李書行Li, Shu-Hsing臺灣大學:會計學研究所李貴富Li, Kuei-FuKuei-FuLi2010-05-052018-06-292010-05-052018-06-292009U0001-2607200914243000http://ntur.lib.ntu.edu.tw//handle/246246/180013本論文共分為二部分。第一部份，吾人探討盈餘管理（或盈餘品質）是否會影響股利政策及股利的資訊內涵。具體而言，本文探討當期股利變動程度與未來年度未預期盈餘間之關連性。雖然此議題已於文獻中廣泛探討及進行實證分析之，然而，仍並未有足夠實證證據支持之關連性。 文認為經理人利用裁決性應計項目增加私人利益時，股利政策可能成為誤導投資人的工具之一。本文發現，相較於高盈餘品質的經理人，低盈餘品質的經理人增加股利的機率較低，但這些經理人當面臨時股票選擇權將逾期時，其增發股利的機率會提高。或當經理人即將卸任時，相較於年紀較輕的經理人，年紀較長的經理亦會提高增發股利的機率。顯示出低盈餘品質的公司，當經理人的私有利益可能減少時或欲增加私人利益時，股利政策成為誤導投資人判斷績效的工具之一。此之外，本文亦發現在高盈餘品質公司，其股利變動程度與未來年度未預期盈餘之變動成正向關係，此實證結果與股利訊號假說一致。不過，於低盈餘品質公司，其股利變動程度與未來年度未預期盈餘之變動間關連性，未達統計上顯著水準。整體而言，上述結果意含在低盈餘品質公司，股利訊號假說並不是經理人決定股利政策時的第一考量。第二部分，本文探討生產力是否為導致不同種類的盈餘管理行為。吾人採用生產理論中所提出的創新方法：資料包絡分析法及勞動生產力成長因素分解法，以衡量生產力成長的持續性，進而估計未來維持高生產力的潛力。同時，本文亦參考Tucker and Zarowin(2006)，構建損益平穩化的衡量指標。證結果顯示生產力與損益平穩化程度指標呈正向關係。此發現意涵當公司生產力高時，經理人利用裁決性應計項目構建出平穩化盈餘數列，但當生產力低時，經理人卻投機性運用裁決性應計項目。吾人進一步分析，相較於低生產力公司，高生產力的盈餘及裁決性應計項目價值攸關性較高。here are two parts in this doctoral dissertation. In the first part, I examine whether the behavior of earnings management (or earnings quality) affects the dividend policy and information content of dividends. Specifically, I investigate the relationship between dividend changes and unexpected changes in future earnings. This issue has been the subject of numerous empirical studies. However, these studies have failed to document a significant relationship. allow for the fact that dividends would be as tools to mislead investors when managers could benefit from the income-increasing discretionary accruals. I find that firms with low earnings quality are less (more) likely to raise (cut) dividends, but further investigate and find the managers whose options will expire are more likely to raise dividends than managers whose options won’t in the firms with low earnings quality. The result also indicates the older managers are more likely to raise dividends than younger managers for CEOs facing turnover in low earnings quality firms.Otherwise, I also find that in high earnings quality firms, there is a significant positive relation between the magnitude of dividend change and unexpected future earnings changes, consistent with the signaling theory of dividends. However, in low earnings quality firms, the relation between dividend increases and unexpected future earnings changes is insignificant. Overall, this finding implies that signaling effect is not first order of dividend policy in low earnings quality firms. n the second part, this paper investigates whether magnitude of productivity induces different behaviors of earnings management. I use an innovative method proposed by in production theory to measure the persistence of growth to measure the potential to be high productivity in the future by applying Data envelopment Analysis (DEA) method and decomposing the labor productivity growth. Meanwhile, I follow Tucker and Zarowin (2006) to construct an income-smoothing measure. he data reveal the relations between the magnitude of income smoothing and productivity is positive. This result implies firms with high productivity (or high earnings persistence) manage discretionary accruals to make a smooth earnings series, but firms with low productivity (or low earnings persistence) use discretionary accruals opportunistically. I further investigate and find earnings and discretionary accruals in high productivity firms have more value relevance than ones in low productivity firms目 錄 (Table of Contents)試委員會審定書 i 辭 ii 要 iiibstract v. Does Earnings Quality Matter in Dividend Signaling Hypothesis? 