Psychometrics has long played a significant role in the transformation of human emotions into numbers (Michell 2008), and is being transformed by innovations in ‘digital phenotyping’ and ‘algorithmic psychometrics’ which treat bodies as transmitters of affective data (Stark 2018). By re-presenting affective experiences as quantitative inscriptions and digital data, psychometrics is a key technique for the diagnosis, management, and governance of human emotion and personality (Dror 2001). In this paper I examine the cross-organizational collaborations, forms of expertise, and psychometric techniques that are integral to the formation of the OECD’s most recent international assessment, the Survey on Social and Emotional Skills (SSES). SSES is an online survey of social-emotional learning based on a particular classification model for coding and quantifying ‘non-cognitive skills’. Drawing on documentary analysis, the paper will demonstrate how SSES metrics have been co-produced by an expert network (consisting of the OECD, economists, psychometricians, and software developers) and in the process how the quantification of affects is changing the priorities of the OECD itself. Together, this cross-sector expert network has brought together interdisciplinary expertise into a new kind of ‘policy knowledge’ (Littoz-Monnet 2017), encompassing human capital theory, labour market analytics, personality theory, behavioural economics, and positive psychology. The objective of SSES is to capture social-emotional indicators of socio-economic outcomes through psychometric techniques of emotional coding and personality classification. Through the survey, the OECD is also reinventing itself by extending its testing infrastructure from the calculation and presentation of cognitive skills to noncognitive skills. Its rationale is a radical shift in economies as jobs become increasingly automated and robotized through advances in artificial intelligence. As the OECD’s Schleicher (2018) argues, ‘routine cognitive skills, the skills that are easiest to teach and easiest to test, are exactly the skills that are also easiest to digitise, automate and outsource’. Whereas the ‘knowledge economy’ was focused on knowledge workers with high cognitive capacity as human capital, in the ‘robot economy’ computerized ‘cognitive intelligence’ will be paired with human ‘emotional intelligence’ (Mantas 2018), bringing about a need for education systems to produce measurable ‘human-computer capital’ to prevent humans ‘losing their economic value’ (Schleicher 2018). SSES represents an advance in international assessment to measure economically-valuable information about the emotions and personality, thereby linking measurable socio-emotional indicators to socio-economic outcomes in the robot economy, and instantiating digital psychometrics in transnational education governance

Dror, O.E. 2001. Counting the affects: Discoursing in numbers. Social Research 68(2): 357-378. Littoz-Monnet, A. (ed.) 2017. The Politics of Expertise in International Organizations: How international bureaucracies produce and mobilize knowledge. London: Routledge. Mantas, J. 2018. Why artificial intelligence is learning emotional intelligence. World Economic Forum, 12 September: https://www.weforum.org/agenda/2018/09/why-artificial-intelligence-is-learning-emotional-intelligence/ Michell, J. 2008. Is Psychometrics Pathological Science? Measurement 6(1-2): 7-24. Stark, L. 2018. Algorithmic psychometrics and the scalable subject. Social Studies of Science 48(2): 204-231. Schleicher, A. 2018. World Class: How to Build a 21st-Century School System. Paris: OECD