1: \documentclass[12pt,twoside,reqno]{amspaper}

2: \usepackage[edges]{mathdef}

3: \usepackage{macros}

4: \usepackage{natbib}

5: \bibpunct{(}{)}{,}{a}{}{,}

6: \newcommand{\bibfont}{\small}

7: \renewcommand{\pastx}{\mathbf{X}}

8: \renewcommand{\ffilt}[1]{\mathscr{X}_{\!{#1}}}

9: \usepackage{comment}

10: %

11: \begin{document}

12: \title{Graphical modelling of multivariate time series}

13: %\title{Markov properties for graphical time series models}

14: \author{Michael Eichler}

15: \thanks{The paper was written while the author was working at the

16: Institut f\"ur Angewandte Mathematik, Universit\"at Heidelberg, Germany.}

17: \thanks{{\em E-mail address:} m.eichler@ke.unimaas.nl (M.~Eichler)}

18: \affil{Department of Quantitative Economics,

19: University of Maastricht\\P.O.~Box 616, 6200 MD Maastricht, The Netherlands}

20: \dedicatory{{\upshape\today}}

21: \begin{abstract}

22: We introduce graphical time series models for the analysis

23: of dynamic relationships among variables in multivariate

24: time series. The modelling approach is based on the notion of strong

25: Granger causality and can be applied to time series with non-linear

26: dependencies. The models are derived from ordinary time series models

27: by imposing constraints that are encoded by mixed graphs.

28: In these graphs each component series is represented

29: by a single vertex and directed edges indicate possible Granger-causal

30: relationships between variables while undirected edges are used to

31: map the contemporaneous dependence structure. We introduce

32: various notions of Granger-causal Markov properties and

33: discuss the relationships among them and to other Markov

34: properties that can be applied in this context.

35: \\[1em]

36: {{\itshape Keywords:} Graphical models, multivariate time series,

37: Granger causality, global Markov property}

38: \end{abstract}

39: \maketitle

40: 

41: \input{nsect1}

42: \input{nsect2}

43: \input{nsect3}

44: \input{nsect4}

45: %\input{sect5}

46: \input{conclusion}

47: %\section*{Acknowledgement}

48: %Part of the results presented here have been taken from the

49: %author's doctoral thesis \citep{eichlerthesis}. The author thanks

50: %his supervisor Rainer Dahlhaus and two referees for helpful

51: %comments.

52: \begin{appendix}

53: \input{appendixa}

54: \input{appendixb}

55: \input{appendixc}

56: \input{appendixd}

57: \end{appendix}

58: \bibliography{papers,application}

59: \bibliographystyle{stat}

60: 

61: \end{document}

62: \end{input}

63: