How a product warranty influences customers’ willingness to pay
Vincent Eijsbouts 10253610 BSc Bedrijfskunde
2 How a product warranty influences customers’ willingness to pay
Vincent Eijsbouts Vincentt__@hotmail.com
10253610 BSc Bedrijfskunde
Supervisor: Anouar El Haji
3 Abstract
This study is concerned with the relation between a warranty and customers’ willingness to pay. The objective is to study to what extent and how a warranty and its framing influence the customers´ willingness to pay. We hypothesize that the presence of a warranty positively influences the willingness to pay, and we also hypothesize that this effect is mediated by perceived risk. Furthermore, we posit that different warranty frames influence the strength of the relationship between a warranty and willingness to pay. Finally, hypotheses are formulated concerning two specific elements of perceived risk: performance and financial risk. These hypotheses suggest that warranties influence perceived performance risk, and that perceived performance risk affects perceived financial risk. We test all hypotheses through the means of an online experimental auction with a sample of 512 Dutch participants. All hypotheses but one about differing warranty frames are confirmed. The results of this study suggest that perceived risk mediates the relation between a warranty and willingness to pay. Also, additional analysis shows that a warranty positively moderates the relation between performance and financial risk.
4 Table of contents Page 1. Introduction 6 2. Theoretical framework 8 2.1 Warranty framing 8 2.2 Willingness to pay 8 2.3Perceived risk 10
2.4 Gaps in current literature 12
3. Conceptual model 14 4. Methodology 18 4.1 Sample 17 4.2 Design 18 4.3 Method 20 4.4 Procedure 20 4.5 Measurements 21 4.5.1 Willingness to pay 21 4.5.2 Perceived risk 21
4.5.2.1 Perceived performance risk 21
4.5.2.2 Perceived financial risk 21
4.5.3 Manipulation check 22
4.5.4 Other control variables 24
5. Results 25
5.1 Main analyses 26
5.1.1 Relation between warranty and willingness to pay 26 5.1.2 Relation between framing warranties and willingness to pay 27
5.1.3 Mediation of perceived risk 28
5.1.4 Relation between perceived warranty and perceived performance risk 33 5.1.5 Relation between perceived performance risk and perceived financial risk 33
5.2 Additional analyses 33
6. Discussion 36
6.1 Key findings 36
6.2 Limitations and alternative explanations 39
5 6.2.2 Non-measured variables explaining the results 40
6.2.3 Low explained variance 41
6.2.4 Methodological limitations 41
6.3 Practical value of the study 42
6.4 Concluding remarks 42
6 1. Introduction
In today’s world, marketing can be found anywhere and is one of the key activities of many companies. This is reflected in the fact that it is no exception when a company spends up to 12 percent of its budget on marketing activities (Moorman, 2011). One extrinsic product feature often mentioned in marketing is that of a warranty. The most common form of a warranty is the
standardized consumer product warranty. Priest (1981) explains why products often include this types of warranties. The theory explained by Priest states that warranties are provided to customers to help them deal with the difficulty of estimating the risk of product defects at the time of purchase. It states that warranties supposedly function as tools to inform the customer about product quality, product durability, and product reliability; warranties have a signaling function.
This study focuses on whether mentioning a warranty increases perceived value. Many studies identify various effects of warranties on marketing activities (Bearden & Shimp, 1982; Wirtz, Kum & Lee, 2000; Erevelles, Roy & Vargo, 1999; Boulding & Kirmani, 1993; Lwin & Williams, 2006). Also the underlying process of this influence is subject to some research. Perceived risk supposedly mediates this relation since warranties signal value and therefore reduce the customers’ perceived risk (Bearden & Shimp, 1982; Sweeney, Soutar & Johnson, 1999; Erevelles, Roy & Vargo, 1999). These studies suggest that it is this lowered risk perception that increases customers’ willingness to pay rather than the warranty itself.
Companies can present a warranty to the customer in many ways. Companies can for example emphasize different characteristics, or use different words to communicate the same thing when presenting a warranty. How a warranty is presented is what we call the warranty frame in this study. Current literature barely addresses warranty framing and its effects on the customers’
perception of value. Besides, the mediating role of perceived risk has not been sufficiently tested and elaborated in the context of manufacturer warranties. Furthermore, research is needed in different types of risk and how these are related. This study tries to contribute to filling these research gaps. The main research question is: To what extent and how does a product warranty and its framing influence the customers´ willingness to pay in an online real-life auction setting? These variables are studied by assessing the direct influence of several warranty frames on willingness to pay, and studying the underlying cognitive processes involved.
This study contributes to the literature by integrating the existing marketing theories and testing the conceptual model in practice. Practically, the results can also be relevant to businesses since these are likely to be helpful in improving their marketing activities.
A field experiment was conducted to test the hypotheses. Specifically, an experimental online auction was executed via Veylinx. Members, all from the Netherlands, were contacted
7 through e-mail and invited to participate. This resulted in 512 participants. Participants were shown an advertisement of an innovative product where the warranty framing was manipulated. The manipulation of the warranty framing resulted in four experimental conditions. After bidding, the participants were asked five multiple choice questions. Since the auction was a field experiment, the participants with the highest bids actually bought the auctioned product.
The next section discusses the relevant literature, define crucial concepts. and address relevant research gaps. Following, a conceptual model will be presented and elaborated which results in five hypotheses. Subsequently, the research methodology will be mentioned and explained. Thereafter, the analyses will be described and the results will be presented. The final section will cover the discussion of the results, limitations of the study, alternative explanations for the results, and some concluding thoughts.
8 2. Theoretical framework
This section starts by defining warranty framing. Following willingness to pay will be elaborated. Subsequently, the concept of perceived risk, consisting of five elements, are discussed. Finally, the gaps in current literature that this study contributes to filling, are addressed.
2.1 Warranty framing
Padmanabhan and Rao (1993) define a warranty as a concept that “refers to the coverage provided by the manufacturer with the purchase of the product”. This is also the definition used in this study. An example of products that always includes such warranties are mobile phones. These warranties often say that if the buyer is not to blame for a phones malfunction, the manufacturer has to repair or replace the mobile phone without charging the buyer (much). Although not exclusively, a product warranty is typically provided by the manufacturer of that product. This warranty is a contractual obligation carried by the manufacturer often entered at the same time when the product is sold. This product warranty specifies the standards the product has to meet and establishes liability in case of a malfunction. The source of such a product malfunction may be the inability of the product to
perform its intended function or the premature failure of the product. Such a warranty also specifies the redress available to the customer as compensation (Blischke, 1993). These manufacturer
warranties differ per product in, for example, what they cover, but also the extent and duration of this coverage.
