股利偏好與股票流動性共移性

Abstract

本文探討股利偏好是否會影響股票流動性共移性。既有研究指出，公司初次發放股利後，股票可能被投資人重新歸類為配息股票，進而提高其與配息股票群體之報酬共移性。本文延伸此一觀點，進一步檢驗股利初次發放是否亦會改變個股與配息股票及不配息股票群體之流動性連動關係。 本文以 1981 年至 2022 年美國上市公司之股利初次發放事件為研究對象，資料取自 Compustat 與 CRSP。首先參考 Hameed and Xie（2019）之研究設計，初步驗證股利初次發放與報酬共移性之關係，作為資料建構與事件設定之基準。接著，以 Amihud illiquidity 之日變化衡量個股非流動性變化，並採用納入前一日、當日與後一日投資組合非流動性變化之模型，以考量領先落後效果，估計個股非流動性變化對配息股票與不配息股票投資組合非流動性變化之敏感度。為降低公司基本面差異對結果之影響，亦透過傾向分數配對法建構控制組公司，並比較股利初次發放公司與控制組公司之事件前後變化。 實證結果顯示，股利初次發放公司於事件後與配息股票投資組合之流動性共移性上升，與不配息股票投資組合之流動性共移性下降；相較之下，控制組公司並未呈現相同型態。此結果支持股利初次發放可能使股票流動性變化更接近配息股票群體，並相對遠離不配息股票群體。補充分析顯示，2003 年股利稅改事件在方向上支持主要發現，而安慰劑事件與股票回購事件則降低了替代解釋的可能性。但 Citizens Utilities 事件與單變量模型之證據較弱，因此本文主要結論仍以全樣本中考量領先落後效果之模型結果為核心依據。 整體而言，本文結果支持股利偏好不僅可能影響股票報酬共移性，也可能反映於股票流動性共移性。本文補充股利政策、投資人分類與股票共移性之相關文獻，並指出投資人依股利特徵進行分類與交易的效果，可能進一步影響股票流動性之共同變動。

This thesis examines whether dividend preferences affect stock liquidity comovement. Prior studies suggest that after firms initiate dividends, their stocks may be reclassified by investors as dividend-paying stocks, leading to stronger return comovement with dividend-paying firms. Extending this argument, this thesis investigates whether dividend initiations also alter the liquidity comovement between individual stocks and dividend-paying or non-dividend-paying stock portfolios. Using dividend initiation events of U.S. listed firms from 1981 to 2022, this thesis employs data from Compustat and CRSP. The analysis first follows Hameed and Xie (2019) to preliminarily verify the relation between dividend initiations and return comovement as a benchmark. It then uses daily changes in Amihud illiquidity to measure changes in stock illiquidity and estimates liquidity comovement through a lead-lag specification. To mitigate differences in firm fundamentals, matched control firms are constructed using propensity score matching. The empirical results show that after dividend initiations, initiating firms exhibit higher liquidity comovement with dividend-paying stock portfolios and lower liquidity comovement with non-dividend-paying stock portfolios. In contrast, matched control firms do not show the same pattern. Additional analyses indicate that the results from the 2003 dividend tax cut are directionally consistent with the main findings, while placebo tests and stock repurchase events do not exhibit similar patterns. However, the evidence from the Citizens Utilities event and the univariate model is weaker; therefore, the main conclusion is primarily based on the full-sample lead-lag results. Overall, the findings suggest that dividend preferences may affect not only stock return comovement but also stock liquidity comovement. This thesis contributes to the literature on dividend policy, investor categorization, and stock comovement by showing that trading based on dividend characteristics may influence the commonality in stock liquidity