Schedule - Parallel Session 6 - Intertemporal Choice 3

WMG IDL 1st Floor Syndicate Room - 15:40 - 17:10

Revealing Sophistication and Naivete from Procrastination

David Freeman

Abstract

Consider a person with time inconsistent preferences who must complete a task, and is given a set of options for completing the task at different times. The person cannot commit her future behaviour except by completing the task. This paper shows that comparing a person’s completion time across different sets of completion opportunities can reveal her sophistication or naivete about her dynamic inconsistency. I show that adding an extra opportunity to complete the task can lead a naïve (but not a sophisticated) person to complete it even later, and can lead a sophisticated (but not a naïve) person to complete the task even earlier, even if the extra opportunity is not used. This result can be obtained with or without parametric assumptions about utility. In the quasi-hyperbolic discounting model with partial naïveté, such “doing-it-earlier” reversals are shown to exist for any partially sophisticated individual, while “doing-it-later” reversals always exist for any partially naïve individual. Additional results completely characterize models of naïve and sophisticated individuals in this environment. These results provide the framework for revealing the preference and sophistication types studied in O’Donoghue and Rabin (1999) from behaviour in a generalization of their environment.

David Freeman

Assistant Professor, Simon Fraser University

Intertemporal Choice - Discounting and Projection Bias

Kirsten I.M. Rohde; Chen Li

Abstract

Intertemporal choices concern tradeoffs between optimal sizes of outcomes and their optimal timing. A popular model to describe such tradeoffs is the discounted utility model. Discounted utility evaluates an outcome to be received in the future by multiplying its utility by a discount factor. Utility captures the expected attractiveness of an outcome irrespective of its distance to the present. The discount factor captures the attractiveness of the timing of the outcome irrespective of its expected utility. Utility refers to the psychological process where people project themselves to the future and imagine how valuable an outcome will be to them once consuming it. Discounting refers to the value one derives today from foreseeing to experience the value of the outcome once consuming it. Traditional measurements of discounted utility assume that utility is constant over time. Thus, it is commonly assumed that the value one derives from an outcome today equals the value one expects to experience in the future once consuming the outcome. The difference between the value of receiving an outcome today or at a later point in time, it thereby entirely attributed to the discount factor applied to future values. Yet, it is well possible that the value one derives from an outcome today is different from the value one expects to derive from it in the future. If individuals, for instance, project themselves to the future and expect their future consumption level to be higher than their current one, it is well imaginable that they expect the value of an outcome in the future to be different from its value today. In this paper we study intertemporal decisions and disentangle discounting and the expected changes in utility. We measure whether people expect their utilities to change with time. This allows us to measure discounting independently from utility and expected changes in utility. We employ a methodology from measurements on subjective well-being to assess strength of preferences. This allows us to assess the extent to which subjects expect their utilities to remain constant over time. We also assess the extent to which the expectations about future utilities are correct. That is, we assess the extent of projection bias in predicting future utility.

Kirsten I.M. Rohde

Professor, Erasmus University Rotterdam

It Will Be Worth It, in the End

Philip Newall

Abstract

We embark on ventures, investments, or other projects with the common-sense saying: it will be worth it, in the end! Anticipated outcomes are evaluated against inputs. A required rate of return, or more general rate of time preference, r, assesses the worth of anticipated future outcomes. Comparing anticipated future outcomes against inputs with a rate of time preference is formally consistent with the discounted utility model (Samuelson, 1937), but notice that the rate of time preference need never discount anticipated future outcomes to equivalent present values. “Discounted utility” conceals the flexibility of the economic logic. A rational actor need never discount for finite time horizons. A voluminous literature shows that intertemporal choice behavior cannot be unanimously described between-experiments or even within-individuals by a single “discount rate,” as the rate of time preference is usually called (Frederick, Loewenstein, & O’donoghue, 2002). Numerous modeling attempts have sought to provide better descriptions of intertemporal choice behavior by modifying the parameters of the discounted utility model (Doyle, 2013). Perhaps best-known is the quasi-hyperbolic discounting model, used to describe present-biased behavior (Laibson, 1997). Although discounting models can be fitted to describe choices in an intertemporal choice task, this approach is yet to provide a unifying description of intertemporal choice behavior. (Dohmen, Falk, Huffman, & Sunde, 2012) elicited choices from 2,003 participants in three commonly-used intertemporal choice tasks. They found that, “Overall, about 65 percent of individuals are inconsistent with all discounting types.” p.21. Imposing a discounting framework on descriptive theories of intertemporal choice has so far failed. Reframing our intertemporal choice models is suggested as a way of moving forward, of describing most people’s intertemporal choices. People make intertemporal choices on the basis of anticipated future outcomes, and can achieve normatively appropriate results if they do so consistently. Discounted present values need never enter the equation. Of course many people do not achieve normative results, so we should model the psychological processes that undermine accurate predictions of the future.

Philip Newall

PhD Student, University of Stirling

Present Bias and Everyday Self-Control Failures: A Day Reconstruction Study

Leonhard Lades; Liam Delaney

Abstract

Present bias is the economist’s favourite explanation for self-control problems (O’Donoghue and Rabin, 2015). Various studies elicit present bias in monetary delay discounting tasks and show that experimentally elicited present bias (weakly) predicts various behaviours related to self-control (Sprenger, 2015). However, recently monetary delay discounting tasks have come under attack (e.g. Augenblick et al, 2015). Importantly for this paper, also the empirical link between present bias and self-control problems as they occur in everyday life has not yet been established. We directly test whether individual differences in present bias are linked to individual differences in self-control in everyday life. We elicit present bias in a double multiple price list as it is typical procedure using money, avoiding confounds recently highlighted in the literature (Andreoni and Sprenger, 2012). We measure everyday temptations, self-control attempts, and self-control failures using a day reconstruction methodology (DRM) as introduced by Kahneman et al. (2004). The DRM is well-known from happiness research, but this is the first paper that uses the DRM in order to measure everyday decision-making and in particular everyday self-control using the questions suggested by Hofmann et al. (2012). In a sample of 142 participants we find that experimentally elicited present bias is not associated with self-control problems in everyday life. Present bias is also not associated with psychological trait measures of temptation and self-control. These psychological trait measures, however, do predict everyday self-control problems, indicating that the DRM provides valid information about temptations and self-control as they occur in everyday life. The results are in line with a clear distinction between time discounting and visceral influences as determinants of decision making. Present bias as typically measured in delay discounting tasks might be a specific type of dynamic inconsistency resulting purely from the passage of time. Self-control failures in everyday life are more likely to be the result of visceral states (Loewenstein, 1996) that increase the motivational power of short-term desires and are not necessarily correlated with the passage of time. Accordingly, the results can explain why recent studies find only weak empirical associations between present bias elicited in monetary delay discounting tasks and life outcomes related to self-control.

Leonhard Lades

Early Career Fellow, Behavioural Science Centre, University of Stirling