Imposing some form of short'sales constraints is a common response of regulators facing stock market downturns. The latest example is the global restrictions on short'selling in September/October 2008 due to the credit crunch. Although widely adopted, the impact of short'sales constraints on stock prices is still debated. Miller (1977) argues that if short' sales constraints are imposed, stock prices are set only by optimistic investors and these stocks are thus overpriced. With a rational expectation model, however, Diamond and Verrecchia (1987) argue that short'sales constraints do not necessarily bias prices upward. Investors rationally incorporate the constraints into prices. Bai, Chang and Wang (2006) take the rational expectation model one step further. They argue that risk averse investors ask a higher premium as compensation for bearing more risk due to less informative trades. In this paper, we shed light on the theories by providing a unique empirical strategy to study the effect of short'sales constraints on the prices.

We take advantage of a natural experiment in the Taiwan stock market. On 4th of September 1998, the Taiwanese Securities and Futures Bureau prohibited investors from short'selling a stock at a price below its close price of the previous trading day. It was a stability measure for the aftermath of the Asian financial crisis. The spirit of the rule is similar to the U.S. uptick rule since both try to prevent short'sellers from driving stock prices further downward. The major difference is that the rule in Taiwan is stricter and clearer. Unlike the multiple reference prices in the uptick rule, there is only one price (determined and fixed before the market opening) in Taiwan per day. Therefore, investors are aware of whether short'selling is banned or not during the trading hours. Consequently, this rule creates daily dynamics of short'sales constraints.