使用馬可夫轉換矩陣分析股利政策之持續性

On the Stickiness of Dividend Policy: A Study Using Migration Matrix

本研究利用新方法分析公司股利政策之持續性，樣本是2005 至2012 年之臺灣上市櫃公司。討論主題包含：公司發放傾向、發放規模，及除息日至發放日之間隔長短。本研究主要使用馬可夫機率矩陣分析工具，依照除息日至發放日間隔天數，區格五種狀態：不發放、除息日後22 日內發放、23 至28 日發放、29 至36 日發放與超過36日發放。然後，本研究再將發放事件分成四組發放規模，以分析股利政策之改變過程。研究發現，無論就統計顯著性或經濟顯著性，公司之股利政策呈現持續性與平滑性。因此，公司股利政策之持續性比傳統文獻討之內容更明顯。研究結果對股利政策持續性之原因具有啟發性解釋，本研究主張發放股利之行為已經成為一種規範。

This article studies the stickiness and smoothness of dividend policy from a new perspective, through examining the dividend policy of Taiwanese firms over 2005-2012. The issues we investigate include: the propensity to paying dividends, the magnitude of dividends and the periods between the ex-dividend date and the payment date. We exploit the Markov migrating probability matrix to explore these issues. This study defines five states for the cash dividend-paying policy: paying none, paying within 22 days, paying between 23 and 28 days, paying between 29 and 36 days, and paying after more than 36 days, with respect to the various periods between the ex-dividend date and the payment date. After further dividing the sample of dividend payers into four groups according to the quartiles of dividend magnitude, we find firms reveal a high degree of stickiness and smoothness in terms of economic significance. Hence, the preponderance of smooth dividend policy is more prevalent than we thought of. This study sheds light on the reason why a dividends policy is sticky or smooth. We suggest that it has become a social norm to pay and receive dividends.