This paper proposes a new classification of occupations based on the extent to which they put workers at risk of being infected by aerial-transmitted viruses. We expand on previous work that mainly focused on the identification of jobs that can be done from home by providing a more nuanced view of infection risks: in particular, we identify jobs that, although impossible to be done from home, expose workers to a low risk of infection. Jobs that cannot be done from home and that present a high risk of infection are labelled ‘unsafe jobs’. We then combine our classification of infection risk with a list of ‘essential occupations’ that have been carried out even during the most severe lockdown measures: this provides a taxonomy ranking jobs along two dimensions, one related to workers’ health and the other related to economic conditions. Using both survey and administrative data, we show that this taxonomy successfully predicts outcomes along these two dimensions, such as sick leaves, COVID-19-related work injuries, recourse to short-time work (STW) schemes and work from home. We also find that unsafe jobs are very unequally distributed across different types of workers, firms and sectors. Workers who are more vulnerable economically (women, youngsters, low educated, immigrants and workers on fixed-term contracts) are more likely to hold unsafe jobs and therefore more at risk of suffering from the economic consequences of a prolonged pandemic. We finally discuss possible paths to reform social protection systems, so that they can better support workers during the labour market adjustments that are likely to be spurred by the COVID-19 pandemic.

RésuméLes auteurs examinent la relation entre les gains, le capital humain et les tâches. Ils reprennent le modèle théorique d'Autor et Handel (2013), tout en élargissant l'analyse empirique à 21 pays, grâce aux données de l'enquête de l'OCDE sur les compétences des adultes (PIAAC). Celles‐ci permettent en outre de caractériser plus précisément le capital humain des travailleurs, en fournissant une mesure directe des compétences cognitives. L'analyse empirique confirme l'influence du contenu en tâches de l'emploi sur les écarts de salaire, à la fois entre les catégories professionnelles et en leur sein. Les prédictions du modèle se vérifient dans la grande majorité des pays.

This article investigates the links between earnings, human capital and job tasks, using internationally comparable information from the OECD Survey of Adult Skills. The authors use the theoretical framework developed by Autor and Handel (2013) and extend their empirical results to 21 OECD countries. The data allow for a richer characterization of workers' human capital by including both educational attainment and a measure of cognitive skills. Their findings confirm the predictive power of job tasks in explaining wage differences, both between and within occupations, and provide empirical support for the model's prediction in the vast majority of countries.

Numeracy and literacy are important foundation skills which command significant wage premia in modern labour markets.The existence of gender differences in these skills is therefore of potential concern, and has spurred a large amount of research, especially with respect to numeracy skills.Still, little is known about the moment in which such gaps emerge, how they evolve, and if this evolution differs across countries.We use data from large scale international assessments to follow representative samples of birth-cohorts over time, and analyse how gender gaps in numeracy and literacy evolve from age 10 to age 27.We find that the advantage of boys in numeracy is small at age 10, but grows considerably between age 15 and 27.The gender gap in literacy follows a very different pattern: it is small at age 10, large and in favour of girls at age 15, and negligible by age 27.

This paper illustrates how process data can be used to identify behavioral patterns in a computer-based problem-solving assessment. Using sequence-mining techniques, we identify patterns of behavior across multiple digital tasks from the sequences of actions undertaken by respondents. We then examine how respondents' action sequences (which we label "strategies") differ from optimal strategies. In our application, optimality is defined ex-ante as the sequence of actions that content experts involved in the development of the assessment tasks identified as most efficient to solve the task given the range of possible actions available to test-takers. Data on 7462 respondents from five countries (the United Kingdom, Ireland, Japan, the Netherlands, and the United States) participating in the Problem Solving in Technology-Rich Environment (PSTRE) assessment, administered as part of the OECD Programme for the International Assessment of Adult Competencies (PIAAC), indicate that valuable insights can be derived from the analysis of process data. Adults who follow optimal strategies are more likely to obtain high scores in the PSTRE assessment, while low performers consistently adopt strategies that are very distant from optimal ones. Very few high performers are able to solve the items in an efficient way, i.e. by minimizing the number of actions and by avoiding undertaking unnecessary or redundant actions. Women and adults above the age of 40 are more likely to adopt sub-optimal problem-solving strategies.

