An Application of Cluster Analysis to the
Understanding of Consumers’ Financial Services Behaviour
by
Robert Hamilton[1], Paul Hewer[2] and Barry Howcroft[3]
September 2000
Acknowledgements
The authors would like to thank all those consumers who took part in the research, and the financial assistance provided by NCR’s Knowledge Laboratory, which made this research possible.
ABSTRACT
1. Introduction
The market for financial services has become increasingly competitive with unprecedented levels of new entrants and developments in delivery channel strategies. In this ever-changing climate financial providers are seeking to create sustainable competitive advantage. One possible solution focuses upon the concept of ‘customer value’ (Woodruff, 1997) which involves financial providers adopting an external orientation and seeking to understand more fully the nature of their customers. A critical feature of this approach is the attempt to not only anticipate, but also to influence and determine consumers’ buying behaviour. Accordingly this paper outlines a conceptual model for understanding consumer behaviour in the context of financial services. This model is subsequently tested and appraised through the use of cluster analysis procedures. Section 2 of the paper provides a review of the relevant literature and develops the conceptual model. Section 3 outlines the research methodology, Section 4 assesses the findings from the study and Section 5 draws some conclusions.
2. Understanding Consumer Behaviour and Financial Services
The need to further academic understanding of consumer behaviour within the financial services sector has become a critical issue for a number of reasons, not least being the impact of deregulation and the emergence of new forms of technology. At the same time academics have suggested that a conceptual framework is currently lacking which is able to explain how consumers purchase financial services (McKechnie, 1992; Betts, 1994).
A review of the literature on consumer-buyer interactions (Bateson, 1989; McKechnie, 1992; Harrison, 1994; Ennew and McKechnie, 1998) suggests that two factors principally motivate and determine individual purchasing or contracting choices, namely involvement and uncertainty. Consumer “involvement” in the buyer-seller interchange incorporates a number of subsets: customer control (Bateson, 1989), customer participation and level of contact (Chase, 1978). Chase (1978), for example, offered a framework to distinguish between services based upon level of contact during the service delivery stage. Consumer uncertainty has been analysed in terms of consumers’ perceptions of risk and the research suggests that this is ultimately related to the characteristics of the product (Shostack, 1977; Ennew and McKechnie, 1998). Such characteristics directly influence consumer uncertainty by determining their perceptions of complexity associated with different financial products (Harrison, 1994).
By placing consumer involvement and consumer uncertainty onto a simple continuum running from high to low, it is possible to construct a two dimensional matrix of consumer behaviour (see Figure 1). This matrix describes the purchasing/contracting alternatives available to financial service consumers and, as such, reflects a range of interaction modes between financial service providers and their customers. Each quadrant represents a different combination of involvement and uncertainty (referred to as consumer confidence in the matrix) and, therefore, different modes of banker-customer interactions when purchasing different types of financial product.
INSERT FIGURE 1
These considerations are significant because the limitation of previous consumer behaviour models (Nicosia, 1966; Engel, Kollat, and Blackwell, 1968; Howard and Sheth, 1969; Bettman, 1979) stems from the factors hypothesised to explain consumer motivation and behaviour, particularly in the creation and maintenance of relationships. Previous models have tended to focus on the extensive role of information in shaping and directing rational choice. Nicosia (1966), for example, identifies the search and evaluation of information; Howard and Sheth (1969) link the amount of information held with the extent of learning and thus problem solving; Engel, Kollat and Blackwell (1968) identify information input and processing as central elements in decision-making; Bettman (1979) recognises that individuals have limited capabilities for processing information and this limitation affects their decision-making activities. The common causal link in all of these approaches can be summarised as: information attitude purchase. As an explanation of behaviour this form of linkage is not only predominant in the social sciences, but can also be clearly identified in behavioural economics (Katona, 1960; Scitovsky, 1976). All of these models are based on information which is necessarily assumed to be freely and readily available with little reference made to the context in which information is found and used. Academics have thus argued that such models are largely unverifiable empirically because they do not offer testable hypotheses (Tuck, 1976; Foxall, 1991). These earlier models also ignored consumer involvement with the purchase and the consumer’s ability or confidence when making purchase decisions (Morgan and Hunt, 1994). Their specific applicability to the field of financial services can also be questioned because they typically refer to one-off purchases rather than the recurrent and long-term relations common in financial services (McKechnie, 1992).
