Communication Task Difficulty in Investment Risk Profiling: A Linguistic Perspective

Working out a client’s appetite for risk is an essential role for financial planners in determining appropriate asset allocations for investment decisions. A client’s risk profile is “broadly defined as a person’s emotional and financial capacity to take on risk” (Brayman, et al. 2017, p.71), and it is assumed to comprise both objective and subjective attributes having relatively stable parameters.

Objective factors are those elements that can be measured quantitatively. Examples include an individual’s capacity to incur financial losses and the time horizon associated with the accomplishment of a financial objective. Subjective factors include concepts such as risk perception and risk preference, both of which are based on a client’s idiosyncratic evaluations of the riskiness of a situation or choice. Historically, financial advisors have emphasised objective attributes of a client’s risk profile when making investment recommendations. (Brayman, et al. 2017, pp.72-3)

In surveying the ways a client’s risk profile can be measured and assessed, Brayman, et al. 2017 (pp. 75-76) found that there are three ‘marketplace’ approaches in common use among financial advising firms and their planners: (1) a comprehensive risk profile tool providing objective and subjective questions, which is psychometrically designed and adapted to match each financial adviser’s need; (2) subjective risk tolerance questionnaires which are psychometrically designed to measure a client’s willingness to take financial risk; and (3) an asset allocation calculator based on subjective and objective measures, and designed ad hoc by financial consultants or in-house within a financial advising firm. This third model is an economic approach based on income and investment gamble preferences and adviser experience. Brayman, et al. (2017) note that while the first two approaches provide a platform for further relevant discussions with the client before finalising their risk profile, the third approach provides a single-step solution with minimal professional judgement input by the adviser. One-issue telephone advice, as we shall see in the data analysis below, fits the criteria of this third approach.

‘Risk tolerance’ is the term most widely used by financial advisers when talking about client risk taking (Nobre and Grable, 2015, p.19) and refers to “an individual’s willingness to engage in a financial behaviour in which the outcome is both unknown and potentially negative” (Brayman, et al. 2017, p. 75. As Nobre and Grable (2015, p. 19) note, “accurately measuring financial risk tolerance is the most important task a financial adviser faces when thinking about risk profiling”. However, it is often unclear which aspect of risk a planner is actually dealing with. Nobre and Grable (2015) systematically address the issue of confusion surrounding risk terminology (i.e. the interchangeable and/or inconsistent use of terms such as risk tolerance, risk preference, risk need, and risk perception) and provide clear definitions of each type. Moreover, they integrate risk profile parameters (i.e. risk composure, capacity and preference) with a model of financial risk taking behaviour (i.e. risk profile, perception and need which directly influence risk tolerance which in turn leads to risk behaviour) resulting in improved clarity for financial planners trying to make sense of the various terms and how they are related to one another.

In parallel with confusion clouding a clear understanding of financial risk terminology is some confusion regarding the regulatory demands on financial planners in relation to how they handle a client’s risk profiling and risk tolerance assessment. According to Brayman, et al. (2017, p.72), nearly all advisory regulators worldwide have refrained from prescribing how financial advisers should measure and evaluate a client’s risk profile and/or risk tolerance. The reason for this is that

…nearly all worldwide advisory regulators already place a professional responsibility on financial advisory firms and advisors to gauge a client’s risk profile….even those [regulators] with the most prescriptive guidelines…use a principles-based approach that requires the advisor, dealer, representative, or financial advisory firm to determine how a client’s risk profile should be assessed. (Brayman, et al. 2017, p.72)

Thus the fiduciary duty placed on financial planners to put their clients’ interests first would appear to include conducting an appropriate risk tolerance assessment. Such assessments are commonly performed in whole or in part by means of a questionnaire instrument, the topic of the next section.

The first published risk tolerance questionnaire dates back to 1984 (Droms and Straus, 2003). MacCrimmon and Wehrung (1986, p.65) note the relative advantages of a questionnaire instrument compared to other assessment instruments: (1) they exclude any subtle influences in the assessment by the assessor (although this is perhaps only true for written-mode instruments and not spoken-mode ones, as we shall see in the data analysis below); and (2) they provide flexibility to concisely include a range of complementary question items. In subsequent breakthrough research in risk evaluation, Grable and Lytton (1999) presented a model risk assessment instrument in the form of a questionnaire comprising 13 items in a multiple choice format. The authors detail the development of the questionnaire, including verification of its validity and reliability, and propose that it form the basis for creating an index that could be widely accepted in the field of financial planning.

