Neoclassical consumer choice models presume individual rationality. Important decisions are made by weighing costs and benefits with respect to stable preferences. Psychology, in contrast, emphasizes the importance of context and cognitive limitations. Preferences are malleable, and bounded rationality makes problem-solving conflicted and error-prone. A growing body of evidence from laboratory psychology experiments supports this view of consumer choice. It suggests that choices can be manipulated by frames, cues, and other “features” of a choice set that change the presentation of the choice, but not its content or inherent value.
Economists are often skeptical of the external validity of findings from laboratory experiments. Critiques tend to emphasize that lab studies typically use non-representative and inexperienced subjects, small stakes, relatively static environments, and inherently artificial settings that do not extrapolate readily to the “real world (for a recent elaboration of these criticisms, see Levitt and List, 2005). Consequently, several recent studies have looked for violations of neoclassical consumer choice models in observational data from the field. But few field studies so far have taken the strength of the laboratory methodology a carefully controlled research design featuring random assignment and precise manipulation of specific treatments to test for systematic deviations from neoclassical consumer choice.
Even the marketing literature, which presumably stands to gain considerably from measuring the relative influence of psychological and economic treatments on consumer choices, has a dearth of field experiments (Simester 2004), a few notable exceptions being Dreze, Hoch and Purk (1994), Ganzach and Karsahi (1995), Dhar and Hoch (1996), and Wansink, Kent and Hoch (1998). Conjoint analysis, which is by far the most widely used tool by marketing researchers to assess how different bundles of attributes for a product (such as price, packaging or marketing) are preferred by customers, is typically conducted through hypothetical surveys in academic or corporate laboratories (Green, Krieger, and Wind, 2001).
This paper tests for the existence of psychological deviations from a neoclassical consumer choice model using a large scale field experiment. A consumer lender in South Africa (the “Lender”) mailed over 50,000 loan offers to former clients containing randomly assigned psychological features (marketing treatments). The borrowing decision involves high stakes for these consumers, as interest rates are very high (from 50 to 200% APR) and the median oan size is about one-third of the borrower’s monthly gross income.
A unique feature of the experimental design is that we can price any effects of psychological features on loan demand. The Lender simultaneously randomized interest rates that were orthogonal to the marketing treatments. As such, we can scale sensitivity to marketing treatments by sensitivity to interest rates and thereby price psychology. Specifically, suppose that some psychological feature increases take up by and a one point decrease in interest rate raises take up by y. Then the ratio measures the economic importance of this psychological feature: how large a change in interest rate is needed to produce the same size effect.
We designed the marketing treatments to mimic cues and frames that have been shown to influence choice in the lab. For example, we varied whether the Lender’s rate was compared to a competitor’s (thereby establishing a reference level), and whether this comparison was presented as a loss or a gain. We also experimented with suggested loan uses and with the addition of photographs to the loan offer letter, since psychology has found that cues can be used to arouse emotions that are conducive to consumption. None of the marketing treatments changed the economic terms of the loan offer; they only varied the fashion in which the loan offer was presented. The consumers in our study are experienced borrowers (the median client has three prior loans from the Lender) so it is unlikely that they sought to infer anything about the economic content of the offer from its presentation.
We tried ten types of “psychological” treatments that have been shown to impact demand in prior mostly lab settings. There is evidence that four of them had significant effects in the full sample. Two additional treatments significantly change demand in meaningful sub samples. The magnitude of the effects, relative to the demand elasticity with respect to price, is large; e.g., displaying a single example loan rather than multiple ones increased demand by 9%, the equivalent of a 2.3 percentage point decrease in the monthly interest rate.
We also report on several additional findings that speak to how our main results may play out in general equilibrium. First, demand inducing psychological features appear relatively more effective when the interest rate is relatively high. In other words, it appears that psychological factors matter more for less attractive offers. Second, there is no discernible difference in the effect of the marketing treatments across income or education groups. Third, credit bureau data suggests that psychological features might induce net new borrowing as there is no evidence that other sources are crowded-out. Fourth, marginal borrowers brought in by the marketing treatments do not appear to be worse default risks.
In all, we find some violations of the neoclassical model that appear to be economically important. More constructively, our findings are consistent with an important role for psychology in market contexts, and lend support to behavioral models of consumer choice that incorporate frames (e.g. Kahneman and Tversky, 1979), cues (e.g., Laibson, 2001), and bounded rationality (e.g., Simon, 1955). Moreover, the fact that many psychological features did not have statistically significant effects demonstrates the importance of context in understanding and predicting when certain laboratory findings will hold in the real world. Incorporating psychological features into field studies will be a challenging but necessary step for forming more general theories on when, why and how such manipulations influence important real decisions.
Download
PDF Ebook What’s Psychology Worth? A Field Experiment in the Consumer Credit Market