Resumen En este artículo se investigan los vínculos entre ingresos, capital humano y tareas laborales, utilizando información internacionalmente comparable del PIAAC (OCDE) sobre competencias de adultos. Tomando el marco teórico de Autor y Handel (2013), se amplía el análisis empírico a 21 países de la OCDE. Los datos permiten tener en cuenta tanto el nivel de instrucción como una medida de las competencias cognitivas. Los resultados confirman que las tareas que implica cada ocupación tienen influencia en las diferencias salariales tanto entre ocupaciones como dentro de cada una y proporcionan pruebas empíricas del poder de predicción del modelo para un gran número de países.

This paper analyses several dimensions of workers' safety that are relevant in the context of a pandemic. We provide a classification of occupations according to the risk of contagion: by considering a wider range of job characteristics and a more nuanced assessment of infection risk, we expand on the previous literature that almost exclusively looked at feasibility of working from home. We apply our classification to the United States and to European countries and we find that roughly 50% of jobs in our sample can be considered safe, although a large cross-country variation exists, notably in the potential incidence of remote working. We find that the most economically vulnerable workers (low-educated, low-wage workers, immigrants, workers on temporary contracts, and part-timers) are over-represented in unsafe jobs, notably in non-essential activities. We assess the nature of the reallocation of workers from unsafe to safe jobs that is likely to take place in the years to come, and the policies that could mitigate the social cost of this reallocation.

This paper illustrates similarities and differences between two international surveys that assess adults’ skills: the Programme for the International Assessment of Adult Competencies (PIAAC) and the Skills Towards Employment and Productivity (STEP) survey. In particular, the paper highlights the issues that can arise for researchers interested to jointly analyse the data from the two surveys and to compare their results. The paper finds that, in spite of the many similarities, important differences exist between PIAAC and STEP, both in the way the data are collected, and in the way the proficiency of respondents is estimated. These issues can indeed affect the cross-country comparability of results from the two surveys. There is instead little evidence that the literacy assessment used in the two surveys is not adequate to form a basis for a valid assessment of adults’ proficiency on a global scale.

Using data on multiple cohorts of Italian students we investigate how the ordinal rank in the within-school-cohort age distribution affects the probability of being bullied (age rank effect). Identification is achieved exploiting variation in the age composition of different school cohorts, and through an IV strategy based on the discontinuity in the probability of enrolling in a given school year generated by an end-of-year cut-off rule. We find that a one-decile increase in the within-school-cohort age distribution decreases the probability of being victimized by about one percentage point. Age rank effects are stronger when calculated within groups defined by observable characteristics.

The value of cross-country comparisons is at the heart of large-scale international surveys. Yet the validity of such comparisons is often challenged, particularly in the case of latent traits whose estimates are based on self-reported answers to a small number of questionnaire items. Many believe self-reports to be unreliable and not comparable, and indeed, formal statistical procedures very often reject the assumption that the questions are understood and answered in the same way in different countries (measurement invariance). A methodological conference on the comparability of questionnaire scales was hosted by the OECD on 8 and 9 November 2018. This meeting report summarises the discussions held at the conference about measurement invariance testing and instrument design. The report first provides a brief introduction to the measurement models and the accompanying invariance analyses typically used in the industry of large-scale international surveys and points to the main limitations of these current standard approaches. It then presents classical and novel ways to deal with imperfect comparability of measurements when scaling and reporting on continuous traits and on categorical latent variables. It finally discusses the extent to which item design can improve the cross-country comparability of the measured constructs (e.g. by adopting innovative item formats such as anchoring vignettes and situational judgement test items). It concludes with some general considerations for survey design and reporting on invariance analyses and survey results.

In most countries urban workers enjoy higher wages than non urban ones, and this premium increases with the size of the city. In this paper we show that in Italy hourly wages of private-sector workers are 6% higher in urban areas than in non-urban local labor markets (less than 2% controlling for observable workers’ characteristics); this premium is higher for more educated workers and for women. More generally, as the local population grows, hourly wages tend to increase: doubling population increases wages by 2.1% (less than 1% net of workers’ characteristics). Even larger gaps are usually estimated in other developed countries. Using an employer-employee dataset and a standard AKM wage decomposition, we divide Italian wages into two components, one proxying for worker’s skills and the other one proxying for firm’s quality. We find that better workers and better firms both tend to sort themselves in urban areas. Nevertheless, the sorting of workers seems to be more relevant than the sorting of firms, resulting in a larger urban premium for the workers’ component. The sorting of firms is almost entirely explained by a few characteristics of the local labor market, such as higher educational attainment and labor market participation.

