The economic value of breaking bad: How childhood misbehaviour can signal something good

Nicholas W. Papageorge, Victor Ronda, Yu Zheng 04 May 2019



Educators and policymakers are increasingly interested in developing students’ non-cognitive skills to boost their academic performance, employability, as well as long-term physical and mental well-being. In England, the national curriculum states that personal, social, health, and economic education, which aims to build young people’s skills in confidence, resilience, self-esteem, communication, and teamwork, should be provided in all schools and an ongoing campaign aims to make this curriculum statutory (Long 2019). In the US, schools in eight California districts have already started testing students on socio-emotional skills such as self-control and conscientiousness and have included such measures in the school accountability system (West et al. 2016).

Interest in these policies follows a growing body of research in economics and psychology that demonstrates the importance of socio-emotional skills for a wide range of economic outcomes, such as employment and lifetime earnings (Almlund et al. 2011, Borghans et al. 2008). Such research also shows that socio-emotional skills are malleable during childhood and adolescence (Heckman and Kautz 2014). 

The conventional wisdom is that socio-emotional skills are another dimension of human capital, whose return may differ across occupations or sectors in the same way manual skills and academic skills do (Willis and Rosen 1979). However, conventional wisdom is also that these skills are generally productivity-enhancing, that socio-emotional skills typically improve education attainment as well as labour market outcomes. Our research challenges this view as we demonstrate that some socio-emotional skills associated with lower schooling are productive in the labour market (Papageorge et al. 2019). This finding implies that well-intentioned policies aimed at reducing so-called ‘negative’ behaviours can instead harm children in the long run.

The socio-emotional skills we study are externalising behaviour and internalising behaviour, both of which have been widely studied in psychology and child development literature. Externalising behaviour is known to be linked to aggression and hyperactivity, while internalising behaviour is known to be connected to anxiety and depression (Ghodsian 1977, Duncan and Magnuson 2011, Duncan and Dunifon 2012). 

Using a longitudinal dataset from Britain, the National Child Development Survey, we measure these skills from teacher reports of children’s behaviour in the classroom when the children were 11 years old. We rely on a measurement error model and a latent-factor approach to relate the measurements of the two socio-emotional skills to schooling and labour market performance measured in adulthood. We perform the analyses separately for males and females.

Our key empirical finding is that, for both genders, externalising behaviour lowers educational attainment but is also associated with higher earnings. For males, externalising behaviour boosts earnings mainly through raising wages, while for females it boosts earnings primarily through an increase in hours worked. In our benchmark specification, one standard deviation increase in externalising behaviour raises the hourly wage by 6.4% for males, while it raises hours worked by 4.7% for females, both effects significant and larger in magnitude than the effects from cognition. 

We verify that these empirical patterns hold not only for a 1958 British cohort, but also hold for a younger 1970 British cohort (using the 1970 British Cohort Study), and for US cohorts (using the National Education Longitudinal Study of 1988 and the Panel Study of Income Dynamics). In other words, we provide robust and compelling evidence, across cohorts and countries, that externalising behaviour, which is bad for schooling, carries an earnings premium. A penchant for ‘breaking bad’ can be good.

Given the basic empirical pattern, we further ask two questions. First, we ask how intermediate choices and outcomes – such as occupational sorting, marriage, and fertility – affect the way socio-emotional skills impact earnings. Though externalising behaviour does affect life-cycle outcomes realised before hours and wages are measured, conditioning on the intermediate outcomes does not eliminate the externalising behaviour’s earnings premium. In fact, for females, conditioning on the intermediate outcomes tends to increase externalising behaviour’s positive impact on earnings. This is because highly externalising women are also more likely to marry and have children, which tends to decrease earnings. After controlling for its negative effect on earnings through the fertility channel, externalising behaviour by itself predicts even higher earnings for women. The evidence supports the view that externalising behaviour represents a skill that is valuable in the labour market.

The second question we ask is whether the labour-market returns to externalising behaviour vary across socioeconomic groups. In previous literature, Heckman et al. (2013) show that the Perry Preschool Program raised earnings among a group of American schoolchildren partly through reducing the children’s externalising behaviour. In contrast, we find that for a 1958 British cohort, externalising behaviour on average increases earnings. To explore this difference, we construct a subsample from the British cohort that experienced financial difficulties during childhood to mimic the financially disadvantaged participants of the Perry Preschool Program. We show that indeed among individuals who grew up in poverty, externalising behaviour no longer carries a significant earnings premium. This finding, though reassuring from an academic point of view, is troubling from a policy point of view, since it suggests that individuals who are already disadvantaged due to childhood poverty are excluded from realising the full returns to skills that are lucrative for individuals born into wealthier families.

