Working Paper: CEPR ID: DP3119
Authors: Michael J. Artis; Anindya Banerjee; Massimiliano Marcellino
Abstract: Time series models are often adopted for forecasting because of their simplicity and good performance. The number of parameters in these models increases quickly with the number of variables modelled, so that usually only univariate or small-scale multivariate models are considered. Yet, data are now readily available for a very large number of macroeconomic variables that are potentially useful when forecasting. Hence, in this Paper we construct a large macroeconomic data-set for the UK, with about 80 variables, model it using a dynamic factor model, and compare the resulting forecasts with those from a set of standard time series models. We find that just six factors are sufficient to explain 50% of the variability of all the variables in the data set. Moreover, these factors, which can be considered as the main driving forces of the economy, are related to key variables such as interest rates, monetary aggregates, prices, housing and labour market variables, and stock prices. Finally, the factor-based forecasts are shown to improve upon standard benchmarks for prices, real aggregates, and financial variables, at virtually no additional modelling or computational cost.
Keywords: factor models; forecasts; time series models
JEL Codes: C22; C51; C52; C53
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| identified factors (C38) | variability in macroeconomic indicators (E39) |
| use of factor models (C38) | forecasting outcomes (C53) |
| structural breaks addressed (L16) | accuracy of forecasts (C53) |