No abstract

The empirical analysis of economic growth usually starts with the linearity assumption, implying only linear or trended movement in output growth and its relation with the explanatory variables. This is relevant when the countries under analysis follow the "Solow" balanced growth pattern, characterized by no significant fluctuations in the macroeconomic data series in the long run and usually described as variation around a single trend, which means that the variations are negligible and do not affect the linear trend in the data. However, sometimes big exogenous shocks such as Global Financial Crisis, pandemic, wars and conflicts around the World impact growth pattern, causing big shifts in growth process in the countries. This impact is especially relevant in the case of developing or transition countries, where the big shocks cause shifts in growth pattern, named growth regimes with specific properties for each regime. The growth process observed through various growth regimes instead of singular growth path was supported by the findings of many scholars who called for specification of a nonlinear data generating process for analysing the impact of big shocks on economic growth. In this paper the main objective is to examine the deviations of real economic activity, measured by the GDP growth rate from some linear trend, by the use of Markov Switching model in the case of North Macedonia. North Macedonia is good example to test for non-linearities in growth patten due to the big shocks and adjustment happening in the course of last three decades such as structural changes of transition, conflicts, Global Financial Crisis, Covid-pandemic. The results suggest that the real economic activity changes before and after some shock or regime shift occurs, characterised with specific mean growth rate and specific volatility within the regime. Hence, the conclusion is that the possible nonlinear notion of economic growth should be taken into account when conduction growth analysis, but also when defining the economic growth programmes in the countries, especially developing ones.

Empirical investigation on growth, especially in the course of big exogenous shocks is still relatively ambiguous. While models of developed economies describe the growth process as a smooth movement along the balanced growth path, this pattern is altered in the cases of big exogenous shocks that hit economies. The later accumulation of evidence, including the big shocks due to pandemic and Russia - Ukraine war led to more realistic specifications of growth models as well, putting the emphasize on the impact of shocks on growth processes. Hence, the main objective of this paper is to provide a review of the papers that investigate empirically growth and its main determinants, with the accent on analysis of the impact of shocks on growth patterns. A distinction is made between studies that investigate developed economies and those that concentrate on developing or transition economies. The former ones usually apply the modelling strategy based on the neoclassical linear framework and, their review offers valuable insight and overview of the variables mostly used in growth studies. On the other side, the studies concerning developing or transition countries are rather focused on finding new ways of empirically modelling growth in specific conditions of shocks. Although still the majority of the analyses are based on the linearity assumption, in this review we shall treat only the ones that introduce non-linearity in the growth studies, assessing the ways they address the non-linearity observed in the data generating processes. Finally, the proposed strategy for modelling big shocks in the growth pattern is adjusted growth accounting formula with Markov Switching Vector Autoregressive modelling. The modest number of variables is due to the intention of the empirical exercise to put a focus on the shifts in growth rather than on the detailed determinants behind the shifts. Additionally, it should be emphasized that the informative purpose of this empirical model is limited by the lack of data for other possible variables and by the modelling procedure.