In sum, we have amassed considerable empirical evidence that one socio-emotional skill, externalising behaviour, has opposite returns in two sectors, the educational and the labour market sectors. The opposite returns reveal a mismatch between the types of skills promoted in school and the skills valuable at work. Schooling environments designed to achieve better academic performance by moulding certain socio-emotional skills – such as in the direction of obedience and conformity – may stifle or penalise skills that create value in the labour market. More generally, our findings illustrate that it is not meaningful to think of socio-emotional skills as either good or bad per se. This offers a sharp contrast to cognition or health, which would improve outcomes on almost any conceivable economic dimension. Thus, investment in socio-emotional skills should be evaluated in light of the possibility of mixed effects. 

More research is needed to understand the impact of socio-emotional skills across sectors of life and across different socioeconomic groups of children, before becoming a target of education policymaking. In the case of the externalising behaviour we study, a relationship-oriented paedagogical method proposed by Allen at al. (2011) could be a good solution. Different from control-oriented teaching methods, relationship-oriented methods are designed to strengthen the learning environment for externalising children by channelling their outwardly expressed behaviour to educational use instead of suppressing the behaviour.


Allen, J P, R C Pianta, A Gregory, A Y Mikami and J Lun (2011), “An interaction-based approach to enhancing secondary school instruction and student achievement”, Science 333(6045): 1034–1037. 

Almlund, M, A L Duckworth, J Heckman and T Kautz (2011), “Personality psychology and economics”, Handbook of the Economics of Education 4(1). 

Borghans, L, A L Duckworth, J J Heckman and B Ter Weel (2008), “The economics and psychology of personality traits”, Journal of Human Resources 43(4): 972-1059. 

Duncan, G J, and R Dunifon (2012), “‘Soft-skills’ and long-run labor market success”, in S W Polachek and K Tatsiramos (eds.), 5th Anniversary Retrospective (Research in Labor Economics, Volume 35), Emerald Group Publishing Limited, 313–339. 

Duncan, G J, and K Magnuson (2011), “The nature and impact of early achievement skills, attention skills, and behavior problems”, in Whither Opportunity? Rising Inequality, Schools, and Children’s Life Chances, Russell Sage Foundation, 47–69. 

Ghodsian, M (1977), “Children’s behavior and the BSAG: Some theoretical and statistical considerations”, British Journal of Social and Clinical Psychology 16(1): 23–28. 

Heckman, J J, and T Kautz (2014), “Fostering and measuring skills: Interventions that improve character and cognition”, in J J Heckman, J E Humphries and T Kautz (eds.), The Myth of Achievement Tests, University of Chicago Press, 341-430. 

Heckman, J J, R Pinto and P Savelyev (2013), “Understanding the mechanisms through which an influential early childhood program boosted adult outcomes”, American Economic Review 103(6): 2052–86. 

Long, R (2019), “Personal, social, health and economic education in schools (England)”, Briefing Paper 07303, House of Commons Library.

Papageorge, N W, V Ronda and Y Zheng (2019), “The economic value of breaking bad: Misbehavior, schooling and the labor market”, NBER Working Paper 25602.

West, M R, M A Kraft, A S Finn, R E Martin, A L Duckworth, C F O Gabrieli and J D E Gabrieli (2016), “Promise and paradox measuring students non-cognitive skills and the impact of schooling”, Educational Evaluation and Policy Analysis 38(1): 148–170. 

Willis, R J, and S Rosen (1979), “Education and self-selection”, Journal of Political Economy 87(5): S7–S36.



Topics:  Education Labour markets

Tags:  education, US, UK, earnings, personality traits, non-cognitive skills, socio-emotional skills

Broadus Mitchell Assistant Professor of Economics, Johns Hopkins University

Assistant Professor of Economics, Aarhus University

Lecturer, School of Economics and Finance, Queen Mary University of London and CEPR Research Affiliate


CEPR Policy Research