The goal of this paper is to conduct a meta-regression analysis (MRA hereafter) regarding the effects of the Kaldor-Verdoorn effect – the relation between output/demand and productivity. The Kaldor-Verdoorn effect has been subject to many econometric studies and while the overwhelming majority of them finds a positive overall effect, there is no consensus on its size – the results vary quite a bit, especially according to the chosen econometric specification.
This MRA estimates a ‘true value’ of the Kaldor-Verdoorn effect without interference from potential publication selection bias via the use of multivariate MRA. A series of moderator variables is used to check for their effect of excess variation, including amongst others the year of publication, the sectors and the countries studied.
This MRA study uses available data from 22 published studies with 303 estimations of the Kaldor-Verdoorn effect. When examining the primary literature as a whole, there seems to be publication bias. While there seems to exist a genuine Kaldor-Verdoorn effect, its size varies considerably depending on the specification chosen.
Consulter ce document de travail