Consistent Factor Estimation in Dynamic Factor Models with Structural Instability
By Brandon Bates, Mikkel Plagborg-Møller, James H. Stock, Mark W. Watson. 2013
140 citations recorded by the source corpus.
This paper considers the estimation of approximate dynamic factor models when there is temporal instability in the factor loadings. We characterize the type and magnitude of instabilities under which the principal components estimator of the factors is consistent and find that these instabilities can be larger than earlier theoretical calculations suggest. We also discuss implications of our results for the robustness of regressions based on the estimated factors and of estimates of the number of factors in the presence of parameter instability. Simulations calibrated to an empirical application indicate that instability in the factor loadings has a limited impact on estimation of the factor space and diffusion index forecasting, whereas estimation of the number of factors is more substantially affected.
Subfields: Economics and Econometrics
View Paper: https://www.semanticscholar.org/paper/c42c28cf66df71b193f846e28b6eb7a7d040e97f
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Economics and Econometrics
Citations in the corpus
- A new index of financial conditions (2014, unknown): “πt|t−1,j = παt−1|t−1,j ∑J l=1 π α t−1|t−1,l , (4)”
- TESTS FOR PARAMETER INSTABILITY IN DYNAMIC FACTOR MODELS (2013, background): “While the estimated factors still consistently span the original factor space if the size of break is small enough (Stock and Watson, 2002b; Bates et al., 2012), such results do not hold when the size of breaks is large.”
- Detecting big structural breaks in large factor models (2014, background): “Finally, it is easy to verify that ∑N i=1 ∑T t=1 ∥ηi∥E∥Ft∥ = O(1) and thus the new idiosyncratic errors εt satisfy the necessary condition for the consistency of r̂ (See Observation 1 of Bates et al. 2013). Once it has been shown that the new factors: F ∗ t , the new loadings: Γ and the new errors: ”
- Simultaneous multiple change-point and factor analysis for high-dimensional time series (2016, unknown): “…in Assumptions 2 (ii) and 4 (ii), along with the mixing condition in Assumption 5 (ii), allow us to control the deviation of sample covariance estimates from their population counterparts, via Bernstein-type inequality (see e.g., Theorem 1.4 of Bosq (1998) and Theorem 1 of Merlevède et al. (2011).”
- Dynamic factor model with infinite‐dimensional factor space: Forecasting (2016, background): “…Han and Inoue (2014), Yamamoto and Tanaka (2015), Chen et al. (2014), Barigozzi et al. (2016), (ii) consistent factor estimation under breaks, Bates et al. (2013), Ma and Su (2016), Cheng et al. (2016), Massacci (2016), (iii) modeling time varying factor models Del Negro and Otrok (2008),…”
- State-Varying Factor Models of Large Dimensions (2018, methodology): “It includes Andrews (1993), Chen, Dolado, and Gonzalo (2014), Breitung and Eickmeier (2011), Cheng, Liao, and Schorfheide (2016), Baltagi, Kao, and Wang (2020), Bai, Han, and Shi (2020), Ma and Su (2018), Barigozzi, Cho, and Fryzlewicz (2018), and Chen, Wang, and Wu (2020).9 A factor structure,…”
- Identification and estimation of a large factor model with structural instability (2017, methodology): “Our design roughly follows that of Bates et al. (2013), with the focus switching from small change to large change and from forecasting to estimating the whole model, i.e., estimating the change point, the number of pre-break and post-break factors and the pre-break and post-break factor spaces.”
- Testing for Factor Loading Structural Change under Common Breaks (2015, methodology): “To see the size, the data is generated by (5) with the factor loadings (j) i = i for all j drawn from 14The cases Bates et al. (2013) consider include this example.”