Can Splits Create Market Liquidity? Theory and Evidence

This paper joins practitioners in predicting that firms split their stocks to move them into an optimal trading range, thereby creating market liquidity. Consistent with empirical evidence, our theory predicts that splits should follow a period of stock price increase and should have positive announcement effects. The driving force behind our result stems from the minimum price variation restriction self-imposed by organized exchanges. This restriction results in additional transaction costs which are time varying and mean reverting. In the model, discretionary liquidity traders strategically time their trades to periods with relatively low discreteness-related execution costs. The ensuing temporal aggregation creates market depth and allows the market maker to breakeven at lower commissions. These commissions can be lower than those charged in an otherwise identical economy with continuous prices. We show that the benefits of temporal aggregation are higher for less liquid stocks. The optimal effective tick size (defined as the ratio of the tick size to the stock price) is, therefore, decreasing in natural liquidity. Since the exchange sets the tick size, a firm can choose its effective tick by splitting (or reverse splitting) its stock. In an economy that values liquidity, firms will split (reverse split) their stocks to enhance liquidity. The empirical evidence is consistent with our model. To maintain a constant effective tick, firms need a constant nominal price. Indeed, despite positive inflation, positive real interest rates and positive risk premiums, the average (nominal) price of stocks traded on the NYSE during the last 70 years has remained almost constant. Further, our model predicts that Japanese firms should split less frequently than American firms due to a more rigid tick size regulation on the Tokyo Stock Exchange. The evidence presented is consistent with this prediction. In the U.S. 56% of stock distributions can be classified as splits (i.e., stock dividends larger than 20%) whereas in Japan splits constitute only 6.2% of the sample of stock distributions