Methods for Pastcasting, Nowcasting and Forecasting Using Factor-MIDAS with an Application to Real-Time Korean GDP *

We discuss a variety of recent methodological advances that can be used to estimate mixed frequency factor-MIDAS models for the purpose of pastcasting, nowcasting, and forecasting. In order to illustrate the uses of this methodology, we introduce a new real-time Korean GDP dataset, and carry out a series of prediction experiments, using a two step approach. In a first step, we estimate common latent factors (i.e., diffusion indices) from 190 monthly macroeconomic and financial series using various estimation strategies. Second, we use these factors, along with standard variables measured at multiple different frequencies, in six varieties of factor-MIDAS prediction models. Our key empirical findings are that: (i) When using realtime data, factor-MIDAS prediction models outperform various linear benchmark models. Interestingly, the ‘MSFE-best’ MIDAS models contain no AR lag terms when pastcasting and nowcasting. Indeed, AR terms only begin to play a role in model specification at 6-month ahead horizons. (ii) Models that utilize only 1 or 2 factors are ‘MSFE-best’ at all forecast horizons, except those associated with so-called pastcasting and nowcasting. (iii) Real-time data are crucial for forecasting Korean GDP, and the use of ‘first available’ versus ‘most recent’ data ‘strongly’ affects model selection and performance. (iv) Recursively estimated models are almost always ‘MSFE-best’, models estimated using autoregressive interpolation dominate those estimated using other interpolation methods, and factors estimated using recursive principal component estimation methods have more predictive content than those estimated using a variety of other (more sophisticated) approaches. (v) Factor-MIDAS models which constitute the ‘MSFE-best’ models, across many forecast horizons, estimation schemes, and data vintages, perform best when factors are estimated recursively.