3 Dynamic Factor Models (DFM) with different factor loadings fitted, and their insample and psuedo out-of-sample prediction accuraries are explored.
The Covid-19 pandemic has had a profound and long lasting impact on the global economy. It caused a global economic recession starting in February 2020. Many macroeconomic indicators exhibited a high volatility during the first half of 2020, but many also quickly rebounded to pre-covid levels just a few months later. This paper explores how different models perform in capturing the strong volatility of macroeconomic indicators during this period, using pseudo out-of-sample forecasting for the period of January 2020 to December 2020. We find that certain models were able to capture the volatility better than others. We also find evidence that this may be due to non-stationary relationships between indicators of different categories. The results of this paper highlights the importance of model specification and to ensure that a model is consistent with the underlying macroeconomic relationships that are more likely to break during periods of high volatility.