تحليل التحول النظامي لأثر النفقات العمومية على مؤشرات الاقتصاد الكلي في الجزائر (2000-2023) باستعمال نماذج Markov-switching

Abstract

This study aims to analyze the effect of public expenditure on macroeconomic indicators in Algeria over the period from the first quarter of 2000 to the fourth quarter of 2023, testing the hypothesis that this effect is non-linear and varies across economic regimes. To overcome the shortcomings of conventional linear models (VAR and ARDL), which fail to capture the structural shifts of a rentier economy, the study adopted a Markov-Switching Vector Autoregression model, MS(2)-VAR(1), estimated through the Expectation-Maximization (EM) algorithm, using five variables: public expenditure, growth rate, inflation, unemployment, and oil price. The preliminary tests (BDS and LR) confirmed the existence of systematic non-linearity and the superiority of the regime-switching model over its linear counterpart. The results reveal two distinct economic regimes: an oil-boom expansionary regime accounting for about 68.75% of the period, and a shock-and-recession regime accounting for 31.25%. Most importantly, the fiscal multiplier proved to be regime-dependent: it reaches about 1.12 in the boom regime versus only 0.35 in the shock regime, a ratio close to 1:3.2; moreover, the employment effect of spending vanishes while its inflationary cost doubles during crises. The study concludes that the effectiveness of fiscal policy in Algeria is contingent on the availability of oil rent, and that linear modeling provides a misleading picture of the true relationship. It recommends adopting binding counter-cyclical fiscal rules and restructuring expenditure toward productive quality.