I study the effects of short sale constraints in a rational framework with asymmetric information. I consider the cases of Bernoulli-distributed (a la Glosten and Milgrom) and continuously distributed (a la Kyle) liquidation values, and focus on the latter case. In this case my model is able to explain the following features of stock returns: (i) Black's "leverage effect" (a.k.a. asymmetric volatility), (ii) persistent volatility, and (iii) more negative skewness for longer horizon returns. Model implications for price impact are as follows: (i) the impact has typically lower magnitude compared to the unconstrained case, (ii) the impact of extremely positive news has higher magnitude than the impact of extremely negative news, (iii) the impact of moderately positive news has lower magnitude than the impact of moderately negative news.