Large-scale surveys such as the Programme for International Student Assessment (PISA), the Teaching and Learning International Survey (TALIS), and the Programme for the International Assessment of Adult Competences (PIAAC) use advanced statistical models to estimate scores of latent traits from multiple observed responses. The comparison of such estimated scores across different groups of respondents is valid to the extent that the same set of estimated parameters holds in each group surveyed. This issue of invariance of parameter estimates is addressed in model fit indices which gauge the likelihood that one set of parameters can be used across all groups. Therefore, the problem of scale invariance across groups of respondents can typically be framed as the question of how well a single model fits the responses of all groups. However, the procedures used to evaluate the fit of these models pose a series of theoretical and practical problems. The most commonly applied procedures to establish invariance of cognitive and non-cognitive scales across countries in large-scale surveys are developed within the framework of confirmatory factor analysis and item response theory. The criteria that are commonly applied to evaluate the fit of such models, such as the decrement of the Comparative Fit Index in confirmatory factor analysis, work normally well in the comparison of a small number of countries or groups, but can perform poorly in large-scale surveys featuring a large number of countries. More specifically, the common criteria often result in the non-rejection of metric invariance; however, the step from metric invariance (i.e. identical factor loadings across countries) to scalar invariance (i.e. identical intercepts, in addition to identical factor loadings) appears to set overly restrictive standards for scalar invariance (i.e. identical intercepts). This report sets out to identify and apply novel procedures to evaluate model fit across a large number of groups, or novel scaling models that are more likely to pass common model fit criteria.

In many countries urban workers enjoy higher wages than non urban ones, and the wage premium increases with the size of the city. In this paper we show that this wage premium can go beyond the city boundaries and affect also workers located elsewhere through spillovers in productivity between plants/firms. In particular, we analyze spillovers in the urban wage premium along: i) the internal structure of the firm (i.e. between different plants belonging to the same firm), ii) the value chain (i.e. different firms linked through client-supplier relationships). We rely on a matched employers-employee dataset in order to control for firms’ and workers’ characteristics. Our results confirm that workers enjoy a wage premium when they work in an urban area; this premium is larger for workers in headquarters than in branches. Despite that, workers in branches benefit from the localization of their headquarter: when firm’s headquarter is in an urban area, workers in non-urban branches receive a wage premium of about 3%. Furthermore, working for a firm supplying goods or services to urban firms provides a wage premium that is almost as large as the one a worker would get in a urban non-supplier firm. According to our results, these spillovers are not driven by workforce composition; they reflect the fact that plants/firms connected to cities have larger rents to share with workers. We also conclude that the presence of such withinand between-firms spillovers, by equalizing wages across urban and non-urban locations, results in a downward bias in standard estimates of the urban wage premium and in an under-appreciation of the economic importance of cities.

This paper analyses proficiency in literacy and numeracy in the countries that have participated in the International Adult Literacy Survey (IALS, administered between 1994 and 1998), the Adult Literacy and Life Skills Survey (ALL, administered between 2003 and 2007) and the Survey of Adult Skills (PIAAC, administered in 2012). While many countries experienced small to modest changes in literacy proficiency between IALS and PIAAC, others saw sizeable variations, mostly on the negative side. In the shorter span that separated ALL and PIAAC, numeracy proficiency clearly declined (except in Italy), while literacy moved less on average (except for the large increase registered in Italy and the large decline experienced by Norway). Changes in the composition of the population have had little impact on observed changes in scores. Larger variations took place within different socio-demographic groups, but these tended to cancel each other out on aggregate. In particular, large variations are observed by age and levels of education. Older adults in PIAAC are generally more proficient than their IALS counterparts, probably due to the increase in educational attainments that took place over recent decades. On the contrary, tertiary-educated individuals appear to be on average less proficient than in the past, which may signal that the expansion of tertiary education has been accompanied by a decline in the average quality of university graduates (or of university instruction). There is also no evidence that the change in delivery mode, with a switch to a computer-based assessment in PIAAC, had any significant effect on performance. However, the OECD is unable to ascertain how differences in implementation and technical standards affect the comparability of the data, so that a certain degree of caution should always be exercised in interpreting these results. Amongst the countries that experienced larger changes in literacy proficiency between surveys, a close inspection of IALS data (in particular through an investigation of response patterns at the item level) highlights some anomalies in Italy and Poland (and, to a lesser extent, in England and Northern Ireland), suggesting that particular caution should be exercised in interpreting the evolution of proficiency in these countries.

