Time-varying general dynamic factor models and the measurement of financial connectedness

Abstract

We propose a new time-varying Generalized Dynamic Factor Model for high-dimensional, locally stationary time series. Estimation is based on dynamic principal component analysis jointly with singular VAR estimation, and extends to the locally stationary case the one-sided estimation method proposed by Forni et al. (2017) for stationary data. We prove consistency of our estimators of time-varying impulse response functions as both the sample size T and the dimension n of the time series grow to infinity. This approach is used in an empirical application in order to construct a time-varying measure of financial connectedness for a large panel of adjusted intra-day log ranges of stocks. We show that large increases in long-run connectedness are associated with the main financial turmoils. Moreover, we provide evidence of a significant heterogeneity in the dynamic responses to common shocks in time and over different scales, as well as across industrial sectors.

Dynamic network connectedness of BRICS equity markets during the Covid-19 era

“…As extensively examined by scholars, connectedness between financial assets prones to intensify around financial/geopolitical bursts (Barigozzi et al, 2020;Fernández-Rodríguez & Sosvilla-Rivero, 2020;Lee & Lee, 2020;Polat, 2020). The COVID-19 pandemic sets an example for such a geopolitical turmoil and connectedness between financial markets has dramatically soared following the pandemic (Bissoondoyal-Bheenick et al, 2020;Bouri et al, 2020;So et al, 2020).…”

Manfred Deistler and the General Dynamic Factor Model Approach to the Analysis of High-Dimensional Time Series

“…Proposition. Let {Y t | t ∈ Z} be a tall process satisfying (9) for some rational Q × q filter D(L). Then, for Π P -generic values of D(L) (P unspecified), Y t admits, for some K < ∞, a VAR(K) representation of the form…”

“…that is, generically, 9 for some A n (L), R n and v t , we have the block-diagonal VAR representation…”

A Neural Network Ensemble Approach for GDP Forecasting

“…We compute one-quarter to four-quarter ahead forecasts to evaluate differences between the time-varying DFM and neural networks over different time horizons, as well as assessing when it is suitable to put them together in an ensemble model to forecast along the good and bad turns of the business cycle. As a matter of fact, dynamic factor models are widely used within the context of macroeconomic nowcasting and forecasting, and many specifications include dynamics in the parameters as well as the possibility to model breaks along the economic cycle [Del Negro and Otrok, 2008, Camacho et al, 2012, Lee, 2012, Korobilis, 2013, Barigozzi et al, 2020. We adopt a score-driven approach with a GAS, similarly to Creal et al [2013], as a way of capturing parameter dynamics in the DFM specification.…”