[For part I, see ibid. 35, No. 5, 1790-1810 (1997; Zbl 0891.93085).]
The authors investigate the infinite horizon, partially observed risk-sensitive control problems with finite state space and long run average cost. Defining an appropriate information state, they solve the dynamic programming equation, which is a nonlinear eigenvalue problem, and derive an optimal output feedback control. They also analyze stochastic dynamic games and robust control problems related to their results.