Growth and volatility in the European Union: a linear or a non-parametric approach?

Abstract

The paper reports an investigation of the relationship between business-cycle volatility and economic growth in the European Union before (EU-15) and after broadening (EU-25) using cross-sectional models. Results based on the superiority of the pseudo-likelihood ratio test as a more reliable tool for testing the robustness of the model used, indicate that the above relationship is linear only in the case of EU-15.

Notes

¹ FDI is a variable which influences positively the GDP (Crespo-Cuaresma et al., 2001).

² S ₁ follows a chi-square distribution with one degree of freedom. In addition the normality assumption for both models was verified via the normality test of Kolmogorov–Smirnov. There are two possible explanations for heteroscedasticity. First of all, entry to the EU brought substantial modifications to certain sectors, such as tax policy. The impact of these changes is reflected through the non-constant variance of the disturbance term. Another reason of non-constant variance is that we do not take account of changes to the ‘within’ variability of GDP. That is, the variability among provinces for every country.

³ Durio and Isaia (2004) demonstrate concrete examples.

⁴ The relationship between growth and volatility in EU-25 when dependent variable is median growth rate is not tested as long as is not supported the linearity via OLS.