the difference between the car's reliability and the reference reliability was used ('the curve term'). Cubic root function produces a shape similar to that which is conceptually predicted by the prospect theory.

To model the shift in the slope past the reference point, we have included an interaction effect ('slope term') with the sign of the transformed deviation from the reference point and the 'curve' term achieve the effect using piecewise regression (Smith, 1979).

The hypotheses were tested using a series of regressions. First, regression was run using only the control variables, price new and wheelbase. Second, the curve term was added. Third, the slope term was added into the regression. Fourth, reliability was added.

4.3 Results

The results of the regressions are presented in Table 1. In Table 1, Model 1 is the base model with only price new and wheelbase as the independent variables. Model 2 adds the curve term, Model 3 the slope, and Model 4 reliability.

Adding the Brand Curve term to the baseline model significantly enhances the model fit, with [R.sup.2] growing from .561 to .724. Adding the Brand Slope term causes a statistically significant but somewhat trivial increase of 0.002 in [R.sup.2]. Adding the reliability instead of the Brand slope brings about a significant change in [R.sup.2] of 0,2, and ever term is statistically significant.

Thus, Hypothesis 1 and Hypothesis 3 are confirmed. However, Hypothesis 2 has not been confirmed. Several reasons could account for the lack of finding with regards to the Hypothesis 2 if the theory indeed holds: small number of points around the reference point might not allow for the subtle effect to be observed in our sample; the limited nature of the Kelley blue book prices might not allow for the uncovering of such dependencies.

CONCLUSIONS AND IMPLICATIONS

In this study, we have shown that consumers' evaluation in the used car market is influenced by the reliability of used cars in a manner consistent with that predicted by prospect theory (Kahneman & Tversky, 1979). This study bears important implications for researchers, educators and practitioners. For researchers it is a contribution to our understanding of consumer driven pricing, specifically the relationship between quality and price. Another contribution is to a largely under-developed area of application of prospect theory in examining the factors, which go into a price decision. The proposed research design al offers simple methodology utilizing available secondary data sources, thus overcoming the generalizibility and external validity issues inherent in traditionally used in the field experimental studies and being easy to extend into practical applications. Also, as an alternative to experimental studies, it allows for the triangulation of findings.

For educators in the field of marketing and consumer behavior, this study can be used as supplemental material in a classroom discussion related to pricing, promotion, consumer decision-making, and/or product quality.

For practitioners it explains quality reference points and emphasizes the importance of identifying the proper quality reference points. Such knowledge could be practically applied in positioning and/or promotional efforts. Sellers of cars of superior quality could attempt to upwardly influence consumer's reference quality point. This would cause the quality of other cars to be coded as losses and potentially bring along much steeper premium than if the reference point were at the lower level.

Stephen C Betts, William Paterson University, Wayne, New Jersey, USA

Zinaida Taran, Siena College, Loudonville, New York, USA

TABLE 1--MODEL COEFFICIENTS AND FIT

                     Model 1               Model 2

                beta                  beta
                (SE)        sign.     (SE)        sign.

(constant)      31097       0.000     33061       0.000
                (5337)                (42420

Price New       0.580       0.000     0.576       0.000
                0.027                 (0.022)

Wheelbase       -318.5      0.000     -336.3      0.000
                (52.8)                (41.9)

Brand Curve                           1443.9      0.000
                                      (97.6)

Brand Slope

Reliability

[R.sup.2]       0.561                 0.724

Partial F                             219.001
(models)                              (1,2)       0.000

                     Model 3               Model 4

                beta                  beta
                (SE)        sign.     (SE)        sign.

(constant)      33096       0.000     17377       0.001
                (4226)                (4990)

Price New       0.581       0.000     0.556       0.000
                (0.022)               (0.021)

Wheelbase       -350.2      0.000     -273.1      0.001
                (42.4)                (42.0)

Brand Curve     726.7       0.058     599.80
                (381.9)               (180.6)

Brand Slope     1440.1      0.053
                (741.5)

Reliability                           181.8       0.000
                                      (33.2)

[R.sup.2]       0.726                 0.744

Partial F       3.772                 29.962      0.000
(models)        (2,3)       0.053     (2,4)

Dependent variable is the 'private-party price' (Kelley, 2002)