Sparse dynamic factor modelling in macroeconometrics
Sylvia Kaufmann ()
Additional contact information
Sylvia Kaufmann: Study Center Gerzensee
No 26.05, Working Papers from Swiss National Bank, Study Center Gerzensee
Abstract:
Factor models condense the common dynamics of high-dimensional data into a few factors. Inducing sparsity in the factor loading matrix identifies strong factorseries associations, so that factors load on distinct groups of series and can be interpreted. The paper reviews an order-invariant Bayesian approach to sparse factor estimation. The model is estimated without pre-imposing rotational identification restrictions, and factor position and sign are identified by post-processing the MCMC output. Since extracted factors are distinct, the restriction of independent factor innovations can be relaxed. The resulting reduced-form state equation, also in a factor augmented VAR with observed factors, allows for further structural analysis. To handle the recent pandemic period, we introduce a Covid factor that loads on the series only from 2020 onwards, while we assume the remaining factors and their dynamics to stay the same. An application to FRED-QD data, spanning the period 2000Q1 to 2024Q1, identifies interpretable group-specific factors in addition to the pervasive ones, and the Covid factor captures the transitory, extreme volatility at the onset of the pandemic across many series. Impulse responses to a policy shock identified by maximizing the forecast error variance of the term spread are consistent with a monetary policy shock, although uncertainty remains considerable. Allowing for weakly correlated idiosyncratic innovations is a promising extension that would bring parametric sparse factor models closer to approximate factor models.
Pages: 48 pages
Date: 2026-11
New Economics Papers: this item is included in nep-ets
References: Add references at CitEc
Citations:
Downloads: (external link)
https://www.szgerzensee.ch/fileadmin/Dateien_Anwen ... g_papers/wp-2605.pdf Full text (application/pdf)
None
Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.
Export reference: BibTeX
RIS (EndNote, ProCite, RefMan)
HTML/Text
Persistent link: https://EconPapers.repec.org/RePEc:szg:worpap:2605
Ordering information: This working paper can be ordered from
Studienzentrum Gerzensee, Postfach 21, 3115 Gerzensee
The price is Free.
Access Statistics for this paper
More papers in Working Papers from Swiss National Bank, Study Center Gerzensee Studienzentrum Gerzensee, Postfach 21, 3115 Gerzensee.
Bibliographic data for series maintained by library ().