The Impact of Limit Order Anonymity on Liquidity: Evidence from Paris, Tokyo and Korea