市場品質、流動性提供與價格叢聚---價格升降單位之影響研究

Abstract

本計畫是一個三年期計畫，主旨在探討市場微結構領域中的價格升降單位 (minimum price variation, tick size)對於市場品質和流動性提供的影響。理論研究 指出，最小價格升降單位使市場報價和成交價格皆呈現間斷(discrete)現象，直接 決定交易成本中的買賣價差，也影響市場參與者提供流動性的意願；多數實證研 究也發現降低價格升降單位可以顯著縮小買賣價差的幅度，然而卻也同時降低報 價深度，影響市場對流動性的提供。本研究探討台灣股票市場的價格升降單位， 對其影響從事全面且深入的分析，在三年的研究中，分別從三個不同的角度探討 價格升降單位改變對市場品質、流動性提供、以及價格叢聚(clustering)現象的衝 擊。 台灣市場為一全面自動化訂單驅動(order-driven)市場，且訂有多級距的價 格升降單位，有別於美國的市場結構，故可提供研究此議題的不同視角。本研究 首先採取和Ahn, Cao, and Choe (1996) and Bessembinder (2003)相同的觀點，著眼 於比較不同升降單位下的市場品質，以包括報價價差、有效價差、報價深度、成 交量、波動性等變數衡量市場品質的改變。比較的方式涵蓋2005 年升降單位制 度改變前後的市場品質，以及股票因本身價格漲跌而跨越某升降單位級距之前和 之後的市場品質。第二年的研究將深入分析流動性提供受升降單位改變的影響層 面。研究主題涵蓋成交量、報價深度、整體委託簿(limit-order book)所呈現的市 場深度、提供流動性的競爭與獲利、流動性的日內型態、改變升降單位的最終獲 利者等等，企圖從眾多面相中呈現市場提供流動性的意願與升降單位的關係。本 研究第三年預訂探討升降單位對價格叢聚現象的影響。若投資人符合價格解析假 說(price resolution hypothesis)所述，對於真實均衡價格之解析度有限，因而採用 較粗略的價格級距議價，使得市場報價和成交價皆呈現價格叢聚的現象，則降低 升降單位將使價格叢聚現象更為明顯。觀察升降單位降低前後的價格叢聚程度， 可以得知某升降單位是否過寬而限制市場的價格發現，此分析觀點亦有助於最佳 升降單位的擬定。

The minimum price variation (tick size) mandated by exchange restricts the quote prices to fall on a discrete grid and limits the prices at which a trade can take place. Theories suggest that the change in the tick size may significantly affect market quality. Empirical researches have found declined bid-ask spreads and reduced market depth after the exchanges adopt a smaller tick size. However, evidence is mixed regarding the effect of a finer price grid on volume, volatility, and other variables. This three-year project conducts a comprehensive study for the effect of tick size on the Taiwan stock market. The first-year study focuses on the impact of tick size changes on market quality, where market quality may be assessed by quoted bid-ask spread, effective bid-ask spread, quoted depth, trading volume, and intraday volatility. Similar to Ahn, Cao, and Choe (1996) and Bessembinder (2003), the investigation primarily centers on whether changes in tick sizes reduce costs of trading stocks. The study will perform several tests including the comparison of market quality for stocks in different tick size groups, the before-versus-after examination of stocks that pass through a particular tick size threshold, and a preversus post-event test of market quality surrounding the market-wide adjustment in tick sizes. In the second year, the research investigates liquidity provision under different tick sizes. The study uses a variety range of variables and methodologies to assess the willingness of liquidity provision, including the quoted depth, depths away from the inside quotes, the frequency of quote revision, and the intraday price variation/reversals that reflect the extent to which quoted depth is consumed by liquidity demanders (market orders). The study also reveals the extent to which a finer price grid enhances the competition of liquidity provision and therefore, reduces the profits of limit-order strategy. Finally, by assessing realized costs for traders who submit market orders in exceed of the depths of prevailing quotes, the study sheds light on whether tick size reduction benefits all liquidity providers/demanders or only some. In the last year, this proposal will link the minimum price variation to the study of price clustering. Price clustering indicates the tendency for asset prices (trade or quote) to occur more often at certain fractions or integers (e.g., 0 and 5) than the others. According to the price resolution hypothesis, price clustering may arise because traders can only resolve prices to coarser price grids than the minimum tick size. Tick size restricts the finest price resolution can be obtained. If tick size is binding (too large) for finer price discovery, there should be no significant price clustering across all available price grids. In that case, a smaller tick size would allow trades and quotes to take place on more possible price grids, which, in term, leads to greater price clustering. The study views the price clustering as the needs for finer price grids and better price discovery. The impacts of tick size on trade and quote price clustering are examined by comparing measures of price clustering across tick size categories as well as before and after the changes of tick-size rule.