Because people are limited in the amount of information they can process cognitively, marketing campaigns become less effective if they include too much information (Eisenhardt & Zbaracki, 1992). Therefore, it would be ineffective for companies to include their whole warranty policy in their marketing activities. This implies that companies often have to announce and clarify their warranty policies in several words. How they choose to present these warranties, hence, which words they use to describe the manufacturer warranty, is called the warranty frame in this study. When framing a warranty a company can place the emphasis on many different characteristics of the warranty. For example a company could focus on who offers the warranty service to the customer: “Including a manufacturer warranty”. On the other hand a company could also highlight the duration of the warranty: “Including a two year warranty”.
2.2 Willingness to pay
The perceived value of a customer is the monetary amount a customer feels an offering´s benefits are worth. Anderson, Jain and Chintagunta (1992) define customer perceived value in business markets as the “Perceived worth in monetary units of the set of economic, technical, service, and
9 social benefits received by a customer firm in exchange for the price paid for a product offering, taking into consideration the available alternative suppliers' offerings and price”. Willingness to pay then, is a measure of perceived customer value prior to purchase and is also measured in monetary units. This expectation of value explains how much customers want to pay for a product. Homburg, Koschate and Hoyer (2005) define willingness to pay as “the maximum amount of money a customer is willing to spend for a product or service”. This implies that if a consumer perceives a product as more valuable, his or her willingness to pay will increase.
In theory, consumers assess the value of a product by evaluating and adding up all the potential benefits and costs of the purchase. This is what Brent (2007) identifies as a cost-benefit analysis. When the customer perceives the benefits to be higher than the costs, the customer will make the rational decision to buy the product. A cost-benefit analysis is in theory a rational, clear, straightforward and seemingly logical process in which both the buyer and the seller have access to all information which leads to a predictable evaluation. In practice however, this process often deviates from this theory mainly because of two reasons.
Firstly, people have cognitive limitations and are boundedly rational (Eisenhardt & Zbaracki, 1992). In most situations this limits their ability to process all information available and therefore limits their ability to make a fully rational evaluation. This will make the valuation and thus the decision of a customer less predictable and straightforward. To illustrate this, consider a consumer looking to buy a washing machine online. If this consumer was to make a completely rational decision he or she would have to collect all information for all the washing machines offered on the internet, both new and second hand. Since there is an immensely large supply of washing machines online, it is impossible for the consumer to take all information about every washing machine in consideration when deciding. Hence, consumers have cognitive limitations and are not completely rational. If they were, a consumer would always decide to buy the best washing machine at the cheapest supplier. However, since they do not have the capacity to consider all information, consumers make a decision as soon as they perceive to be sufficiently informed and they see a washing machine that satisfies their needs. However, since there are often lots of washing machines that satisfy a consumers’ needs it cannot be predicted which of those washing machines will be bought by the consumer. Hence, the cognitive limitations and bounded rationality reduce the consumers’ predictability. This is confirmerd by Merz, Jon and Fischhoff ´s (1990) study on cognitive limitations and bounded rationality which concludes that the two often have a negative influence on decision making.
Secondly, in practice the process of business is prone to information asymmetry. Multiple studies confirm that there are information asymmetries within markets (Nayyar, 1990; Porter &
10 Millar, 1985) This implies that mostly buyers, but sometimes also sellers, are not able to access all information. A seller may therefore have an information advantage that it could use to its advantage in case the buyer is not aware of the sellers advantage. A seller can benefit from the information advantage by not mentioning characteristics of a product which would lower the product’s value. In this way the seller will be able to sell the product for a higher price compared to a situation of no information asymmetry (Aboody & Lev, 2000). Recall the washing machine example. When a customer is looking to assess the value of a second hand washing machine offered online, he or she takes in account age, visual wear and tear, visual rust, price of alternatives etcetera. The seller, as opposed to the buyer, is familiar with the particular washing machine up for sale. The seller may therefore know that the inside of the washing machine is completely wore down. Because this is not observable to the buyer, the seller has an information advantage which he or she may choose to exploit. This is done by not telling the buyer that the machine is on the fringe of breaking down which would allow the seller to sell the washing machine for a higher price. Hence, information asymmetry can be exploited.
Concluding, cognitive limitations, bounded rationality and information asymmetry influence the customers´ perception of costs and benefits. Furthermore, they also influence the customers´ willingness to pay and cause practice to deviate from the original theory proposed by Brent (2007).
Something else that affects the customers´ perception of value is the customers´ appreciation of free products. The zero-price theory states that the benefits of a product are estimated to be higher when it is free (Shampanier, Mazar & Ariely, 2007). The results of several studies support the zero-price theory. For example Shampaner and Ariely (2006) studied the effects of zero-pricing on chocolates and found that the benefits of free chocolates were perceived as greater, supporting the zero-price theory. Furthermore, Nicolau and Sellers (2011) studied the presence of a zero-price effect amongst tourists by comparing free breakfasts to paid breakfasts, also their results support the theory.
2.3 Perceived risk
When speaking of risk, we have to make the distinction between objective and subjective risk. Objective risk is determined by physical facts and are exogenous. Subjective risk on the other hand is based on perception and are not necessarily dependent on physical facts; subjective risk differs per person (Hansson, 2010).
Customers experience risk when they purchase a product. This risk experienced by
customers, known as perceived risk, is a “subjective expectation of a loss” (Stone & Grønhaug, 1993). Mitchell (1999) defines risk as the “variation in the distribution of possible outcomes, their likelihood
11 and their subjective values”. Perceived risk in this study thus refers to how the consumer assesses the likelihood and value of these possible variations in outcome, as opposed to the actual likelihood, value and distribution of variations. This implies that perceived risk includes a large portion of subjective risk, since these estimates are different for everyone and often not based on facts.
Perceived risk can be predicted by five types of possible consequences. This suggests that there are five different types of perceived risk that together form overall perceived risk: physical, psychological, social, financial, and performance risk (Kaplan, Szybillo & Jacoby, 1974).
Next, each of these five risk types are shortly explained and illustrated by the example of buying a jetski. Perceived physical risk refers to the possibility of obtaining a physical injury or any physical harm because of the assessed behavior, in this case buying a product (Hassan, Kunz, Pearson & Mohamed, 2006). The physical risks a consumer could take into account when buying a jetski are for example, an headache from bouncing on the water, or bruises and broken bones from crashing into the rocks.
The next type of perceived risk is the psychological risk, which is sometimes referred to as ego risk. This comprises of the risk that the buying of a product and the following consequences will have a negative effect on the consumers’ self-perception, self-confidence, ego or peace of mind (Mitchell, 1992). An example of a psychological risk the consumer may experience when buying a jetski is the probability that he or she will not be able to properly control the jetski. This could result in repeatedly falling in the water and a lowered self-confidence.
