偵測盈餘管理

Detecting earnings management

張惇喬(Chang Tun-Chiao)

指導教授 ： 葉疏

國立臺灣大學/管理學院/會計學研究所/碩士(2005年)

https://doi.org/10.6342/NTU.2005.01736

摘要

目前臺灣學者進行盈餘管理方面研究主要採用方法為應計基礎模型，其中以Healy模型、DeAngelo模型、Jones模型、Modified Jones模型、與Industry模型最為廣泛被使用，但是仍未有一研究評估此五種模型於臺灣資本市場之適用情形。
本文採用DeChow與Sloan(1995)之研究方法，建立兩組可能導致型一錯誤的樣本以測試五個模型分別之精確度(Specification)，與另外兩組可能導致型二錯誤的樣本以測試五個模型的辨別力(Power)，描述、解釋所得之結果，並與Dechow與Sloan以美國資本市場資料進行實驗所得之結果互相比較。
結果發現：
一. 第一組樣本顯示，Jones與Modified Jones模型較不易產生型一錯誤；DeAngelo模型較易產生型一錯誤。
二. 第二組樣本顯示，五個模型皆未能產生與給定信賴水準接近的型一錯誤比例，表示五個模型精確度皆不佳。
三. 第三組樣本中，普遍而言，Modified Jones模型較能產生不偏的估計值，且辨別力曲線上升較快，是較佳的估計。
四. 第四組樣本中，五個模型皆導致型二錯誤，但是這可能是樣本過小，或是實施情況與應計基礎模型前提假設不符合的結果。

關鍵字

盈餘管理 ； 裁量性應計數 ； 模型選擇

並列摘要

When Researchers in Taiwan deal with earnings-management-related issues, they adopt accrual- based models frequently. Especially, the Healy, DeAngelo, Jones, Modified Jones, and Industry Model are most commonly considered among these accrual-based models. However, there is no study about the fitness of these five models when they are applied to the Taiwan capital market.
This article adopts Dechow and Sloan (1995)’s methodology. We build two samples which probably lead to the type I error to access the relative specification of these five models. Also, we prepare another two samples which could give rise to the type II error to test the relative power of each model. We describe and explain the empirical results and compare them with the original Dechow and Sloan’s experiment. The conclusions we draw are as follows:
1 In the first sample, the Jones and Modified Jones Model are more seldom to lead to type I error; DeAngelo Model has the lowest specification.
2 Sample (ii) shows that all these five models cannot produce the reject frequency close to the given confidence level. This result indicates that all these models perform not well.
3 Sample (iii) demonstrates that the Modified Jones Model produces more unbiased estimators than the other four and its power curve rises most steeply. We can infer that the Modified Jones Model estimate nondiscretionary accruals better.
4 In the last sample, all five models fail to reject the null hypothesis and make the type II error. This may be the result of the too small sample size or the lack of the assumptions accrual-based models require.

並列關鍵字

Earnings management ； Discretionary accruals ； Models selection

參考文獻

11.余智蓉，民國93年，遞延所得稅費用與盈餘管理關係之研究，國立台灣大學會計學研究所碩士論文。

1.Brown, S., and J. Warner. 1980. Measuring security price performance. Journal of Financial Economics 14: 3-31.

4.Dechow, P. M., and R. G. Sloan. 1991. Executive incentives and the horizon problem: An empirical investigation. Journal of Accounting and Economics 14: 51-89.

5.Dechow, P. M., and R. G. Sloan. 1994. Detecting earnings management. The Accounting Review 70:193-224.

6.Defond, M.L., and J. Jiambalvo, 1994. Debt covenant violations and manipulation of accruals. Journal of Accounting and Economics 17: 145-176.