Empirical Economics Социальные науки (прочее)
Identifying Useful Indicators for Nowcasting GDP in Sweden
Аннотация
This paper focuses on identifying useful indicators for nowcasting GDP in Sweden. We analyze 36 monthly indicators and their nowcast performance for the period 2003Q1 to 2025Q4 in a pseudo real-time setting. Additionally, we examine two distinct periods: from 2003Q1 to 2018Q4 (pre-pandemic) and from 2019Q1 to 2025Q4 (pandemic and post-pandemic). The analysis is conducted using mixed-data sampling (MIDAS) and mixed-frequency VAR models and we evaluate the pooled predictions from the model for each indicator to robustify against model misspecification. For nowcasting we identify 24 indicators that improve on the AR-benchmark and 11 that are particularly useful. Splitting the evaluation sample in the pre-pandemic and pandemic and post-pandemic periods we find that the results differ substantially with more indicators improving on the benchmark in the pandemic and post-pandemic period than in the pre-pandemic period. In addition, the set of highly informative indicators is quite different in the two periods.