Quelle est la valeur du coefficient d'Okun? Une analyse de type meta-regression

We selected a sample of 522 estimates of the Okun’s Law coeffi cient from literature to uncover the reasons for the diff erences in empirical results across studies and to estimate the ‘true’ OLC with meta-regression methodology. We show that there is evidence of both type I and type II bias amongst the papers studied. Taking into account those biases and then evaluating the 95% confi dence intervals of the true eff ects estimated with various methodologies, we fi nd that the estimated true OLC vary between a minimum value of –0,40 and a maximum value of –0,12. Moreover, the estimated OLCs are signifi cantly larger (in absolute value) with models using regional data or a linear time trend fi ltering procedure. On the other hand we fi nd that lower absolute values of the OLC are obtained with models using quarterly (or semestrial) data bases, data for developing countries, or non-linear models.