Ricardian equivalence and the media narrative of fiscal policy: a time series study of the Swedish economy

Sammanfattning

This study tests whether the Ricardian Equivalence Hypothesis holds in Sweden when the private sector faces an intertemporal budget constraint skewed by the media narrative. Time series analysis of cointegrated vector autoregressive processes is utilized to identify Ricardian consumer behavior in three different models, in order to find differences in consumer behavior when media narrative bias is accounted for versus when it is not accounted for. Large language models are used to create an index of media perceptions of government fiscal policy. Cointegrating vectors are studied for testing of the no-Ponzi scheme condition, and the structured vector error correction model is used to model the impact on consumption from fiscal expansion and government financing changes. The results show that the no-Ponzi scheme condition is violated because the private and public sector over-compensate by saving. Furthermore, results show no significant effect from government spending on consumption, but do show that consumption increases in the short run following a financing restructuring towards a higher degree of taxation. The effects of media narrative bias are undetermined, but there are indicators that available information is crucial for a consumer response to policy changes.