Central Bank of Luxembourg publishes research proposing a VAR-based approach to measuring business cycles

The Central Bank of Luxembourg has published a research paper (Cahier d’études No. 201) by Patrick Feve and Alban Moura that proposes a new way to measure the business cycle using a vector autoregression (VAR) model that jointly describes the dynamics of macroeconomic variables. The approach is intended to address the lack of consensus in business-cycle measurement, where different methods can yield divergent readings of cyclical conditions. The method proceeds in two steps: it first isolates and removes unstable components in the data, then identifies shocks whose effects have a duration close to that conventionally associated with business cycles. The resulting cycle measure is defined by the contributions of these shocks to past fluctuations in the observed variables. The authors highlight four advantages, including anchoring trend-cycle decomposition in the data’s statistical properties, producing cycles that revert to a constant mean, targeting fluctuations of typical business-cycle length while preserving time-series structure, and distinguishing typical from atypical episodes. Applied to US data, the paper finds that two shocks can account for the main features of business cycles, with estimated cycles closely matching the official recession and expansion chronology and identifying the Great Recession as the most pronounced episode. The results also suggest a tighter link between inflation and real activity than is commonly assumed, with robustness tests presented as supporting the reliability of the approach; the publication notes that the views expressed are those of the authors and not necessarily those of the Central Bank of Luxembourg or the Eurosystem.