Third, perceived social risk differs from perceived psychological risk since it is concerned with the image of the consumer in the eyes of its social environment, instead of the self-image of the consumer. Perceived social risk refers to the possible negative influences that the consumption of a product may have on the consumers’ status amongst its social relations. Featherman and Pavlou (2003) define social risk as the “potential loss of status in one’s social group as a result of adopting a product or service, looking foolish or untrendy”. A social risk associated with purchasing a jetski may be that the friends of the buyer perceive that the jetski is primarily bought to brag, something that could harm the buyers reputation.
Financial risk is the fourth type of perceived risk and is primarily concerned with monetary consequences of buying a product. This mainly involves two types of financial costs. First, the initial acquisition price is a factor contributing to the total financial consequences of buying the product. Second, certain products have follow-up costs. These include all the additional expenditures consumers (are forced to) make because of buying the product but excluding the initial purchasing price that was agreed upon. The expectation of the consumer about the financial consequences associated with buying the product, are referred to as perceived financial risk. Consequently, Grewal,
12 Gotlieb and Marmorstein (1994) define perceived financial risk as “the potential monetary outlay associated with the initial purchase price as well as the subsequent maintenance cost of the product”. Financial risks associated with buying a jetski are obviously the purchasing price, the fuel costs, maintenance costs, insurance costs, storage costs, and repair costs in case of a crash or product breakdown.
Finally, the fifth type of perceived risk, performance risk, refers to whether a product functions as promised or expected. This implies that perceived performance risk is concerned with the uncertainty of customers whether the product will deliver the desired and promised
performance and benefits (Mitchell, 1999). If consumers perceive a product to be of inferior quality, by definition they expect this product to be less capable of fulfilling desired and promised
performance, and thus having a higher performance risk as compared to a product that consumers perceive as being of superior quality. Consequentially Grewal, Gotlieb and Marmorstein (1994) define performance risk as “the possibility of the product malfunctioning and not performing as it was designed and advertised and therefore failing to deliver the desired benefits”. When purchasing a jetski, examples of associated performance risks are a jamming motor, sinking, capsizing, stalling, uncontrollability of the jetski, and all other forms of product malfunctions.
According to the signaling theory, consumers will try to reduce the perceived risk associated with the recognition of having incomplete information (Boulding & Kirmani, 1993). They will do this by searching for the missing cues in the form of available prepurchase information that signals value or costs. Examples of signaling information are brand reputation, quality awards or consumer reviews. A negative brand reputation may cause the consumers to make negative inferences about the missing information, hence, it induces a negative signaling effect. The signaling effect of brand equity is amongst others studied by Erdem and Swait (1998), they observed a positive signaling effect and therefore confirm the signaling theory. Also for warranties the signaling theory has been
supported. Boulding and Kirmani (1993) found that warranties can signal value to the consumers, subscribing the signaling theory.
2.4 Gaps in current literature
Although the current literature studied much of the previously explained concepts in detail, there are still gaps that need to be addressed. The first gap concerns warranty framing. Current literature has studied the effects of different warranties (manufacturer warranty versus retail warranty), warranty length, and warranty scope (Boulding & Kirmani, 1993). However, current literature has not studied whether presenting the same warranty in different ways affects the customers’ perception of value. Literature has not even addressed the different ways in which a warranty may be presented.
13 Second, the mediating role of perceived risk has not been sufficiently tested and elaborated in the context of manufacturer warranties. Although current literature implicitly suggests that the relation between a warranty and willingness to pay is mediated by perceived risk, no study has directly assessed this relation. The conceptual model will elaborate this relation further. For now it is only relevant to mention that the studies concerned with this mediation have either another
independent or dependent variable. Often perceived quality and perceived value are used rather than warranties and willingness to pay. Although these concepts are related, we believe that
warranties and willingness to pay are more clear and measureable concepts which would add to the validity of the model. Furthermore, research is needed in different types of risk and how these are related. Even though the different types of risk have been identified, as was mentioned, the relations among them have not yet been studied.
Finally, what is also missing in the literature is a field experiment that studies the influence of manufacturer warranties on consumers´ willingness to pay and identifies the underlying cognitive processes in an actual market environment. There is no current literature yet with an experimental nature that studies this phenomenon in practice. This study tries to contribute to filling these four research gaps. The conceptual model that will be used to fill those gaps will be introduced and elaborated in the following paragraph.
14 3. Conceptual model
According to previous research, warranties signal quality and therefore increase the consumers´ perception of product quality (Boulding & Kirmani, 1993; Wirtz, Kum & Lee, 2000; Erevelles, Roy & Vargo, 1999; Lwin & Williams, 2006). This signaling effect of warranties is because warranties would prove too costly for low-quality sellers to provide. Therefore, warranties signal the seller´s
confidence in the product, because the seller incurs costs if the product does not function as advertised. A warranty thus represents a cue that signals quality. Following, since quality is a valuable characteristic of a product (Zeithaml, 1988; Snoj, Korda & Mumel, 2004), it is implied that warranties increase the consumers´ perceived value of a product and thus the consumers´
willingness to pay. Hence, hypothesis one is set as the following:
Hypothesis 1: The presence of a warranty positively influences the customers´ willingness to pay.
The signaling effect of a warranty however, may differ in strength as to how the warranty is presented. This study assesses the influence of three different possibilities for framing a warranty cue. The first warranty frame is basically no manipulation. The warranty is presented as it is with as complete information as possible and without trying to influence the customers´ perceptions any further than already done by the signaling effect described earlier. For example the warranty is presented as: “Including a two year warranty”.
Secondly, leaving out crucial information such as the duration of the warranty will have negative effects on the effectiveness of its signaling influence. There are two explanations for this negative effect, which Ebenbach and Moore (2000) called the penalty effect. First, consumers find the warranty cue to have less of a signaling effect because consumers perceive information asymmetry which leads to a higher perceived risk and uncertainty as mentioned already. Second, consumers will perceive missing information as negative if they believe that the seller intentionally left out crucial information. This negative cue will negatively affect the overall evaluation of the product or service (Simmons & Lynch, 1991). An illustration of this is framing the warranty as the following: ”Including a warranty”, even though the seller knows that the manufacturer provides a warranty for three months. Compared to the first manipulation, it is expected that this way of presenting a warranty will have less positive consequences on the consumers´ perception of product and service value. On the other hand, there still is a warranty signaled by the seller. Therefore it is expected that within this frame the consumers´ perception of product and service value will be higher than when no warranty is mentioned.