Subsequent reports of further validation studies of their instrument support its continued usability as a sound measure for determining a client’s risk profile (Grable and Lytton, 2003; Kuzniak, et al. 2015) and its wide uptake amongst consumers, financial advisers and researchers (Kuzniak, et al. 2015). This questionnaire has a positive correlation with the widely used yet unsophisticated Survey of Consumer Finance (SCF) risk assessment item, and adheres to requirements suggested by MacCrimmon and Wehrung (1986) that the instrument (1) cover a variety of risky financial situations in a multidimensional manner; (2) be consistent and non-redundant; (3) be interesting; and (4) not be too time consuming to complete. Yet, despite its validity and reliability, the Grable-Lytton questionnaire has not been universally adopted; many financial advising firms and planners seem to demand more flexibility and seek bespoke instruments that they feel are better suited to their individual needs.

Linguistics, the study of language, can inform the task of risk profiling in ways unlikely to have been considered by financial planning researchers or practitioners, but which stand to improve current professional practices. Skehan (1996, pp. 52-53), drawing on the work of Candlin (1987) and Nunan (1989), proposes that three key factors contribute most to communication task difficulty: (1) code complexity (i.e. concerned with grammar and vocabulary difficulty and range); (2) cognitive complexity, affected by cognitive processing and cognitive familiarity; and (3) communicative stress, including time pressure, stakes (i.e. high versus low), and the control a person exercises over completing a task. The research reported below shows how a linguistics-informed perspective can benefit telephone-based financial planning consultations.

The five recordings were professionally transcribed by a third party and then the transcripts were checked for accuracy and completeness by the researcher. Some adjustments were made wherever errors were noted (e.g. un-deciphered words; misheard words; misallocated turns). Each consultation covered a range of topics but the risk profile assessment activity was the object of study for this research. The researcher used various linguistic theories related to communication task difficulties (detailed below) to analyse the discourse of the risk-profiling segments within each of the five consultations.

As noted above, financial advisers appear to use one of three main approaches to determine a client’s risk profile. All three involve the use of an assessment instrument (often a risk profile questionnaire), which might typically be administered as follows:

Establish that a risk assessment is needed

The data investigated in the present study fall within the ‘asset allocation calculator’ approach, in which the risk profile questionnaire forms the basis of the client’s risk tolerance assessment. Table 2 sets out the risk profile questionnaire used by the financial planners in this study².

In the following sections, the linguistic features of code complexity, cognitive complexity and communicative stress are introduced in turn, along with an analysis of the risk profiling data in relation to these three features.

If planners or clients were asked which language issues make dealing with a financial planner difficult, both would likely mention financial jargon. This specialised vocabulary is salient and therefore easily detectable as an issue of potential importance in planner/client interactions. However, it is not simply a matter of knowing what a financial term such as “negative year” means³. There is also the matter of planners using terms that are interchangeable (synonyms) to them, but not necessarily understood as synonyms by their clients. For example, “investment choice”, “investment options” and “investment structure” might be understood by a client to mean three different things. Furthermore, there is the matter of words that have everyday meanings but also more specialised meanings in financial planning discourse. For example, “returns”, “risk” and “debt” (as in “debt asset”). Such terms may be taken for granted by planners, but pose considerable problems for clients who do not anticipate them and are unfamiliar with their specialised meanings.

Beyond the obvious issue of understanding vocabulary and common expressions used in financial planning consultations, there are less visible but equally important code complexity issues impacting on communication at the ‘higher’ meaning levels of grammar and discourse. Differences between spoken and written English are manifest and to some extent conflicting in telephonebased consultations, and hybridised modes of talk also complicate planner/client interactions. These will now be explained and exemplified from this study’s data.

It is important for planners to be aware that although English grammar is the same whether it is spoken or written, the patterns of spoken and written English are in fact distinctive (Paltridge, 2006; Gotti, 2011), and this can be consequential. Grammar can best be thought of as a ‘meaning-making’ system comprised of grammatical ‘function’ words (e.g. “the”, “his”, “by”, “with” etc.) and ‘content’ (lexical) words (e.g. “invest”, “shares”, “markets”, “value” etc.). It is the content words that carry explicit meaning in interactions, while the function words enable meanings to be made efficiently. One key difference between spoken and written English is shown by a ratio known as ‘lexical density’, which is simply the number of lexical words in a sentence divided by the total number of words (Ure, 1971, p. 445). Spoken English is typically less lexically dense than written language because the latter is more carefully crafted since there is usually more time to prepare what is communicated. When written language is spoken, as when a planner reads aloud from a risk profile questionnaire, there are comprehension consequences since its grammar makes it more difficult to process aurally (i.e. in listening mode) than by reading. In other words, language that a client might find comprehensible to read, could be much less comprehensible to them if they hear it rather than read it. So this issue becomes quite relevant when planners follow any sort of script by reading aloud. There are two key aspects to be considered: (1) has the task designer written the questions to be read or to be spoken? (see Example 1), and (2) when a planner speaks the script to the client, do they adhere closely (verbatim) to the script, or do they make modifications such that the script becomes more ‘spoken’ in nature? (see Example 2). These issues raise implications for how clients might be better prepared in advance for consultations with their planners, a point taken up in the Recommendations section of this paper.

