Self-regulation composite (preschool & ages 17-37) predicts capital formation at 46, but preschool delay of gratification alone does not predict such capital formation

Predicting mid-life capital formation with pre-school delay of gratification and life-course measures of self-regulation. Daniel J.Benjamin et al. Journal of Economic Behavior & Organization, Volume 179, November 2020, Pages 743-756. https://doi.org/10.1016/j.jebo.2019.08.016

Rolf Degen's take: The renowned marshmallow test goes belly-up, failing to forecast individuals' future

Highlights

• Self-regulation composite (preschool & ages 17-37) predicts capital formation at 46.

• Preschool delay of gratification alone does not predict capital formation at 46.

• The composite is more predictive partly because it consists of many items.

• No evidence of more predictive power for self-regulation reported later in life.

Abstract: How well do pre-school delay of gratification and life-course measures of self-regulation predict mid-life capital formation? We surveyed 113 participants of the 1967–1973 Bing pre-school studies on delay of gratification when they were in their late 40’s. They reported 11 mid-life capital formation outcomes, including net worth, permanent income, absence of high-interest debt, forward-looking behaviors, and educational attainment. To address multiple hypothesis testing and our small sample, we pre-registered an analysis plan of well–powered tests. As predicted, a newly constructed and pre-registered measure derived from preschool delay of gratification does not predict the 11 capital formation variables (i.e., the sign-adjusted average correlation was 0.02). A pre-registered composite self-regulation index, combining preschool delay of gratification with survey measures of self-regulation collected at ages 17, 27, and 37, does predict 10 of the 11 capital formation variables in the expected direction, with an average correlation of 0.19. The inclusion of the preschool delay of gratification measure in this composite index does not affect the index’s predictive power. We tested several hypothesized reasons that preschool delay of gratification does not have predictive power for our mid-life capital formation variables.

Keywords: Self-regulationDelay of gratificationMid-life capital formation

JEL: D910D140I310I210I120

6. Concluding remarks

We have reported pre-registered analysis of the latest survey wave of the Bing pre-school study on delay of gratification. Respondents were in their late 40’s at the time of this latest wave and were asked to report 11 mid-life capital formation outcomes (e.g., net worth and permanent income). Our analysis plan both described our methods and predicted what we would find. As predicted, a newly constructed measure derived from preschool delay of gratification does not predict the 11 capital formation variables (i.e., the sign-adjusted average correlation was 0.02). By contrast a composite self-regulation index, combining preschool delay of gratification with survey measures of self-regulation collected at ages 17, 27, and 37, does predict 10 of the 11 capital formation variables in the expected direction, with an average correlation of 0.19. The inclusion of the preschool delay of gratification measure in this composite index does not affect the index’s predictive power for two reasons. Most importantly, the index of self-regulatory measures is comprised of 86 responses per participant, whereas the preschool delay of gratification task is a single behavioral task. In addition, the preschool delay of gratification task is measured using a diagnostic variant of the task for 34 of our 113 participants; the remaining 79 participants experienced a non-diagnostic variant of the pre-school delay of gratification task.

The data we have analyzed is unique because the Bing cohort is the only sample where preschool delay of gratification has been studied long enough to examine relationships with mid-life outcomes. While the tests of our primary hypotheses were well powered, we caution that our sample is relatively small and not representative of the overall population—e.g., 97% of our sample has a four-year college degree (the exceptions are one participant who has a two-year college degree and two who have some college but no degree)—limiting the generalizability of the results.

However, we can compare our results to the small set of overlapping analyses that have been conducted using the Dunedin cohort, which began collecting data in 1972–1973 and has childhood self-regulation measures but no preschool delay of gratification measure.11 The cohort is from a small town in New Zealand with much lower levels of educational attainment (Moffitt et al., 2011). Specifically, 29% of the Dunedin sample is college educated (Belsky et al., 2016). Despite the stark socioeconomic differences, the self-regulation measures used in the Dunedin study have similar predictive power to the self-regulation measures in the Bing sample. For example, Moffitt et al. (2011) found that a 1 SD increase in childhood self-control as measured in the Dunedin study predicts a 0.24 SD increase in income at age 32. In the Bing sample, a 1 SD increase in RNSRI predicts a 0.32 SD increase in rank-normalized permanent income. Similarly, a 1 SD increase in the Dunedin self-control measure predicts a 0.12 SD decrease in credit card problems, while a 1 SD increase in the Bing RNSRI predicts a 0.18 SD decrease in rank-normalized credit card misuse. A 1 SD increase in the Dunedin self-control measure predicts a 0.14 SD decrease in money management difficulties, while a 1 SD increase in the Bing RNSRI predicts a 0.24 SD increase in financial health.12 Despite these intriguing similarities across the two samples, the issue of generalizability remains an important question to be addressed in future research as mid-life data becomes available in more childhood longitudinal cohorts.