15 Finally, according to the zero-price theory, consumers perceive a product as more valuable if it is free. This theory argues that consumers perceive not only the costs of a free product to be lower, but also the benefits of a free product to be higher than for products with a positive price, no matter how small it is. This is explained by the positive effect that the free product prompts (Shampanier, Mazar & Ariely, 2007). Sellers are able to insinuate their warranties to be free of charge which will thus lead to a higher valuation of the warranty. One way in which this can be done is by presenting a standard three month warranty as: “Including a one year warranty, extended to two years for free”. Hence, it is expected that framing an additional warranty cue as being free of charge will signal a more valuable warranty and consumers will thus value the product more than when this zero-price additional warranty insinuation is absent. This results in the second hypothesis:
Hypothesis 2: (A) Framing a warranty cue as having additional zero-price benefits influences the customers´ willingness to pay more positively than framing a warranty as it is, (B) which influences the customers´ willingness to pay more positively than intentionally withholding information about the warranty.
However, the influence of the warranty cues mentioned in the first two hypotheses on the consumers’ willingness to pay may not be a direct influence. As mentioned in the previous section, the signaling theory states that if a cue signals value, this cue will reduce the consumers´ perceived risk. Since it is argued earlier that quality, which is signaled by a warranty, is a valuable cue, it may be assumed that a warranty decreases the consumers´ perceived risk (Bearden & Shimp, 1982;
Sweeney, Soutar & Johnson, 1999; Wirtz, Kum & Lee, 2000).
Recall that perceived risk is defined as the subjective expectation of a loss. This implies that increased perceived risk or uncertainty associated with a product will lead the customer to estimate the potential costs of a purchase higher, or the potential value of a purchase lower than it would be estimated without perceived risk or uncertainty. The cost benefit analysis often (implicitly) used by consumers in a purchasing decision as mentioned before will therefore be more negative if perceived risk or uncertainty is present. A decrease in this perceived risk will therefore be beneficial for the evaluation of the product´s value (Sweeney, Soutar & Johnson, 1999; Erevelles, Roy & Vargo, 1999). This implies that a customer is willing to pay a higher price and illustrates that the consumers´ perceived risk affects the consumers´ willingness to pay. Concluding, it is this decrease in the consumers´ perceived risk caused by the warranty, which increases the consumers´ willingness to pay. The effect of warranties on willingness to pay is thus indirect. Hence, hypothesis three is set as the following:
16 Hypothesis 3: The positive relation between a warranty and the consumers´ willingness to pay is mediated by perceived risk of the consumer.
However the influence of warranties on perceived risk may be specified even further. The influence of a warranty on a consumer’s perceived risk will be mainly attributable to two of the five types of perceived risk: the performance and the financial risk (Sweeney, Soutar & Johnson, 1999). It is expected that the presence of a warranty will decrease the perceived performance risk of a consumer. The reason for this has been explained earlier: warranties signal quality. Increased quality will by definition reduce performance risk. This leads to the fourth hypothesis:
Hypothesis 4: The presence of a warranty lowers the customers´ perception of performance risk.
To explain why the presence of a warranty reduces perceived financial risk, the relation between performance and financial risk has to be elaborated. By definition if a product has a high performance risk, it will have a higher chance of not fulfilling its desired and promised benefits. If a product does not sufficiently fulfill its intended, the consumer will have to deal with these
consequences. In case of a product breakdown or another malfunction these consequences are paying to repair the product or buying a new one, both of which can be considered follow-up costs and thus a financial consequence. By this reasoning it can be concluded that performance risk leads to financial risk (Featherman & Pavlou, 2003). This implies that when the presence of a warranty decreases the performance risk, it will subsequently reduce the financial risk. This results in the fifth hypothesis:
Hypothesis 5: The customers’ perceived performance risk positively influences the customers´ perceived financial risk.
17 Warranties + ++ +++ Perceived risk Perceived performance risk Perceived financial risk Withholding information Complete information Zero-price benefits Willingness to pay Figure 1: Conceptual model
18
No warranty cue Withholding
information “Including a warranty” Complete information “Including a 2 year warranty” Zero-price benefits “Including a 1 year warranty, extended to 2 years free of charge” 4. Methodology
4.1 Sample
Of the contacted Veylinx-members a total of 512 participated. Since Veylinx is a Dutch website, all members are Dutch speaking and are mostly from The Netherlands. Of the participants 44% was male and 56% female. The mean age was 43.81 (SD=14.96). Incomplete and impossible values were removed from the data before analysis. An example of an impossible value is an age of 103 years. Missing data was excluded for analysis.
4.2 Design
The independent, categorical variable of warranty framing consists of four categorical manipulations that represent the four experimental conditions of this study (Figure 2).
Figure 2: Four experimental conditions.
The first manipulation is not mentioning anything about the presence of a warranty in the auction. This frame includes an empty section in the same place where the others have a warranty cue. The second treatment consists of presenting some information about the warranty (that it is existing), but intentionally leaving out some crucial information as well. This treatment will show the following information to the bidder: “Including a warranty”. The third treatment frames the warranty
19 information as accurately as possible, showing as much of the relevant details possible to the bidder: “Including a 2 year warranty”. The final treatment frames the same warranty as having an additional zero-price benefit. This is operationalized by showing the following information to the bidder: “Including a 1 year warranty, extended to 2 years free of charge”. The advertisement used for the third treatment is shown below in Figure 3.
20 The auctioned product is a waterproof Bluetooth speaker with a suction-cap so it can stick to any flat surface. The product is able to connect to smartphones and enables the user to engage in hands free phone conversations. This product may be classified as an innovative product. This implies that the product is relatively new and by definition differentiable which inherently leads to a high uncertainty among consumers evaluating the product (Fisher, 1997). This choice to auction an innovative product as opposed to a functional product was made because it is expected that the high uncertainty associated with these products by definition means that the consumers have a higher perceived risk. This provides a more substantial opportunity to decrease this risk by framing the warranty, and thus increases the possibility of significant results. This specific innovative product was chosen because of its product characteristics of being a speaker but also waterproof. This
combination is likely to be looked at with scepticism because many consumers associate the
combination of water and electronic devices with short circuits and product malfunctions. Therefore many consumers will experience a relatively high perceived uncertainty when evaluating this
product.
4.3 Method
Data was collected through the means of an experimental auction. This is a suitable way to analyze and test whether the proposed conceptual model holds in practice, since it provides the researchers with the ability to both manipulate the variables to a certain extent but at the same time observe real behavior. To be more specific, the method used was a sealed-bid, second-price online auction. This has been used in previous studies and has been hypothesized to be the most accurate way of measuring true willingness to pay (Geueke, 2013; Groot, 2013; Sintnicolaas, 2014; Dekkers, 2013). Veylinx was used as the platform for this auction.