The question here (Question 1 from the risk profile questionnaire set out in Table 2) addresses the listener as “you”, which is natural when asking someone a question. So, despite the question being read aloud from a written source, it is perfectly clear and accepted by the listener as ‘normal’ spoken language. The options, however, are not all consistent with the stem question, nor consistent with one another. Options A and B have omitted the subject of the question “you”, so the listener infers it when hearing the planner ask the question. Thus, for option A, they infer “(you’ve) never invested”; and for option B (“you’ve) held investments once or twice”. When the planner subsequently reads aloud options C and D, with their inclusion of a different subject, “I”, this immediately generates a need for the listener to adjust their understanding of “I” as referring also to them, and not to the planner. The point being illustrated here is that the written form of Example 1 causes no problem, but as soon as it is read aloud to someone who can only hear it and not read it, the inconsistencies of the grammar, innocuous though they may seem, increase the level of difficulty in processing and responding to the question.

Appendix 1 illustrates in detail the variation among five financial planners asking this same standardised question from the risk profile questionnaire. It shows that variation from script is common among the planners. Example 2 shows how one planner has modified a scripted question (Question 4 in Table 2) by inserting additional language (highlighted in bold). The additional language serves two purposes: (i) it makes otherwise formal language sound more informal, which helps build rapport with the listener; and (ii) it allows the planner to clarify and educate, as in the explanation given for a “negative year”. Taken together, these modifications are essentially converting written language to spoken language that is easier for the client to process.

Spoken and written English also differ in other significant ways. While the common unit of analysis in written English is a sentence, in spoken English is it a clause (i.e. a unit that is typically smaller than a sentence, and often a constituent of a sentence). So, when we speak, we tend to use grammar that connects lots of clauses. When we write, by contrast, we use a more sophisticated or complex grammar that allows us to make meanings more efficiently, i.e. using fewer words. One common means of achieving this is the process known as ‘nominalisation’ (Halliday and Matthiessen, 2013), in which verbs in spoken English are changed into nouns in written English (e.g. “the rate at which inflation grows” becomes “the inflation growth rate”). This nominalisation process allows the packing of more information into each sentence (and, thereby, increases lexical density). Example 3 illustrates how nominalised statements have been used in the risk profile questionnaire in this study.

The word count between the two versions is striking (11 versus 21), as is their lexical density (0.64 versus 0.52) respectively⁴. By including the “you” in the non-nominalised version, we can clearly see who is responsible for the action (i.e. the client as agent), which is hidden in the more condensed nominalised version.

If we consider spoken English above/beyond the level of vocabulary or grammar we are talking about the ‘discourse’ level. At this level in spoken interactions there are different “modes of talk” occurring at different points in the interaction. Roberts and Sarangi (1999) identify three modes: institutional, professional and personal. For example, in financial planning consultations, some institutional talk always occurs near the beginning of consultations where the planner informs the client that the phone call is being recorded “to ensure the highest level of service” and seeks their consent to continue. This is standard institutional practice in the financial services industry. Professional talk is a different mode of talk in which planners display their knowledge or skills. For example, when a planner explains different investment categories such as “secure assets” or “growth assets”, they are speaking in professional mode. Personal talk features in professional/ client interactions where the planner speaks as a friend. Example 4 illustrates this point as the professional voice of the planner is augmented by a personal comment about being in the same situation as the client (i.e. “people our age”), with reference to the compulsory 9% employer superannuation contributions.

The common use of these different modes of talk leads to a hybridised discourse that is more textured and richer in meanings. The client expects the planner to engage in professional talk, after all, that is why they have sought the planner for advice. Moreover, the client also implicitly understands that there are institutional impositions on the talk, perhaps through regulatory requirements. And, the client anticipates that they are talking to a person who is interested in engaging with them in a friendly, less formal way. Depending on how skilfully these modes are executed by the planner, processing the blend of different modes of talk can present challenges that are more communicatively demanding than processing any single mode on its own. This is especially the case when a client is from a different language or cultural background since hybridised discourse patterns of this type may be quite unfamiliar to them.