In contrast, the proposed consumer behaviour matrix (see Figure 1) is based upon theoretical insights from the work of Dwyer, Schurr and Oh (1987) and Thibaut and Kelly (1959). Methodologically the approach seeks to utilise the notion of the “ideal type” promulgated by Max Weber (1949). Weber sought to characterise phenomena into broad groups or ideal types which represented the self-conscious and ‘one-sided accentuation’ of the most significant features of social phenomena. Once classified into groups, the constituent elements of the phenomena being observed are available for analysis. The rationale for this approach is that complex social interactions rarely, if ever, operate to a set pattern like elements in the physical sciences. Rather than postulate general theories that describe behaviour in all contexts, ideal types describe forms of behaviour in certain contexts, which facilitate the construction of hypotheses. The ideal type expresses what Weber described as an “objectively possible” course of actions within a relevant frame of reference (Parsons, 1951). Behaviour is, therefore, dependent upon both the nature and the context in which it occurs. An ideal type will therefore possess a number of distinguishing characteristics, constructed by the researcher to describe the complexities of behaviour within a particular frame of reference.
Weber argued that the key to developing ideal types was to identify characteristics that shaped both the forms of the ideal type and the frames of reference. It follows that in developing ideal types which characterise consumer buying, it is necessary to identify the underlying constructs which determine consumer behaviour within a particular environment or frame of reference. Ideal types, therefore, reflect certain underlying constructs or imperatives, and individual buying behaviour is determined accordingly. This is consistent with the work of Fishbein (1967) which links attitudes and outcomes, arguing that individuals’ attitudes toward certain outcomes motivate behaviour.
The advantage of the consumer behaviour matrix, as an approach to understanding consumer buying and contracting behaviour, is that it draws upon a diverse source of literature, including economics (Simon, 1957), consumer behaviour (Bloch, 1982; Bloch and Richins, 1983) and psychology (Thibaut and Kelly, 1959), and combines these within a single framework. The rich diversity of literature in these areas is brought together to create ideal types of behaviour which can be applied to actual buying and contracting in financial services. In this way, the consumer behaviour matrix places observed behaviour within a context. By identifying the underlying factors influencing buying and contracting behaviour and linking these directly with consumer needs, the rationale for consumer interactions and relationships can be more readily analysed.
Furthermore, the use of this framework permits the adoption of a deductive approach in the choice of a clustering variate. The technique thereby facilitates segmentation on a range of self-selected dimensions (Christopher, 1969). As Ketchen and Shook highlight “Cluster analysis has played a key role in…[strategic management research] because it allows for the inclusion of multiple variables as sources of configuration definition, thus enabling the creation of potentially rich descriptions” (1996: 451). Speed and Smith (1992) in their review of approaches to segmentation in the financial services sector classify existing research as either a priori or cluster-based. Harrison (1998) has adopted such a cluster-based approach in analysing the financial services consumption patterns of UK bank customers. Using insights from qualitative research (Harrison, 1994), an attempt was made to replicate and validate the analysis through cluster analysis. The research suggests that consumers differ in terms of their perceived knowledge and understanding of financial services. They also differ in their confidence and ability to handle financial matters and in their general interest in the subject matter. On these bases, attention was drawn to four distinct consumer clusters, ranging from the ‘financially confused’ who were characterised by low levels of both perceived knowledge and financial maturity to ‘capital accumulators’ who were characterised by high levels of perceived knowledge and a high degree of financial maturity. McDougall and Levesque (1994) in a similar fashion sought to segment consumers on the basis of service quality. Their research was able to draw attention to two segments, those consumers who were primarily convenience-orientated and those who placed priority upon performance characteristics.
