Identifiability and estimation of structural vector autoregressive models for subsampled and mixed-frequency time series.

Causal inference in multivariate time series is challenging because the sampling rate may not be as fast as the time scale of the causal interactions, so the observed series is a subsampled version of the desired series. Furthermore, series may be observed at different sampling rates, yielding mixed-frequency series. To determine instantaneous and lagged effects between series at the causal scale, we take a model-based approach that relies on structural vector autoregressive models. We present a unifying framework for parameter identifiability and estimation under subsampling and mixed frequencies when the noise, or shocks, is non-Gaussian. By studying the structural case, we develop identifiability and estimation methods for the causal structure of lagged and instantaneous effects at the desired time scale. We further derive an exact expectation-maximization algorithm for inference in both subsampled and mixed-frequency settings. We validate our approach in simulated scenarios and on a climate and an econometric dataset.

[1]  S. Rachev.  Handbook of heavy tailed distributions in finance , 2003 .

[2]  C. Stein,et al.  Structural equation modeling. , 2012, Methods in molecular biology.

[3]  Aapo Hyvärinen,et al.  Causality Discovery with Additive Disturbances: An Information-Theoretical Perspective , 2009, ECML/PKDD.

[4]  Brian D. O. Anderson,et al.  Identifiability of regular and singular multivariate autoregressive models from mixed frequency data , 2012, 2012 IEEE 51st IEEE Conference on Decision and Control (CDC).

[5]  Giorgio E. Primiceri,et al.  The Time Varying Volatility of Macroeconomic Fluctuations , 2006 .

[6]  H. Herwartz,et al.  The macroeconomic effects of oil price shocks: Evidence from a statistical identification approach , 2016 .

[7]  Aapo Hyvärinen,et al.  Estimation of a Structural Vector Autoregression Model Using Non-Gaussianity , 2010, J. Mach. Learn. Res..

[8]  D. Stram,et al.  A METHODOLOGICAL NOTE ON THE DISAGGREGATION OF TIME SERIES TOTALS , 1986 .

[9]  Frank Schorfheide,et al.  Real-Time Forecasting With a Mixed-Frequency VAR , 2013 .

[10]  Markku Lanne,et al.  Modeling Conditional Skewness in Stock Returns , 2007 .

[11]  D. Danks,et al.  Learning Causal Structure from Undersampled Time Series , 2013 .

[12]  Ruslan Salakhutdinov,et al.  Adaptive Overrelaxed Bound Optimization Methods , 2003, ICML.

[13]  J. Boot,et al.  Further Methods of Derivation of Quarterly Figures from Annual Data , 1967 .

[14]  David Cai,et al.  Analysis of sampling artifacts on the Granger causality analysis for topology extraction of neuronal dynamics , 2014, Front. Comput. Neurosci..

[15]  R. Cont.  Empirical properties of asset returns: stylized facts and statistical issues , 2001 .

[16]  Andrew T. Foerster,et al.  Bayesian Mixed Frequency VARs , 2015 .

[17]  Peter A. Zadrozny.  Labor Statistics Office of Prices and Living Conditions Extended Yule-Walker Identification of Varma Models with Single-or Mixed-Frequency Data , 2016 .

[18]  Campbell R. Harvey,et al.  Conditional Skewness in Asset Pricing Tests , 1999 .

[19]  Aapo Hyvärinen,et al.  Causal modelling combining instantaneous and lagged effects: an identifiable model based on non-Gaussianity , 2008, ICML '08.

[20]  Ali Shojaie,et al.  Penalized likelihood methods for estimation of sparse high-dimensional directed acyclic graphs. , 2009, Biometrika.

[21]  E. Oja,et al.  Independent Component Analysis , 2013 .

[22]  Cointegration Analysis with Mixed-Frequency Data of Quarterly GDP and Monthly Coincident Indicators , 2012 .

[23]  Andrew Harvey,et al.  Forecasting, Structural Time Series Models and the Kalman Filter , 1990 .

[24]  Aapo Hyvärinen,et al.  A Linear Non-Gaussian Acyclic Model for Causal Discovery , 2006, J. Mach. Learn. Res..

[25]  H. Luetkepohl,et al.  Structural Vector Autoregressions With Nonnormal Residuals , 2006, SSRN Electronic Journal.

[26]  Massimiliano Marcellino,et al.  A Survey of Econometric Methods for Mixed-Frequency Data , 2013 .

[27]  David Veredas,et al.  Temporal Aggregation of Univariate and Multivariate Time Series Models: A Survey , 2008 .

[28]  Bernhard Schölkopf,et al.  Causal Inference on Time Series using Restricted Structural Equation Models , 2013, NIPS.

[29]  Michael I. Jordan.  Graphical Models , 2003 .

[30]  E. Ghysels,et al.  Macroeconomics and the Reality of Mixed Frequency Data , 2015 .

[31]  L. Kilian,et al.  Structural Vector Autoregressive Analysis , 2017 .

[32]  E. Fox,et al.  A Unifying Framework for Devising Efficient and Irreversible MCMC Samplers , 2016 .

[33]  P. Saikkonen,et al.  Identification and estimation of non-Gaussian structural vector autoregressions , 2015 .

[34]  H. Lütkepohl,et al.  Structural Vector Autoregressions with Markov Switching , 2010 .

[35]  Sergey M. Plis,et al.  Causal Discovery from Subsampled Time Series Data by Constraint Optimization , 2016, Probabilistic Graphical Models.

[36]  B. Anderson,et al.  MULTIVARIATE AR SYSTEMS AND MIXED FREQUENCY DATA: G-IDENTIFIABILITY AND ESTIMATION , 2016, Econometric Theory.

[37]  Filippo Moauro,et al.  Temporal Disaggregation Using Multivariate Structural Time Series Models , 2005 .

[38]  Helmut Ltkepohl,et al.  New Introduction to Multiple Time Series Analysis , 2007 .

[39]  W. Walls.  Modelling heavy tails and skewness in film returns , 2005 .

[40]  Norman R. Swanson,et al.  Temporal aggregation and spurious instantaneous causality in multiple time series models , 2002 .

[41]  Bernhard Schölkopf,et al.  Discovering Temporal Causal Relations from Subsampled Data , 2015, ICML.

[42]  Visa Koivunen,et al.  Identifiability, separability, and uniqueness of linear ICA models , 2004, IEEE Signal Processing Letters.

[43]  Vince D. Calhoun,et al.  Rate-Agnostic (Causal) Structure Learning , 2015, NIPS.

[44]  Peter A. Zadrozny,et al.  AN EXTENDED YULE-WALKER METHOD FOR ESTIMATING A VECTOR AUTOREGRESSIVE MODEL WITH MIXED-FREQUENCEY DATA , 1999 .