4.4 Procedure
In the morning of the auction day Veylinx´s members received an invitation per mail to participate in the auction. Members that decided to participate were randomly assigned to one of the four
treatments and were shown the rules and regulations of Veylinx before participating. All treatments showed the same image to the participants, except for the manipulated phrase. Following,
participants had six minutes to look at the advertisement and place a bid. Following, the participants were asked to confirm their bids to avoid biases. After bidding, the participants were asked to answer five questions. Four of these are concerned with measuring the perceived risk, and one was a
21 manipulation check for the warranty cues. At the end of the auction day, all participants were mailed to inform them about the highest and second-highest offers, and whether or not they had won.
4.5 Measurements
4.5.1 Willingness to pay
The main dependent, continuous variable in this experiment is willingness to pay. This is measured by the sealed-bids made on the product by the participants. A high offer represents a high willingness to pay.
4.5.2 Perceived risk
Perceived risk is the mediating variable in this study and will be measured on the basis of a four-item scale. These four items measure the performance and financial types of perceived risk for two different time spans and are adapted from Sweeney, Soutar and Johnson (1999). These two types of risk are measured because these are expected to be the affected and influential types. Besides, methodological, time, and money constraints make it impossible to measure, check, and exclude the influence of the three other types. These two types of perceived risk are measured for two different time spans to analyze possible differences. The first time span is concerned with the coming two years, which is the period that is covered by the warranty. The second time span covers five years which is thought to be the expected lifetime of an average electronics product (Babbitt, Kahhat, Williams & Babbitt, 2009). This results in the following measures of perceived performance and financial risk.
4.5.2.1 Perceived performance risk
Perceived performance risk will be measured by asking the following two questions for which scales range from (1) 0-20%, (2) 20-40%, (3) 40-60%, (4) 60-80%, to (5) 80-100%: “How high do you
estimate the chance that the speaker will not function properly anymore within two years?” and “How high do you estimate the chance that the speaker will not function properly anymore within five years?”
4.5.2.2 Perceived financial risk
Perceived financial risk will be measured by asking the following two questions for which scales range from (1) 0-20%, (2) 20-40%, (3) 40-60%, (4) 60-80%, to (5) 80-100%: “How high do you estimate the chance that you will spend additional money on the speaker within two years on, for example,
22 repairs?” and “How high do you estimate the chance that you will spend additional money on the speaker within five years on, for example, repairs?”.
4.5.3 Manipulation check
The manipulation of the warranty cues will be checked by asking the following question: “Do you think that you have a warranty on the previously auctioned speaker and if you do, for how long?”. Scales ranged from (1)none, (2) one year, (3) two years, (4) three years, to (5) longer then three years.
The manipulation check makes it possible to assess whether the experimental conditions influenced the participants’ perceptions of a warranty in the predicted ways. As Figure 4 illustrates, it seems that the participants´ perception of the presence of a warranty does not differ substantially across the experimental conditions, hence, warranty frames. In the condition where no warranty cue was shown, only 18.8 percent of the participants thought that no warranty was present. When there was a warranty cue presenting limited information, 13.9 percent of the participants perceived to obtain no warranty. 11.9 Percent of the participants in the complete information condition thought there was no warranty service included. When the warranty was presented as having additional benefits, 10.9 percent perceived no warranty to be included. Unexpectedly, these percentages seem to indicate that the participants perceived a warranty to be present in all experimental conditions, even the no warranty condition (81.2 percent).
23 To test these presumptions a t-test was performed for each of the conditions. This t-test measured whether the participants in a certain condition perceived a warranty. These t-tests showed that participants indeed perceived a warranty in all conditions. Hence, participants perceived a warranty in the no warranty condition (t(132)=33.22, p<0.001), in the withholding information condition (t(107)=32.88, p<0.001), the complete information condition (t(117)=34.25, p<0.001), as well as in the zero-price benefits condition (t(128)=39.67, p<0.001).
These findings are unexpected because it was predicted that the no warranty condition would lead the participants to perceive no warranty. Because participants apparently experienced a warranty cue even in the no warranty condition, which led them to perceive a warranty, it is not valid to use the participants in the no warranty condition to represent the group that was not provided with a warranty cue in further analyses. However, further analyses showed that there are no differences in willingness to pay and perceived risk between the participants that perceived no warranty even though they were assigned to different experimental conditions. This implies that the participants that perceived no warranty from all four experimental conditions may be grouped together for the analyses. Also all participants that did perceive a warranty may be grouped together since neither their bids nor their answers are influenced by the experimental condition they were assigned to. Therefore further analyses will use the participants’ answers on the manipulation check
Figure 4: Relation between warranty frames and perceived warranty
0 20 40 60 80 100 120 140
No warranty cue Withholding information Complete information Zero-price benefits
P ar ti ci p an ts No warranty Warranty
24 question to distinguish whether a participant perceived a warranty or not. Further analyses will thus analyze the consequences of the participants’ warranty perceptions.
4.5.4 Other control variables
Other participant related control variables such as for example age, education, bid duration, marital status and personal preferences are obtained from the Veylinx database.
25 5. Results
Before the main analyses are addressed, we check whether the experimental conditions yielded the expected results. Following, we inspect the correlations between the main variables to give an early indication of the possible effects that we might find during the main analyses. Then after presenting the main analyses, this section ends with describing additional analyses.
Out of the four treatments the participants in the no warranty condition show the highest willingness to pay (M=835.30, SD=797.937) followed by the withholding information condition (M=781.36, SD=819.736). Third, the zero-price benefits condition shows a willingness to pay of on average 780 Eurocents (M=780.46, SD=772.988). On average the participants in the complete information condition had the lowest willingness to pay (M=762.21, SD=843.205). These means can also be found in Table 1.
Table 1: Willingness to pay in Eurocents per treatment
N M SD
No warranty cue 137 835.30 797.937 Withholding information 111 781.36 819.736 Complete information 132 762.21 843.205 Zero-price benefits 132 780.46 772.988
Recall that the perceived risk was measured for two time spans (two years and five years), and for two different types of risk (performance risk and financial risk) which resulted in four questions. The frequencies of the participants´ answers on these four questions can be found in Table 2. The question concerned with the two year perceived performance risk, is most frequently answered with 40-60% by the participants (N=484).The question about with five year perceived performance risk however, is answered most frequently with 80-100% by the participants (N=485). Furthermore, the participants answered the question concerned with two year perceived financial risk most frequently with 0-20% (N=485). The final risk question about five year perceived financial risk, was also most frequently answered with 0-20% by the participants (N=485).
26 Table 2: Answers to risk questions frequencies (mode)
N 0-20% 20-40% 40-60% 60-80% 80-100%
Two year perceived performance risk 484 69 132 (143) 95 45 Five year perceived performance risk 485 14 50 89 123 (209) Two year perceived financial risk 485 (251) 107 80 24 23 Five year perceived financial risk 485 (169) 77 90 60 89
The correlations between most of the different main variables are shown in Table 3.