Cognitive Load Theory (CLT) (see Sweller, 1988 and 1989) deals with the relationship between short-term “working memory” and long-term memory. CLT states that by using schema acquisition (i.e. learning how to structure relevant information to deal with a particular task) combined with automation (i.e. practising until the process becomes virtually effortless), people are able to learn and store information efficiently in long-term memory for future recall, and thus substantially reduce working memory load (important because working memory is much smaller than long-term memory). It is worth reflecting on the considerable gap between what financial planners understand about financial information and what average lay people know about it, as well as the substantial gap in their respective abilities to process financial information with ease and at speed. Although experts and clients converse in the same language (e.g. English) what is understood by each party in the same interaction can be substantially different.

Let us now consider some examples from the investment risk-profile task to investigate whether cognitive complexity for clients is low, medium or high in this task. The eight questions used by the financial planning firm to determine a client’s risk profile (see Table 2) are all in a multiple choice question (MCQ) format (i.e. a statement/question stem followed by, in this case, four options to choose from). This question type is well-known and widely used in Australian schools and other organisations to assess a particular subject matter. Thus, the MCQ format is easily understood and poses no real cognitive processing challenge to clients. The content of MCQs, by contrast, is potentially difficult or even very difficult depending on the familiarity the client has with relevant information and their ability to process and evaluate it efficiently. For example, a client may understand the concept of inflation (see Question 3), but may never have experienced it other than at a very low rate. For them to contemplate a high rate of inflation is an abstraction, moving away from their lived experience; and for them to contemplate how a high level of inflation might impact their retirement savings (which to many young people is a goal so far into the future as to be completely unreal) is itself a further abstraction. To process this information and evaluate their personal feelings about the scenario and reach a conclusion on which much of their future wealth could depend, all in the space of a few seconds, requires a very high level of cognitive processing. Accordingly, it is quite difficult.

Questions 2 and 4 (see Table 2) provide a stark contrast in their cognitive demands of short-term memory. The former is 24 words in total, including an average option length of 4.0 words. The latter, by contrast, is 101 words in total, including an average option length of 19.5 words. Coping with the latter is much more cognitively challenging. However, the repetitive structuring of Question 4’s options illustrate how a schema works: if the client recognises the pattern of a rising rate of return being linked with a rising chance of a negative year, the focus can be on the numbers and not on all the accompanying verbiage.

Question 5 (see Table 2), which asks “How would you feel and react…” presents a different cognitive challenge in blending two questions and options with compound answers (e.g. “Distressed. You’d seek to sell the investment and get your remaining money back.”). The first question is How would you feel?” and second is “How would you react?”. Rather than give the client the simple choice of answering how they would feel, the question also associates one reaction with each type of feeling. Thus, the coupling of option A, just mentioned, between ‘distressed’ and ‘seek to sell the investment’. The client cannot be “distressed” and “look at what other options you may now have” because that reaction is part of option B. Recognising the closed sets for each option, therefore, requires more cognitive processing, perhaps even requiring the reconciling of somewhat mismatching couplets within each option set.

Example 5 illustrates the cognitive demands of a complicated question, and how a client has to process a lot of information, some of it illogically presented, before immediately deciding and reporting their strength of feeling towards achieving their financial goals.

This question is different from the other seven MCQs in the arrangement of its four options, which are out of sequence in terms of ascending or descending order. Also, option A alone has a pre-option statement while the other options do not. (It seems to serve to emphasise the strong commitment being made by any client who selects this option). A very important issue arises from the sequencing of options A and B. They each refer to extremes of a continuum of risk tolerance: option A is the most tolerant of risk; option B the least. This juxtaposing here foreshadows that options C and D are either on a different continuum or are out of sequence for this same continuum.

