Adolescent Time Preferences Predict Lifetime Outcomes

Abstract

This study investigates the relationship between time preferences and lifetime social and economic outcomes. We use a Swedish longitudinal data set that links information from a large survey on children's time preferences at age 13 to administrative registers spanning over five decades. Our results indicate a substantial adverse relationship between high discount rates and school performance, health, labour supply and lifetime income. Males and high-ability children gain significantly more from being future oriented. These discrepancies are largest regarding outcomes later in life. We also show that the relationship between time preferences and long-run outcomes operates through early human capital investments.Every day people make decisions that involve balancing costs and benefits occurring at different points in time. Such choices include whether or not to drop out of school, search for a new job or start saving. Intertemporal decision-making has been a cornerstone in many economic models since Samuelson (1937), and a salient feature in human capital theory, where the notion is that people with high discount rates invest less in their future than people who are more future oriented (Mincer, 1958;Becker, 1964). As the full returns to many human capital investments are not revealed until some time later, it is remarkable that there are few empirical studies which link time preferences to long-term outcomes. 1 This lacuna is especially evident regarding investments made early in life. Needless to say, childhood represents a critical period when many important investments are made with potentially life-long consequences. With a small number of exceptions (Mischel et al., 1989;Cadena and Keys, 2011;Moffitt et al., 2011), the existing evidence on the connection between time preferences and real-world outcomes is cross-sectional in nature and focuses on the adult population.This study investigates the relationship between time preferences during childhood and long-run social and economic outcomes. We use a Swedish longitudinal data set that links survey-based information on 11,907 children's time preferences at age 13 to administrative registers spanning over five decades. Time preferences are measured through a questionnaire in which children are asked to rate the extent to which they prefer SEK 900 (US$ 138) today over SEK 9,000 (US$ 1,380) in five years. 2 We

Challenges in research on preferences and personality traits: Measurement, stability, and inference

“…Suppose one would like to analyze the effect of risk preferences on income. The problem is that (even in a laboratory setting) it is not possible to change risk preferences and exclude the possibility that something else also changed as a result of the intervention (see, e.g., Golsteyn et al, 2014). For instance, when changing risk preferences, one may also change time preferences as the future is inherently more risky than the present.…”

On the interpretation of non-cognitive skills – What is being measured and why it matters

“…Exceptions that connect (incentivized) preference measures and real-life outcomes are e.g. Burks et al (2015Burks et al ( , 2012, Golsteyn et al (2014), Sutter et al (2013), Rustichini et al (2012) and Chabris et al (2008). Burks et al (2015) is most closely related to this paper and considers the relation between education outcomes, personality measures and economic preferences.…”

Patience and long-run growth

“…Firstly, we contribute to the empirical literature trying to identify the impact of time preferences on behavior. While patience has already been shown to be an important predictor of individual behavior, such as health outcomes, school performance (Golsteyn et al, 2014), the likelihood of having credit card debt (Meier and Sprenger, 2010), alcohol consumption, body mass index and individual savings (Sutter et al, 2013), the literature seeking to identify the impact of time preferences on macroeconomic outcomes is still scant. The few existing studies on this topic do not go much beyond testing for mere correlations , Wang et al, 2011, Preis et al, 2012.…”