Table 3: Descriptives and correlations between the variables
N M SD 1 2 3
1 Willingness to pay (Eurocents) 512 790.62 806.44
2 Perceived warranty 488 2.32 .76 .10**
3 Perceived risk two years 488 2.36 .99 -.16*** -.16***
4 Perceived risk five years 487 3.30 1.10 -.08* -.01 0.74***
Note: * p<.10, ** p<.05, *** p<.001.
Surprisingly, not all variables correlate with each other since the independent variable perceived warranty does not correlate with the mediator variable perceived risk five years, r(485)=-0.01, ns. Also the variable perceived risk five years and dependent variable willingness to pay are found to be only marginally negatively correlative, r(485)=-0.08, p=0.087. However, the variable perceived warranty correlates positively with variable willingness to pay, r(486)=0.10, p=0.034, as predicted. Also expected was the found positive correlation between the two mediator variables perceived risk two years and perceived risk five year, r(485)=0.74, p<0.001. Moreover, consistent with the predictions, the variable perceived warranty correlates negatively to mediator variable perceived risk two years, r(486)=-0.16, p=0.001. Finally, the variable perceived risk two years correlates negatively to the variable willingness to pay, r(486)=-0.16, p<0.001, as expected.
5.1 Main analyses
The main analyses are divided into five sections that correspond to the five hypotheses.
27 To test if the perception of a warranty is related to the participants’ willingness to pay, an ANOVA analysis is conducted. This ANOVA indicates that the two variables are related, F(1, 486)=7.57, p=0.006, which is conform predictions. This means that for the participants the perception of a warranty predicts the willingness to pay. Since a positive correlation was found, this outcome implies that participants who think they have a warranty will have higher bids (Field, 2013). Figure 5 shows the effect of a warranty on willingness to pay graphically.
5.1.2 Relation between framing warranties and willingness to pay
A second ANOVA analysis is executed to check whether framing the warranties predicts the
participants’ willingness to pay. Unexpectedly, it shows that none of the four warranty presentations resulted in a different willingness to pay compared to any of the other warranty frames since the ANOVA does not reach significance, F(3, 508)=0.206, ns. This outcome is supported by the post-hoc LSD analysis that confirms that there is no reason to reject that all warranty frames show similar willingness to pay. This means that none of the used warranty frames predicts the willingness to pay for the participants (Field, 2013). This is illustrated in Figure 6 where it can be observed that there are only small differences in willingness to pay that are by far not substantial enough to reach significance.
Figure 5: Relation between perceived warranty and willingness to pay 0 100 200 300 400 500 600 700 800 900 No warranty Warranty W ill in gn es s to p ay ( Eu ro ce n ts )
28 5.1.3 Mediation of perceived risk
In the analysis of perceived risk as a mediator in the relationship between perceived warranty and willingness to pay we will separately study the two different time spans of perceived risk to assess if there are any differences. First, four models each consisting of one linear regression analysis will study the possible mediation of perceived risk two years. Following, a similar second set of four models consisting of one linear regression analysis each will analyze the possible mediation of perceived risk five years. Analyzing a mediation using four models is conform the method described by Baron and Kenny (1986). This implies that in such a four model analysis, the first regression analysis will study the effect of the independent on the dependent variable. Following, the second regression analysis will assess the relation between the independent and the mediator variable. Furthermore, the relation between the mediator variable and the dependent variable will we measured by the third regression analysis. Finally, the fourth regression analysis will study the effect of both the independent and mediator variable together on the dependent variable.
The first regression analysis presents results that are conform expectations of the
relationship between perceived warranty and willingness to pay. Model 1 reveals a positive relation (ß=0.096, p=0.034, R²=0.009) between perceived warranty and willingness to pay with an explained
Figure 6: Influence of the warranty frames
0 100 200 300 400 500 600 700 800 900
Withholding information Complete information Zero-price benefits
W ill in gn es s to p ay ( Eu ro ce n ts )
29 variance of 0.9 percent. This implies that participants who perceived that a warranty was present had a higher willingness to pay, which supports the conclusion of the first ANOVA (Baron & Kenny, 1986).
The second regression analysis, with 2.4 percent explained variance, indicates a negative influence (ß=-0.155, p=0.001, R²=0.024) of perceived warranty on perceived risk two years as predicted. These results from Model 2 reveal that respondents who experienced the presence of a warranty thought to be exposed to less product related risk in the coming two years (Baron & Kenny, 1986).
With 2.5 percent explained variance, Model 3 confirms the expectation that perceived risk two years has a negative effect on willingness to pay (ß=-0.159, p<0.001, R²=0.025). The third regression analysis thus indicates that respondents who feel exposed to more product related risk in the coming two years, have made lower bids (Baron & Kenny, 1986).
The last model which studies the mediation of perceived risk two years, Model 4, shows that there is no longer a relation between perceived warranty and willingness to pay (ß=0.073, ns) while the effect of perceived risk two years on willingness to pay remains strong (ß=-0.147, p=0.001, R²=0.030). This fourth regression analysis has an explained variance of 3 percent, verifies the predictions and provides evidence for the full mediation model. This means that participants’
perception of the presence of a warranty is not directly related to the height of the participants’ bids. Instead, participants who perceive a warranty experience less product related risk. It is this decrease in perceived product related risk, and not the experienced presence of a warranty, that predicts the higher bids of those participants (Baron & Kenny, 1986). The first set of four models that tested the mediation of perceived risk two years and their statistics can be found in Table 4. Figure 7 graphically presents the main results from the first set of four linear regression analyses.
30 Perceived warranty Willingness to pay Perceived risk two years
Table 4: Regression results for the first set of four models
Model 1 Model 2 Model 3 Model 4
Dependent variable
Willingness to pay
Perceived risk two years
Willingness to pay
Willingness to pay
Coefficient SE Beta Coefficient SE Beta Coefficient SE Beta Coefficient SE Beta Constant
566.968*** 115.743 2.828*** .143 1102.296*** 92.270 902.437*** 154.156
Perceived warranty 100.838** 47.430 .096 -.202*** .058 -.155 76.872 47.543 .073
Perceived risk two years -127.739*** 36.068 -.159 -118.610*** 36.448 -.147
R²
.009 .024 .025 .030
Note: N=487, * p<.10, ** p<.05, *** p<.001.
Figure 7: Mediation of perceived risk two years
Note: N=487, * p<.10, ** p<.05, *** p<.001. ß = 0.073 ß = -0.147*** ß = -0.155*** ß = -0.159*** ß = 0.096** Model 4 Model 4 Model 3 Model 2 Model 1
31 However, when we study the possible mediation of the perceived risk five years, hence, the second set of four models, the results are different and not as predicted. The fifth regression analysis is a copy of the first regression analysis, since also for this second mediation analysis perceived warranty is the dependent variable and willingness to pay remains the dependent variable. Therefore Model 5 is exactly the same as Model 1, which is why Model 5 will not be elaborated further.
