Analyzing business cycle asymmetries in a multi-level factor model

- We study the business cycle in the US over 1959-2011 using a large-dimensional multi-level factor model.
- The model features general (symmetric) and phase-specific (asymmetric) factors.
- A simple maximum likelihood method for estimating the factors and factor loadings is proposed.
- Comovement turns out to be higher during recessions than during expansions.
- The recession (expansion) factor is highly correlated with financial (real activity) variables.

We study the business cycle in the US over 1959-2011 using a large-dimensional multi-level factor model. We find notable asymmetries over the business cycle, but the bulk of common dynamics is stable over time. The comovement among variables is larger in recessions compared to expansions. The recession factor is highly correlated with monetary and financial variables, whereas expansion and symmetric factors are mainly related to real activity variables.