Quantifying Human Capital: Applications to the Study of the Institutional Organisation of Capitalism
In the first instance the focus of the paper is twofold. Firstly, it attempts to transcend the singular focus in the comparative political economy literature on the specificity and transferability of skills and complements it with an equal emphasis on their productive value. Secondly, given its centrality to much empirical research in the field, it is highly desirable that measurements of human capital are based on the direct quantitative analysis of empirical data, something which is not indicative of existing methods. The approach which is formulated reconciles both of these issues by developing new quantitative estimators for the asset specificity and the asset value of human capital. Measurements are derived from the estimation of binary logit models and generalised ordered probit models based on cross-sectional data. In addition, binary logit models with random effects as well as generalised ordered probit models with random effects are run on panel data.
The applications of these indicators are subsequently tested in a series of quantitative analyses which relate to four of the most prominent substantive issues in comparative political economy research on the institutional organisation of capitalism. Firstly, OLS regressions are run in order to estimate the relationship between different national skill profiles and national economic growth, productivity and labour mobility using IMF, OECD and panel data. Secondly, time-series cross-section regressions are run in order to estimate the relationship between different kinds of human capital and cross-national patterns of welfare spending and generosity over time using OECD and labour force survey data. Thirdly, logit and probit models are run in order to estimate the relationship between differences in human capital and divisions in access to social protection both within and across countries over time using labour force survey and panel data. Fourthly, logit and probit models are run in order to test the relationship between skill profiles and individual institutional/redistributional preferences using ISSP data.