Estimating the late-life effects of social and emotional skills in childhood using midlife mediators

Social and emotional skills are known to affect health and non-health outcomes, but there is limited evidence on whether these skills in childhood affect late life outcomes because of a shortage of long-running datasets containing this information. We develop a three-stage procedure and use it to estimate the effect of childhood social and emotional skills on health and labour market outcomes in late-life. This procedure makes use of mediators in midlife which are shown to be predicted by childhood skills in one dataset and to predict late-life outcomes in another dataset. We use this method to combine estimates from the National Child Development Survey and the British Household Panel Survey. We find that childhood skills predict marital status, education, home ownership, income and health at age 46 years and these midlife variables predict levels of quality-adjusted life years and labour income accumulated by age 63 years. The combined estimates suggest a standard deviation increase in average Bristol Social Adjustment Guide total score at ages 7 and 11 is associated with 4.2% (standard error = 0.6%) additional quality-adjusted life years and more than 9.9% (£14,539, standard error = £2072) additional accumulated pre-tax earnings by age 63 years. Therefore, childhood interventions to increase social and emotional skills would be expected to reduce future healthcare costs and increase wealth. Our three-stage methodology can be used to predict the life-course effects of investments in childhood skills by combining results from datasets across population cohorts.