Unexpectedly, the sixth regression model presents results different from the predictions. Model 6 reveals that there is no relation (ß=-0.005, ns) between perceived warranty and perceived risk five years. This means that participants who thought to be provided with a warranty do not show a higher nor lower expectancy of product related risk for the coming five years (Baron & Kenny, 1986).
The seventh regression analysis, with an explained variance of 0.6 percent, shows a marginally negative influence (ß=-0.078, p=0.087, R²=0.006) of perceived risk five years on willingness to pay as predicted. These outcomes from Model 7 imply that respondents who
perceived to be prone to a lot of product related risk in the coming five year made lower bids (Baron & Kenny, 1986).
Finally, Model 8 indicates with an explained variance of 1.5 percent that the relation
between perceived warranty and willingness to pay remains significant (ß=0.095, p=0.035, R²=0.015) while the effect of perceived risk five years on willingness to pay is only marginally significant (ß=-0.077, p=0.088, R²=0.015). This contradicts the expectations since it indicates that perceived risk five years does not (fully) mediate the relation between perceived warranty and willingness to pay (Baron & Kenny, 1986). Perceived risk two years is thus the only identified mediator within this relation. Table 5 presents the second set of four models that tested the mediation of perceived risk five years and their statistics. The main findings are also graphically represented in Figure 8.
32 Perceived warranty Willingness to pay Perceived risk five years
Table 5: Regression results for the second set of four models
Model 5 Model 6 Model 7 Model 8
Dependent variable Willingness to pay Perceived risk five years Willingness to pay Willingness to pay
Coefficient SE Beta Coefficient SE Beta Coefficient SE Beta Coefficient SE Beta Constant
566.968*** 115.743 3.314*** .160 987.117*** 114.227 753.658*** 158.620
Perceived warranty 100.838** 47.430 .096 -.007 .066 -.005 100.142** 47.389 .095
Perceived risk five years -56.287* 32.866 -.078 -55.954* 32.750 -.077
R²
.009 .000 .006 .015
Note: N=486, * p<.10, ** p<.05, *** p<.001.
Figure 8: Mediation of perceived risk five years
Note: N=486, * p<.10, ** p<.05, *** p<.001. ß = 0.095** ß = -0.077* ß = -0.005 ß = -0.078* ß = 0.096** Model 8 Model 8 Model 7 Model 6 Model 5
33 5.1.4 Relation between perceived warranty and perceived performance risk
To assess whether the participants’ warranty perception is related to their perception of
performance risk a third ANOVA analysis is conducted. Conform predictions, the analysis reveals that perceived warranty and perceived performance risk are related, F(1, 482)=10.567, p=0.001. Because two variables are negatively correlated, r(482)=-0.146, p=0.001, these results mean that the
participants who experienced the presence of a warranty, perceived the performance risk of the product to be lower (Field, 2013).
5.1.5 Relation between perceived performance risk and perceived financial risk
A ninth linear regression analysis is executed to study if the participants perceived performance risk is related to their perceived financial risk. The analysis, with an explained variance of 19 percent, indicates that perceived performance risk predicts perceived financial risk (ß=0.436, p<0.001,
R²=0.190) which was expected. This implies that participants who experienced a high product related performance risk also perceived the product related financial risks to be high (Field, 2013).
5.2 Additional analyses
Besides the hypothesized relationships between warranties and the two types of perceived risk, it may be that warranties affect the relationship between these two types. This is suggested because it is likely that the perception of a warranty will cause the customer to experience low financial risk even though the performance risk is high. This is because the customer does not have to pay for repairs or replacements if a warranty is granted. When a warranty is perceived to be absent however, the customer has to pay for solving performance issues itself and thus the relation between performance risk and financial risk is expected to be stronger. To analyze this supposedly negative moderation a set of two models, consisting of one linear regression analysis each, is studied following Baron and Kenny’s (1986) guidelines. The first of these models studies the main effects of the independent and moderator variables on the dependent variable. Next, the second model analyses the influence of the interaction effect of the independent and moderator variables on the dependent variable.
Model 9, with an explained variance of 19.2 percent, reveals a main effect for perceived performance risk (ß=0.429, p<0.001, R²=0.192) but not for perceived warranty (ß=-0.045, ns). It is thus only perceived performance risk that affects perceived financial risk directly and does this in a positive way. This confirms earlier findings and indicates that participants who perceived high performance risk, experienced higher financial risk (Baron & Kenny, 1986).
34 Unexpectedly, Model 10 shows a significant positive interaction effect of perceived
performance risk and perceived warranty where a negative interaction effect was predicted. However, this does verify the moderating effect of perceived warranty on the relationship between perceived performance risk and perceived financial risk. This means that for participants who believed a warranty was present, an increase in perceived product related performance risk led to a relatively larger increase in perceived product related financial risk, than it did for participants experiencing no warranty presence. Furthermore, Model 10 (R²=0.204) explains more variance than model 9 (R²=0.192)(Baron & Kenny, 1986). This final set of two models that analyzed the moderation of perceived warranty and their statistics can be found in Table 6. The main findings are also
graphically represented in Figure 9, 10 and 11.
Figure 9: Moderation warranty in relation performance and funtional risk
0 5 10 15 20 25 30 35 40 45 50 No warranty Warranty P er ce iv ed c h an ce o f ad d iti o n al c o sts w ith in tw o y ea rs (p er ce n ta ge )
Low functional risk High functional risk
35 Perceived performance risk Perceived financial risk Warranty Perceived performance risk Perceived financial risk Warranty
Table 6: Regression results of main (Model 9) and interaction (Model 10) effects of Perceived performance risk two years and Perceived warranty on Financial risk two years
Financial risk two years (DV) Model 9 Model 10
Coefficient SE Beta Coefficient SE Beta
Constant
1.886*** .047 1.904*** .047
Perceived performance risk two years
.415*** .040 .429 .421*** .040 .435
Perceived warranty
-.147 .136 -.045 -.252* .140 -.077
Perceived performance risk two years*Perceived warranty .300** .109 .117
R²
.192 .204
Note: Dependent variable is Financial risk two years, N=480, * p<.10, ** p<.05, *** p<.001.
Figure 10: Model 9: main effects Figure 11: Model 10: interaction effect
Note: N=480, * p<.10, ** p<.05, *** p<.001. Note: N=480, * p<.10, ** p<.05, *** p<.001. R² = 0.192 ß = -0.045 ß = 0.429*** R² = 0.204 ß = 0.117**
36 6. Discussion
This study aimed to contribute to the literature concerning warranty cues. It was especially designed to predict more accurately how a warranty affects consumers’ perceptions and willingness to pay. The discussion starts by addressing the key findings and hypotheses. Following, the main limitations and alternative explanations are mentioned. Next, the theoretical and practical value of this study are discussed and the discussion ends with some concluding thoughts.