This paper estimates the impact of college teaching on students' academic achievement and labor market outcomes using administrative data from Bocconi University matched with tax records. The estimation exploits the random allocation of students to teachers in a fixed sequence of compulsory courses. We find that teacher effects on students' academic and labor market outcomes are only mildly positively correlated and that the professors who are best at improving the academic achievement of their students are not always also the ones who boost their earnings the most. For the least able students, the correlation between the academic and labor market effectiveness of teachers turns out to be negative.

Using census data on four cohorts of Italian students enrolled in 5 grade between 2010 and 2013, we investigate how age affects the probability of being bullied, distinguishing between the contribution of absolute and relative age. The absolute age effect is identified by means of an IV strategy that exploits the discontinuity in the probability of enrolling in a given school year generated by an end-of-year cut-off rule. In such a setting, theoretical school starting age is a valid instrument for absolute age. The relative age effect is identified exploiting cohort-to-cohort variation within schools in the average age of enrolled students, also instrumented by (average) theoretical school starting age. We find that only absolute age negatively affects the chances of being bullied, while relative age has no effect. This result implies that policies aimed at preventing bullying through changes in the age composition of schools and classes are unlikely to yield significant benefits. JEL Classification: I21, J24, Z13

In this paper we estimate the short-run elasticity of substitution between male and female workers, using data on employment and wages from Italian provinces from 1993-2006. We adopt a production function approach similar to that of Card and Lemieux (2001a) and Acemoglu, Autor and Lyle (2004). Our identification strategy relies on a natural experiment. In 2000 the Italian parliament passed a law to abolish compulsory military service; the reform was implemented through a gradual reduction in the number of draftees, and compulsory drafting was definitively terminated in 2004. We use data on the (planned) maximum number of draftees at the national level (as stated in the annual budgetary law), interacted with sex-ratios at birth at the provincial level, as instruments for (relative) female labor supply. Our results suggest that young males and females are imperfect substitutes, with an elasticity of substitution ranging between 1.1 and 1.6. Our results have implications for the evaluation of policies aimed at increasing female labor market participation, suggesting that they do not necessarily displace male employment.

This paper exploits data from the Survey of Adult Skills (PIAAC) to shed light on the link between measured cognitive skills (proficiency), (formal) educational attainment and labour market outcomes. After presenting descriptive statistics on the degree of dispersion in the distributions of proficiency and wages, the paper shows that the cross-country correlation between these two dimensions of inequality is very low and, if anything, negative. As a next step, the paper provides estimates of the impact of both proficiency and formal education at different parts of the distribution of earnings. Formal education is found to have a larger impact on inequality, given that returns to education are in general much higher at the top than at the bottom of the distribution. The profile of returns to proficiency, by contrary, is much flatter. This is consistent with the idea that PIAAC measures rather general skills, while at the top end of the distribution the labour market rewards specialised knowledge that is necessarily acquired through tertiary and graduate education. Finally, a decomposition exercise shows that composition effects are able to explain a very limited amount of the observed cross-country differences in wage inequality. This suggests that economic institutions, by shaping the way personal characteristics are rewarded in the labour market, are the main determinants of wage inequality. Ce document exploite les donnees de l'Evaluation des competences des adultes (PIAAC) pour tenter de mieux comprendre le lien entre les competences cognitives mesurees (le niveau de competence), le niveau de formation (dans le cadre institutionnel) et les resultats sur le marche du travail. Apres la presentation de statistiques descriptives sur le degre de dispersion des distributions des niveaux de competence et des revenus, le document montre que la correlation internationale entre ces deux dimensions d’inegalite est tres faible et, le cas echeant, negative. Le document presente ensuite des estimations de l'incidence a la fois du niveau de competence et du niveau de formation dans le cadre institutionnel a differents points de la distribution des revenus. Le niveau de formation dans le cadre institutionnel s’avere avoir une incidence plus importante sur l'inegalite, les rendements de l'education etant en general bien plus eleves dans la partie superieure de la distribution que dans sa partie inferieure. Les rendements du niveau de competence presentent, en revanche, un profil beaucoup plus plat. Ce constat concorde avec le fait que le PIAAC evalue des competences plutot generales, tandis qu’au sommet de la distribution, le marche du travail recompense des connaissances specialisees necessairement acquises dans l'enseignement superieur et universitaire. Enfin, un exercice de decomposition montre que les effets de composition ne sont en mesure d'expliquer qu’un nombre tres limite des differences d'inegalite des revenus observees entre les pays. Ce constat laisse penser qu’en faconnant la maniere dont les caracteristiques personnelles sont recompensees sur le marche du travail, les institutions economiques sont les principaux determinants de l'inegalite des revenus.