Services can be classified in a number of differing ways (Lovelock, 1983; Normann and Haikola, 1986; Storbacka, 1994). They can be grouped in terms of their marketing characteristics, such as the nature of the provider-customer relationship (Lovelock, 1983) or on the basis of the content of the interaction between provider and customer (Storbacka, 1994). Storbacka proposed a typology of interactions distinguishable in terms of the duration and frequency of the interaction, and the level of customer control, contact and participation. Using this typology he drew attention to five main governance systems: transactions, deposit and lending, counselling, specialist services and investment services. This paper utilises the Storbacka typology, but excludes counselling on the basis that it is implicit in the other forms of governance. Four categories of financial services are, therefore, highlighted as representative of the range of interactional contexts: transactions-based services such as a current account; insurance-based services such as house contents, buildings or motor insurance; longer-term lending services such as personal loans or mortgages, and finally, investment services which were believed to be relatively complicated with commensurately high levels of risk, ie PEPs, TESSAs, Stocks and Shares and Pensions. In this manner the research sought to segment consumer behaviour on the two constructs of involvement and uncertainty for a range of differing financial products.
2.1 Forms of Consumer Behaviour
The model outlines four ideal types of consumer behaviour which are labelled: repeat-passive, rational-active, relational-dependent and no purchase. These quadrants have previously been examined in the light of qualitative research (Beckett, Hewer and
Howcroft, 2000), however a brief description of the chief characteristics of each quadrant follows:
Repeat - Passive
Consumers display low levels of involvement with the financial product as they are fully aware of the product’s salient features. Given the low levels of involvement and the limited perception of uncertainty, these consumers can be described as “passive” in the sense that they will make repeated interactions without actively seeking alternatives. This repeated pattern of purchase behaviour, which is described as “behavioural loyalty” in the literature, has been extensively researched (Brown, 1952; Johnson, 1973, 1982).
Rational - Active
These consumers display high levels of involvement in terms of the process dimensions of control, participation and contact. Confidence in terms of understanding the product and certainty of ex poste outcome is similarly high. To purchase in an ‘instrumentally rational’ manner, individual consumers are assumed to possess sufficient ability and information to enable them to make clear comparisons between competing products and thus make an informed choice. If the information is not available or the consumer lacks the ability to make choices, they have to move away from “instrumental” rationality as discrete contracting is no longer an effective means of structuring the transactions.
No Purchase
This quadrant describes consumers who, because they have no involvement with the financial product and do not possess the ability or the confidence to make transaction decisions, make no purchase.
Relational - Dependent
In this quadrant consumers are highly involved, but are not very confident due to the complexity of the product and uncertainty of ex poste outcomes. In order to make choices, the consumer will seek advice and help from banks or third parties and can, therefore, be described as “dependent consumers” who form relationships to reduce uncertainty and structure their pattern of purchases.
Relational contracting does not fit easily into the concept of either an active or a passive interaction, but it is clearly an important aspect of the banker-customer relationship. It emerged from the work of MacNeil (1978) and Williamson (1975, 1985), who recognised that in particular contexts rational-active and repeat-passive contracting were not effective in structuring exchange. It is typically used in highly uncertain environments where consumers lack the information to make rational decisions, but, nevertheless, perceive differences in quality between competing products or services. Under these circumstances, if consumers want to make informed choices, they have to draw on the assistance of more informed third parties. This relationship with third parties effectively replaces the information search and processing activities found in repeat-passive and rational-active contracting. Trust plays a critical role in this relationship and, to a large extent, the role of professional associations is to protect consumers from the opportunistic behaviour of third parties.
3. Research Analysis and Sample
The previous section provided a detailed examination of the relevant literature relating to consumer behaviour in the context of purchasing/contracting financial products and a model outlining four ideal types of consumer behaviour (see Figure 1) was proposed. The main objective of this research was to examine consumer behaviour and ascertain if the conceptual model could be supported empirically.
Following a series of focus group discussions, a questionnaire was designed and piloted using university personnel. The definitive questionnaire was then sent to 4000 UK consumers. This resulted in 244 useable responses, ie out of the 351 responses, 244 respondents had answered all of the 24 questions. The relatively low response rate of less than 10% was not that surprising given the confidential and personal nature of the information requested (see Table 1). A major concern, however, was the representativeness of the sample in the light of this response rate. In this respect, when compared with recent UK statistics (Social Trends, 1999; Advertising Association, 1998), it was found that the sample was under-representative of consumers under the age of 25 and over-representative of consumers in the higher income groups (ie > £30,000). This profile may be explained by the fact that investment products have a tendency to be purchased by higher income groups and by consumers aged over 25 years old.