6.1 Key findings
Overall, the majority of the findings conform to the stated hypotheses. The hypothesis that stated that the presence of a warranty positively influences the willingness to pay of customers was supported. It turned out that the two are related and that customers who think that a warranty is present will exhibit a higher willingness to pay. This supports the explained theories and earlier findings (Boulding & Kirmani, 1993; Wirtz, Kum & Lee, 2000; Erevelles, Roy & Vargo, 1999; Lwin & Williams, 2006; Zeithaml, 1988; Snoj, Korda & Mumel, 2004).
The second hypothesis stated that warranty framing influences the consumers’ willingness to pay. Specifically, framing zero-price benefits was hypothesized to be more positively related to willingness to pay than framing a warranty as it is, which would be more positively related to willingness to pay than intentionally withholding information when framing a warranty. This hypothesis was rejected since none of these warranty frames influenced the customers’ willingness to pay more strongly than any of the other frames. These unexpected outcomes may have three possible causes.
One possible reason is that the manipulation check revealed that the manipulation has failed. This may have affected the results concerned with this hypothesis, and therefore it may have caused the rejection of the hypothesis. Also, the lack of effect of the warranty frames may be explained by the fact that warranties are usually not one of the core attributes that create value for the customer. Therefore, warranties explain a relatively low percentage of the total product value, which reduces the chance that a significant change in perceived total product value, and thus willingness to pay, will be observed between the different warranty frames. However, when the perceived value of just the warranty is measured, it may be possible that the zero-price and penalty effects are observed. Unfortunately, the methodology of this study did not allow that to be tested.
A third possible reason for the rejection is that the suggested warranty frames of zero-price benefits and withholding information do not influence the customers’ willingness to pay any stronger nor weaker compared to when a warranty is presented as it is. This implies that warranties are not
37 prone to the zero-price and the penalty effects. Lee and Olshavsky (1997) suggest that customers do not automatically assume the worst-case scenario when they encounter a scenario of missing information, something alleged by the penalty effect. Instead, customers make inferences about the information based on available brand information. This implies that if customers trust, or have a positive attitude towards in this case for example Veylinx or Freecom, that the missing information is not perceived as negative, which might explain why the penalty effect is not observed for warranties.
The third hypothesis stated that perceived risk fully mediates the relationship between a warranty and the willingness to pay of customers. This hypothesis was partially confirmed. The full mediation of perceived risk was only found within the warranty period, which implies that the customers’ willingness to pay is indirectly, and not directly, related to the presence of a warranty. Hence, a warranty lowers the customers’ perceived product related risk only for the period covered by the warranty. It is this reduction in the customers’ perceived risk within the warranty period, and not the warranty itself, which heightens the customers’ willingness to pay.
Although only partially expected, this conclusion seems logical since the warranty provides the customer with the valued risk reducing benefits of being able to hold the manufacturer liable only within the warranty period. Thus, when the warranty has expired, the customer will not be able to benefit from this risk reducing manufacturers liability, which would imply that the perceived risk after the warranty period is not affected by the warranty and does not mediate the relationship. This may explain why the mediation was not found for the perceived risk during the entire product’s life-cycle since this includes also the portion of perceived risk that is not affected by the warranty, which clouds the earlier found mediation. Concluding, these findings are conform expectations and support the theory and current literature (Bearden & Shimp, 1982; Sweeney, Soutar & Johnson, 1999; Wirtz, Kum & Lee, 2000; Erevelles, Roy & Vargo, 1999).
Hypothesis four was concerned with the supposedly negative relation between a warranty and the customers’ perceived product performance risk. This hypothesis was confirmed because customers who believe a warranty is present, estimate the chance of a product to malfunction to be lower. The results thus support the proposed theory.
Also the fifth and final hypothesis was confirmed. This supports the proposed theory that was based on Featherman and Pavlou´s study (2003). This hypothesis stated that customers’ perceived product performance risk relates positively to the customers’ perceived product related financial risk. It turned out that customers who believe there is a high chance of product failure, also expected the product to cause more future costs.
Furthermore, the study yielded an interesting additional finding. A warranty was found to moderate the relation between the customers’ perceived product performance risk and the
38 customers’ perceived product related financial risk. A negative moderation of a warranty in this relation seemed logical and was to be expected since a warranty’s primary function is to lower the customers financial risk especially when the performance risk is high. However, the found
moderation turned out to be the opposite: positive, which on first sight seems not logical. When the typical procedure of actually using the warranty service in case of product malfunction is assessed more thoroughly, it shows that there are at least two groups of costs associated with acting upon a warranty.
The first group consists of time and opportunity costs, which are related. When a product does not function properly, a customer has to look up the warranty and read the conditions to check whether the warranty service is indeed provided to solve the experienced problem. Following, the customer has to gather receipts and the other needed paperwork and prepare a package to send to the manufacturer. This process may consume a fair amount of time from the customer, time which the customer might have otherwise spent doing something he or she values more. This wasted time thus represents an opportunity cost. Furthermore, when the package is sent, it usually takes a couple of weeks before the customer is provided with the repaired or new product. Within these weeks the customer is not able to use the product and thus cannot enjoy its provided value and benefits. When the customer does not have the warranty, he is often forced to buy a new product, which can typically be done within one day. Again, this time the customer is not able to benefit from the product thus also represents an opportunity cost.
The second group of costs represents purely monetary costs associated with the warranty. Although the warranty is intended to reduce the costs of the customer, there are also costs associated with acting upon a warranty. First of all, the product has to be shipped back to the manufacturer, which will require the customer to pay the shipping costs. Besides, since a customer often does not know what is wrong with the product and a warranty typically does not cover all types of product breakdown, there is always the risk that the manufacturer’s problem analysis reveals that the customer still has to pay a fee for repairing the product. Consumers apparently take these costs into account when they have a warranty and perceive the performance risk to be high while these costs are not relevant when there is no warranty provided. This explains why a warranty causes the relation between perceived performance risk and perceived financial risk to increase in strength.
As already briefly mentioned, the described results of this study mostly confirm current theories and literature. This is because the conceptual model, based on these current theories, is largely supported. Namely, the results confirmed the often mentioned relations between warranties, perceived risk, and willingness to pay (Bearden & Shimp, 1982; Sweeney, Soutar & Johnson, 1999; Wirtz, Kum & Lee, 2000; Erevelles, Roy & Vargo, 1999). Also the literature suggests relations between
