Asymetries in the Response of Unemployment to Self-Employment Shocks

Abstract

The aim of this article is to revisit the effects of self-employment shocks on unemployment at the macro-level, taking into account the potential existence of asymmetric effects across nations. In particular, we examine whether the employment intensity of self-employment depends on national wealth and/or labor market conditions. To this end, we employ three alternative panel threshold regression approaches. First, and preliminary we use a lagged fixed-effect non dynamic panel `a la Hansen (1999) based on the bootstrapped linearity test procedure and endogenous threshold retrieval as proposed by Candelon, Colletaz and Hurlin (2013). Second, a contemporaneity-related endogeneity as well as endogeneity of the regressand and its consequent downward bias in the estimated elasticities is thus accounted for with the Seo and Shin (2016) panel dynamic first differenced GMM estimator. Third, we use a more flexible model, and we test for the actual speed of adjustment of our unemployment estimates with respect to business cycle transitions with Gonzalez et al. (2017) smooth logistic panel estimator. Using a panel of 23 OECD countries over three decades, our results -subjected to a triple robustness check– qualify the results of previous literature. Our empirical findings indicate that: i) a linear relationship is rejected in favor of an asymmetric response of unemployment to self-employment shocks being the GDP per capita level the threshold variable; ii) the estimation of the unemployment elasticity to variations in self-employment- is quite robust regardless of the econometric approach considered; iii) the employment intensity of self-employment, -i.e., its capacity to reduce unemployment, is higher for countries that are above a certain threshold during a specific period. This intensity is highly dependent on the relative national wealth situation rather than unemployment dynamics; iv) as corollary, achieving the same unemployment reduction goal requires a higher intensity of entrepreneurial policy shock if the economy in question is below the threshold. This circumstance should be taken into account in the design of entrepreneurial promotion policies, which should aim to be non-horizontal or neutral across countries.

Keywords: entrepreneurship, self-employment, unemployment, non-linear models, panel threshold models