Comment

Abstract

The objective of the paper by Aruoba et al. is to construct an index of the global business cycle, more precisely, an index tracking the cycle of Group of 7 countries. The index is computed using GDP data, which are available at a quarterly frequency, and monthly business cycle indicators. The latter are included for capturing features of the business cycle that are not summarized in the GDP series and to obtain an indicator that is available at a higher frequency than GDP. The econometric model is based on the assumption that the observable series have a factor structure representing global and countryspecific comovements across series. The model is estimated in two steps: first, for each country separately, the authors extract a common (country) factor; second, they extract the global factor by estimating a factor model on the country factors extracted in the previous step. Following standard practice, the model is written in its state space form, and the Kalman filter is used to handle the mixed-frequency data problem. The global index is defined as the estimated common factor obtained in the second step. Another exercise proposed by the paper is to use the same approach for each single country in order to extract, for each model, the seven national factors and then study the relation between the latter and the global index. Empirically, the results of the paper are not surprising. The factors are generally highly correlated with GDP. The description of depth and length of various recessions as well as the declining volatility of the cycle are well-known facts that do not require more than eyeballing econometrics to be identified. However, there are also some puzzles. For example, what is the interpretation of the fact that the U.S. country factors