Bayesian analysis of dynamic factor models: An ex-post approach towards the rotation problem

Due to their indeterminacies, static and dynamic factor models require identifying assumptions to guarantee uniqueness of the parameter estimates. The indeterminacy of the parameter estimates with respect to orthogonal transformations is known as the rotation problem. The typical strategy in Bayesian factor analysis to solve the rotation problem is to introduce ex-ante constraints on certain model parameters via degenerate and truncated prior distributions. This strategy, however, results in posterior distributions whose shapes depend on the ordering of the variables in the data set. We propose an alternative approach where the rotation problem is solved ex-post using Procrustean postprocessing. The resulting order invariance of the posterior estimates is illustrated in a simulation study and an empirical application using a well-known data set containing 120 macroeconomic time series. Favorable properties of the ex-post approach with respect to convergence, statistical and numerical accuracy are revealed.

[1]  Ajay Jasra,et al.  Markov Chain Monte Carlo Methods and the Label Switching Problem in Bayesian Mixture Modeling , 2005 .

[2]  R. Redner,et al.  Mixture densities, maximum likelihood, and the EM algorithm , 1984 .

[3]  M. West,et al.  High-Dimensional Sparse Factor Modeling: Applications in Gene Expression Genomics , 2008, Journal of the American Statistical Association.

[4]  Andrew R. Barron,et al.  Information-theoretic asymptotics of Bayes methods , 1990, IEEE Trans. Inf. Theory.

[5]  S. Frühwirth-Schnatter.  Fully Bayesian Analysis of Switching Gaussian State Space Models , 2001 .

[6]  C. Robert,et al.  Computational and Inferential Difficulties with Mixture Posterior Distributions , 2000 .

[7]  P. Green,et al.  On Bayesian Analysis of Mixtures with an Unknown Number of Components (with discussion) , 1997 .

[8]  Roger E. Millsap.  When Trivial Constraints Are Not Trivial: The Choice of Uniqueness Constraints in Confirmatory Factor Analysis , 2001 .

[9]  Julius O. Smith,et al.  Introduction to Digital Filters: with Audio Applications , 2007 .

[10]  Gilles Celeux,et al.  Bayesian Inference for Mixture: The Label Switching Problem , 1998, COMPSTAT.

[11]  Jonathan D. Cryer,et al.  Time Series Analysis , 1986 .

[12]  J. S. Tanaka,et al.  Problems with EM algorithms for ML factor analysis , 1983 .

[13]  P. Schönemann,et al.  A solution to the weighted procrustes problem in which the transformation is in agreement with the loss function , 1976 .

[14]  T. W. Anderson,et al.  Generation of random orthogonal matrices , 1987 .

[15]  Jean Boivin,et al.  Measuring the Effects of Monetary Policy: A Factor-Augmented Vector Autoregressive (FAVAR) Approach , 2003 .

[16]  Sylvia Frühwirth-Schnatter,et al.  Finite Mixture and Markov Switching Models , 2006 .

[17]  J. Geweke,et al.  Measuring the pricing error of the arbitrage pricing theory , 1996 .

[18]  P. Schönemann,et al.  A generalized solution of the orthogonal procrustes problem , 1966 .

[19]  Xiao-Li Meng,et al.  The Art of Data Augmentation , 2001 .

[20]  Dorothy T. Thayer,et al.  EM algorithms for ML factor analysis , 1982 .

[21]  A. Gelman,et al.  Using Redundant Parameterizations to Fit Hierarchical Models , 2008 .

[22]  Hedibert Freitas Lopes,et al.  Parsimonious Bayesian Factor Analysis when the Number of Factors is Unknown , 2010 .

[23]  Deborah F. Swayne,et al.  A weighted procrustes criterion , 1991 .

[24]  Xiao-Li Meng,et al.  Seeking efficient data augmentation schemes via conditional and marginal augmentation , 1999 .

[25]  Friedrich Leisch,et al.  Dealing with label switching in mixture models under genuine multimodality , 2009, J. Multivar. Anal..

[26]  R. Kohn,et al.  On Gibbs sampling for state space models , 1994 .

[27]  Dongchu Sun,et al.  Bayesian Estimates for Vector Autoregressive Models , 2005 .

[28]  John Geweke,et al.  Evaluating the accuracy of sampling-based approaches to the calculation of posterior moments , 1991 .

[29]  Christian Schumacher,et al.  Bayesian estimation of sparse dynamic factor models with order-independent identification , 2013 .

[30]  Robert I. Jennrich,et al.  Rotational equivalence of factor loading matrices with specified values , 1978 .

[31]  J. Whitaker,et al.  Ensemble Square Root Filters , 2003, Statistical Methods for Climate Scientists.

[32]  M. Stephens.  Dealing with label switching in mixture models , 2000 .

[33]  B. Turlach,et al.  High-derivative parametric enhancements of nonparametric curve estimators , 1999 .

[34]  M. West,et al.  Bayesian Dynamic Factor Models and Portfolio Allocation , 2000 .

[35]  W. Newey,et al.  A Simple, Positive Semi-Definite, Heteroskedasticity and Autocorrelationconsistent Covariance Matrix , 1986 .

[36]  Arthur S. Goldberger,et al.  Structural Equation Models in the Social Sciences. , 1974 .

[37]  Jushan Bai,et al.  Identification and Estimation of Dynamic Factor Models , 2012 .

[38]  Michael A. West,et al.  BAYESIAN MODEL ASSESSMENT IN FACTOR ANALYSIS , 2004 .

[39]  Herman Rubin,et al.  Statistical Inference in Factor Analysis , 1956 .

[40]  Gene H. Golub,et al.  Matrix computations , 1983 .