1.1 Introduction 1.2 Literature Review and Hypotheses Development 6.2.1 Free Cash Flows Hypothesis 6.2.2 Signaling Hypothesis 6.2.2.1 Price-dividend Type 8.2.2.2 Earnings-dividend Type 9.2.3 Hypotheses Development 10.3 Sample Selection and Research Design 15.3.1 Sample Selection 15.3.2 Research Design: Earnings Management 16.3.3 Hypothesis 1: Earnings Management and Dividend Change 18.3.4 Hypothesis 2: Multivariate Analysis of Future Earnings Change 20.4 Empirical Results 22.4.1 Part 1- Hypothesis 1 22.4.1.1 Descriptive Statistics 22.4.1.2 Regression Results 23.4.2 Part 2- Hypothesis 2 24.4.2.1 Descriptive Statistics 24.4.2.2 Regression Results 26.4.3 Robustness Checks 27.4.3.1 Linearity Mean Reverting 27.4.3.2 Future Profitability 28.4.3.3 Alternative Deflation 29.4.4.4 Alternative Proxy for Dividend Change 31.5 Summary and Conclusion 31 Assessing the Relationship between Earnings Management and Productivity: An Empirical Study 34.1 Introduction 34.2 Literature and Hypothesis Development 39.2.1 Earnings Management 39.2.2 The Functions of Discretionary Accruals 39.2.3 Income Smoothing 40.2.4 Hypotheses Development 41.3 Methodology and Sample Selection 42.3.1 Data Envelopment Analysis 43.3.2 Decomposition of Productivity Growth 46.3.3 Earnings Management Estimation Model 51.3.4 Multivariate Earnings Management Analysis 53.3.5 Value Relevance of Discretionary Accruals 55.3.6 Sample Selection 56.4 Empirical Results 57.4.1 Descriptive Statistics 57.4.2 Regression Analysis 59.4.3 Further Test 60.5 Summary and Conclusion 61eference 80 目 錄 (List of Figures)igure 1: Illustration of Decomposition in Kumar and Russell (2002) 48 目 錄 (List of Tables)able 1: Sample Selection and Sample Composition 63able 2: Summary Statistics 64able 3: The Univariate Results by Earnings Management 65able 4: Logistic Regression of Decision to Increase Dividends or Decrease Dividends 66able 5: Logistic Regression of Incentives to Increase Dividends or Decrease Dividends for low earnings quality firms: Hypothesis 1b 67able 6: Logistic Regression of Incentives to Increase Dividends or Decrease Dividends for low earnings quality firms: Hypothesis 1c 68able 7: Summary Statistics 69able 8: The Distribution of Dividend Changes by Earnings Quality 70able 9: Regression of Future Earnings Changes on Current Dividend Changes 71able 10: Robustness Checks 72able 11: Descriptive Characteristic 73able 12: Efficiency Indexes for 33 Industries, 1990 and 2000 74able 13: Percentage Change of Tripartite Decomposition Indexes, 1990 and 2000 75able 14: Regression Analyses of Income Smoothing Based on Accrual Model 76n the Change in Productivity 76able 15: Regression Analyses of Market Value Based on Book Value and Components of Net Income 77able 16: Testing H1 by Using Alternative Sample Period 78able 17: Testing H2a and H2b by Using Alternative Sample Period 79application/pdf4503702 bytesapplication/pdfen-US股利政策股利訊號盈餘品質損益平穩化勞動生產力成長因素分解價值攸關性Dividends policyDividend signalingEarnings qualityIncome smoothingDecomposition of labor productivity growthValue relevancehttp://ntur.lib.ntu.edu.tw/bitstream/246246/180013/1/ntu-98-D92722004-1.pdf