Asymptotics of the principal components estimator of large factor models with weakly influential factors

By A. Onatski. 2012

275 citations recorded by the source corpus.

Subfields: Economics and Econometrics, Finance

View Paper: https://www.semanticscholar.org/paper/fd71b4c9603c8b5f00ff474cf7f8c06716d0efe8

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

1. Jushan Bai
2. Liangjun Su
3. Serena Ng
4. Jianqing Fan
5. H. R. Moon
6. Markus Pelger
7. Martin P. Weidner
8. Yuan Liao

In Finance

1. Jushan Bai
2. Serena Ng
3. Zhaoguo Zhan
4. Markus Pelger
5. Yuan Liao
6. Lorenzo Trapani
7. F. Kleibergen
8. Liangjun Su

Citations in the corpus

• PROJECTED PRINCIPAL COMPONENT ANALYSIS IN FACTOR MODELS (2014, background): “As is shown in Onatski (2012), when α = 0, the principal components estimator of the factors is inconsistent, because the strength of factors does not dominate the idiosyncratic errors.”
• Statistical inference for principal components of spiked covariance matrices (2020, methodology): “For instance, sparse principal component analysis [15, 41] is proposed to estimate the spiked eigenvalues and eigenvectors assuming some sparsity structure in the population eigenvectors; factor-model based estimators [2, 3, 66] for the eigenvectors are constructed if the population covariance matri”
• NETS: Network estimation for time series (2019, unknown): “However, this is not ruling out either cross-sectional or serial dependence, and weakly influential factors are still possible (Ando & Bai, 2017; Onatski, 2012).”
• Large Dimensional Latent Factor Modeling with Missing Observations and Applications to Causal Inference (2019, unknown): “Note that weak factors are also a form of misspecification, as discussed in Onatski (2012). In this simulation, observations are more likely to miss if they are exposed to the omitted or weak second factor.”
• Sparse Principal Component Analysis and Iterative Thresholding (2011, background): “When both n,p → ∞ with n/p → c ∈ (0,∞), at different levels of rigor and generality, this phenomenon has been examined by a number of authors [9, 14, 17, 19, 21, 25] under model (1.”
• Spiked separable covariance matrices and principal components (2019, background): “High-dimensional data obtained at space-time points has been increasingly employed in various scientific fields, such as geophysical and environmental sciences [32, 36], wireless communications [28, 52, 54], medical imaging [49] and financial economics [41, 42, 59].”
• Shrinkage Estimation of Dynamic Panel Data Models with Interactive Fixed Effects (2016, unknown): “…contains two penalty terms, 1 for the regression coefficients ’s and 2 for the loading vectors ·’s. Noting that −12 k·k = (1) under our Assumption A.1(iii) in Section 3.1 which apparently rules out the case of weak factors studied by Onatski (2012), we divide the second penalty term 2 by √ .”
• Estimation of large dimensional factor models with an unknown number of breaks (2018, background): “2015), Lam and Yao (2012), Onatski (2012) and Wang and Fan (2017) for recent advancement in estimation methods and inference theory in large-dimensional factor modeling.”
• Two-sample tests for high-dimension, strongly spiked eigenvalue models (2016, background): “For eigenvalues of high-dimensional data, Jung and Marron (2009), Yata and Aoshima (2012, 2013b), Onatski (2012) and Fan, Liao, and Mincheva (2013) considered spiked models such as λ ij → ∞ as p → ∞ for j = 1 , ..., k i , with some positive integer k i .”
• A diagnostic criterion for approximate factor structure (2016, background): “FF, CRR, MOM, REV, LIQ and JW refer to the three Fama-French factors, the five Chen-Roll-Ross macroeconomic factors, the momentum factor, the reversal factors, the three liquidity factors of Pastor and Stambaugh (2002), and the three Jagannathan and Wang (1996) factors, respectively.”
• Linear Regression for Panel With Unknown Number of Factors as Interactive Fixed Effects (2015, background): “7See Onatski (2010; 2012) and Chudik, Pesaran and Tosetti (2011) for a discussion of “strong” vs. “weak” factors in factor models. perturbation theory of linear operators to derive the approximation.”
• Forecasting with Factor-Augmented Regression: A Frequentist Model Averaging Approach, Second Version (2013, unknown): “This does not accommodate weak factors as in Onatski (2012).”
• Estimating latent asset-pricing factors (2018, methodology): “As pointed out by Onatski (2012), a weak factors can be regarded as a finite sample approximation for strong factors, i.e. the eigenvalues of factors that are theoretically strong grow so slowly with the sample size that the weak factor model provides a more appropriate description of the…”
• Regional Policy Evaluation: Interactive Fixed Effects and Synthetic Controls (2016, unknown): “Moreover, we adopt the assumption that factors are sufficiently strong so that the consistency condition for the number of factors and consequently for factors and factor loadings is satisfied (for alternative views, see Onatski, 2012, or Pesaran & Tosetti, 2011).”
• On the Marčenko-Pastur law for linear time series (2013, background): “Onatski [29] describes a model for production Yt , at time t , involving p different industries in an economy that is given by the equations”
• Factors That Fit the Time Series and Cross-Section of Stock Returns (2018, unknown): “Onatski (2010) shows that the pattern of successive eigenvalues can be used to determine the number of systematic factors and proposes an eigenvalue difference estimator. He argues that idiosyncratic eigenvalues cluster around a single point, i.e. successive eigenvalues are similar in magnitudes. In”