Note there's an exploding literature that reads these models as causal models: http://ftp.cs.ucla.edu/pub/stat_ser/r350.pdf http://ftp.cs.ucla.edu/pub/stat_ser/

Note there's an exploding literature that reads these models as causal models: http://ftp.cs.ucla.edu/pub/stat_ser/r350.pdf http://ftp.cs.ucla.edu/pub/stat_ser/r350.pdf