The client’s initial expectation prior to hearing the four options for Question 6 is that they would logically be presented sequentially in an ascending or descending order. However, when one processes the meanings of each option (given enough time and memory of what each option was) it becomes clear that they are out of sequence. If option A were placed last, then the whole series of options would be presented in ascending order of risk. Given that lay clients may need to be positioned to see solutions that are most likely to be in their best interests, one can perceive Question 6 as being a two-part question sequence. The first part comprises the two extremes of the highest and lowest risk-taking attitudes (i.e. options A and B), both of which are unlikely to be appropriate for the superannuation fund’s clients; while the second part deals with more ‘sensible’ options, risking either a small amount (i.e. option C) or a large amount (i.e. option D). In other words, clients are positioned by the framing of Question 6 to answer with either option C or D. Should they select options A or B, the planner might well probe this choice further to ensure the client has properly understood its implications. The main point to be taken here is that from the client’s perspective this MCQ is markedly different from all others and, as a result, more cognitively challenging.

Cognitive familiarity concerns the extent to which a task draws on ready-made or pre-packaged solutions already held in a person’s brain by way of schemata stored in long-term memory. Risk profiling is easy for planners because of their familiarity with the concepts and their profiling questionnaire, which are practically second nature to them through schema acquisition and their practical experience of performing numerous risk profile assessments. However, for the client, who may never have thought much about how they perceive financial risk, the risk-profiling questionnaire might feel quite unfamiliar for the statements and options it uses to determine their appetite for risk. (They might actually prefer to discuss financial risk by other means, for example, by asking their own questions about various aspects of risk). This lack of familiarity means the task is cognitively more challenging and, therefore, more difficult for the client.

Time pressure, or the sense of urgency, is a key feature of telephone-based financial consultations. Clients are mindful that planners can only devote a limited amount of time to their needs; planners are under pressure to service as many clients as possible each day. In spoken interactions as opposed to written correspondence, everything takes place in ‘real time’. Clients are under pressure to explain the reason for their contacting the planner, and then to process the information given by the planner, and ultimately to make decisions about their financial future, all within a period of about 30 minutes in one-issue consultations. Clearly, this pressure of time increases the stress of the interaction. Remarkably, the five clients involved in the investment risk profile assessment in this study all responded within a few seconds to each MCQ. How they did this (i.e. were they absolutely clear about each question and its options? Or were they sometimes guessing?) is unknown due to limitations in accessing them for subsequent interviews.

The modality of the interaction for financial planning advice is also an important feature of communicative stress. Listening to a planner is more stressful than reading a financial advice brochure. The client is dealing with unfamiliar content and being asked questions about matters they may never have given much thought to previously. If they do not understand something they may feel reluctant to query it to avoid losing face and sounding ignorant or foolish. They are positioned more as ‘recipients’ of knowledge rather than as collaborators or generators of knowledge, so a passive disposition works relatively well in questionnaire-based risk profiling interactions of this sort. Speaking to a planner is similarly more stressful than writing a message in an online dialogue box. The real-time interaction poses constant threats to a client’s public face. In sum, the client’s ability to process unfamiliar specialised financial information is cognitively challenging, and this is made more difficult if the mode is listening rather than reading.

Scale concerns the number of participants in an interaction and the number of different relationships involved. In telephone-based financial consultations, there are only two people directly involved, the planner and the client, so it is small-scale and the communicative stress is low in this respect. However, as most telephone-based planning takes place between strangers having no prior relationship, this can raise the communicative stress for both client and planner.

The stakes involved in a communication task are also a potential source of communicative stress. The issue here concerns how important it is to do the task correctly. In the case of financial planning and in particular the determination of a client’s risk profile, the stakes are very high. If the risk profile is incorrectly calculated, then there could be long term negative financial consequences for the client and possibly legal consequences for the planner, as noted by Hunt (2016). Thus, communicative stress is increased for the client who is aware that getting appropriate financial advice is a high-stakes event, and this is likely to be the case for most clients.

The final aspect of communicative stress related to context is concerned with the issue of control, and the extent to which the participants in an interaction can exert an influence on the task or how it is done. As we have seen above, a financial planning firm’s investment risk-profile questionnaire is a highly prescribed instrument and, therefore, not one which the planner or client can alter in terms of the task itself. In other words the planner asks set questions, and the client provides their answers by nominating one of four options. How the task is done does vary slightly from planner to planner, as exemplified in Appendix 1, where some planners embellish the questions by introducing additional talk which serves to make the task feel less formal, as shown above in Example 2.

Clients, for their part, have even less control over how the task is done since they are recipients of information rather than producers of it. Nevertheless, there is evidence of some degree of exercising control when clients answer the questions by reference to the wordings of an option rather than the option label (A, B, C or D). Example 6 illustrates this point by reference to one particular client in the dataset.

These responses are substantially different from the options stated for the questions (see Table 2), but provide evidence of how the cognitive load felt by this client may have inhibited their shortterm memory from storing the precise wordings of the various options.