This paper proposes a continuous time asset pricing model where fundamentals follow hidden Markov regime-switching processes. Then we introduce an ambiguity-averse economic agent who puts less (more) posterior probability on the desirable (undesirable) state than those who follows Bayesian updating rule. Such an agent would underestimate the expected growth rates of fundamentals, which lowers equity prices. It is shown that, when the agent is less risk averse, our model generates large equity premium with low risk-free rate. From our simulation results, it is also shown that the current model captures many dynamic properties of asset returns, including the procyclical variation of equity prices, the countercyclical variation of both equity premium and the volatility of equity returns, and the predictability of excess equity returns.