透過您的圖書館登入
IP:18.216.45.133
  • 學位論文

以混合羅吉特(MixedLogit)模式診斷公司資產減損之提列

Diagnosing Assets Write-off with Mixed Logit Model

指導教授 : 武季蔚
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


三十五號公報的施行將是我國財務報告的資產項目從歷史成本原則轉向公平價值重要的指標,使企業的財務報導更能真實反映營運狀況與經濟實質。然因缺乏客觀及可信賴的重置市場,評估資產的公平價值,常使公司管理當局應用賦予的裁量權,決定是否提列資產減損。因此財務報表中資產減損的提列與否及提列是否適當,值得進一步的檢視與診斷;同時若能將公司是否應提列資產減損進行估量,將可減少公司管理當局不當的利用裁量權,以符三十五號公報的精神。 過去關於是否提列資產減損的研究,多數係以Logit模式來估計,但受限於該模式的假設,其推論所得之參數結果為樣本的整體特性。而公司是否提列資產減損,決定於產業、公司營運、公司管理及經濟環境等複雜的因素,且各公司基於本身效用最大化,對資產減損提列的決策考量不盡相同。倘若診斷模式未能處理個體間的偏好變異,對分析結果之推論存在某種程度的偏誤。故本研究應用混合羅吉特(Mixed Logit)模式,探討複雜的資產減損提列。在Mixed Logit模式中,不僅可考慮個體間的異質性的存在,模式亦允許不同公司中不可觀察的隨機項之間具有相關性,可以有效克服以傳統Logit模式進行診斷的缺點,且Mixed Logit模式可以近似任何隨機效用模式,是相當具有彈性的模型。再者定期揭露的資產減損所累積成的追蹤資料(panel data)隱含了各公司決策的習性及偏好,藉此調整模式所推論的參數隨機分配以適用於特定的公司,增進預測診斷的準確性。本研究透過比較Logit模式與配合追蹤樣本的Mixed Logit模式的預測模型對於公司資產減損的預測績效,探討具有個別公司提列資產減損特性的參數模式,以期得到更合乎實際情況的資產減損提列診斷方法。實證結果顯示Mixed Logit模式的參數估計確能有效偵測資產報酬率、負債比率、盈餘平穩化及管理當局異動等因素的異質效果;配合追蹤資料對公司是否提列資產減損的檢測,其預測績效亦顯著優於傳統的Logit模式。

並列摘要


The implementation of Statement of Financial Accounting Standard No.35 in Taiwan is a critical landmark which turns the asset item in financial reports from historical cost principle to fair value and hopes to truly reflect the industrial operating conditions and economic essences. Because the shortage of objective and trusted replacement market to estimate fair value, the Standard No.35 authorized somehow extent discretionary to the manager for assets write-off decision. Thus, whether assets write-off are required and financial disclosures are adequate need to be examined. The strategy of assets write-off is determined complicatedly from industrial properties, corporation’s operation, management incentives and economic conditions. Most of previous researches attacked this problem with Logit model. The fixed parameters assumption of the model limits its modeling behavior which can’t conform to the various considerations between different corporations based on their own maximum utility. The ignorance of individual taste variation will lead the bias in inference. Therefore, in this study, Mixed Logit model is applied to deal with these. The model relaxes the fixed parameters assumption and allows for observed and unobserved heterogeneity which helps to solve the variation of each individual decision. Furthermore, the panel data accumulated from periodic assets write-off contains the potential information of individual habit and preference. With this information, the estimated parameters distributions can be adjusted to meet a particular company and improve the prediction accuracy. Therefore, the comparison of prediction performances between Logit model and Mixed Logit model with panel data is also studied in this work to explore the individual characteristics. The empirical results demonstrate that the Mixed Logit model certainly exhibits the taste variation on some critical factors at the decision of assets write-off and also show that panel data really disclose more information to improve accuracy of the diagnosis.

參考文獻


財團法人中華民國會計研究發展基金會,2004,財務會計準則公報第35號公報:資產減損之會計處理準則,2006年11月30日修訂。
謝宛庭與吳清在,2005,認列資產減損時點與金額之決定因素及其市場反應,中華會計學刊,6(1):59-95。
林有志與廖宜鋒,2006,提前採用資產減損公報之公司特性及盈餘管理動機,文大商管學報,第11卷第1期(6月):11-28。
陳慶隆,2007,選擇性新會計準則實施時點對策略性會計報導與資訊攸關性的影響,臺灣管理學刊,7(1):131-159。
武季蔚、陳慶隆、吳思音,2009,應用支向量機於公司資產減損之預測,2009 第十屆管理學域國際學術研討會,朝陽科技大學。

延伸閱讀