Human Capital Data: Evidence and Suggestions from Italian Time Series (1881-1993)

The controversial results obtained by introducing human capital in the traditional regression growth pushed the researchers to search in different directions: refining the measurement of human capital, adding measures of quality of schooling, etc. The paper copes with the econometric side of the regression of human capital on GDP per capita investigating the characteristics of the time series representing human capital. The usual regression techniques assume a linear relationship between the dependent variable and the independent ones, and the independent and identical distribution hypothesis on residuals. The paper shows that this is not the case, for both the conditions, in the series of secondary school enrollment rate to the population of the same age in Italy from 1881 to 1991. Therefore using standard econometric techniques to investigate the role of human capital on GDP is incorrect. As an alternative approach, the paper proposes a kernel regression which is a non parametric method by which observations are weighted according to a particular function. The results are comfortable. The regression finds a non linear relationship between economic growth and human capital, showing that human capital is significative for Italian GDP growth since 1965, when the Reform of Secondary School (1962) displayed its effects. The paper extends this method to the traditional factors (K;L) and finds a similar relationship and timing for labour, while only capital time series show a linear relationship as is usually presumed in the standard production function. * Address all communications to Renato Giannetti, University of Florence, Dipartimento di Studi Storici e Geografici, via San Gallo 10, 50129, Firenze, Italy. E-mail: giannetti @unifi.it and Margherita Velucchi, University of Siena, e-mail: [email protected].