Your timing structure for capital seems correct, but the one for net worth is almost surely wrong.
If you have
vahat // bank net worth (after the shock) deviations from trend
vbhat // bank net worth (before the shock) deviations from trend
then the second one must be predetermined with respect to the shock at time t. But how can they then be contemporaneously related as in
vahat*vastar=xihat+vbhat*vbstar;
The discussion at Expectations vs realized might be relevant as well.