We aim to make two contributions to the literature on the effects of transaction costs on financial price volatility. First, by using a research design with three ingredients (a common set of companies simultaneously listed on two stock exchanges; binding capital controls; different timing of changes in transaction costs), we obtain a control group that has identical corporate fundamentals as the treatment group and is therefore far cleaner than any in the existing literature. We apply the research design to Chinese stocks that are cross-listed in Hong Kong and Mainland. Second, we entertain the possibility that a given transaction cost can have different effects in immature and mature markets. In an immature market where trading is dominated by retail investors with little knowledge of accounting and finance, a Tobin tax should have the best chance of generating its intended effect. In a more mature market, higher transaction costs may also discourage sophisticated investors, hence impeding timely incorporation of fundamental information into prices. We find a significantly negative relation in the Chinese market, on average, between stamp duty increase and price volatility. However, this average effect masks some important heterogeneity. In particular, when institutional investors have become a significant part of traders' pool, we find an opposite effect. This suggests that a Tobin tax may work in an immature market but can backfire in a more developed market.