Re-Examining the Default Forecasting Performance of Naive Merton Model

View Statistics Email Alert RSS Feed

Information

Details

Project title

Code/計畫編號

NSC99-2410-H019-007

Translated Name/計畫中文名

重新檢驗Naive Merton模型的違約預測績效

Project Coordinator/計畫主持人

Heng-Chih Chou

Funding Organization/主管機關

National Science and Technology Council

Department/Unit

Department of Shipping and Transportation Management

Website

https://www.grb.gov.tw/search/planDetail?id=2141190

Year

2010

Start date/計畫起

01-08-2010

Expected Completion/計畫迄

31-07-2011

Bugetid/研究經費

468千元

ResearchField/研究領域

管理科學

"Bharath and Shumway (2008，以下簡稱BS)提出Naïve Merton模型，任意給定模型參數方程式，以美國上市公司資料進行測試，實證結果證實Naïve Merton模型優於傳統的Merton (1974)模型。因此，BS (2008)的結論指出Merton模型在違約預測上的優異績效主要是來自於Merton模型的函數形式(functional form)，而非決定於Merton模型的參數估計值。然而，BS (2008)的結論有以下問題，值得進一步研究。首先，BS (2008)的研究是以 Ronn and Verma (1986，以下簡稱RV法)估計法所估計的參數作為比較基準，但是RV法僅為計算步驟並沒有統計估計的過程，缺乏統計意涵(statistical inference)，許多文獻皆已經指出RV法的缺失，並不符合Merton模型的假設；因此使用RV法所估計的Merton模型參數作為比較基準並不適當，亟需較佳的模型參數估計法(例如，Duan (1994, 2000)的資料轉換最大概似估計法)。其次，BS (2008)是與Merton模型比較，但是文獻已經指出Merton模型對於公司違約的設定不符合實際違約過程，若能採用障礙選擇權模型較能描述公司的實際違約過程；因此，若能將Naïve Merton 模型與障礙選擇權模型比較，應該有助於了解Naïve Merton模型的的違約預測績效。所以，本計畫擬重新檢驗Naïve Merton模型的違約預測績效，首先擬採用Duan (1994,2000)的資料轉換最大概似估計法推估模型參數，其次將同時採用Merton模型與障礙選擇權模型作為比較模型。此外，由於新興市場與成熟市場具有不同的資本結構，因此首先本計畫擬以1999年至2008年間台灣的上市與上櫃公司資料作為三種模型的測試對象，以進一步了解Naïve Merton模型在新興市場的違約預測績效；其次，本計畫並擬由台灣的上市與上櫃公司之資產價值波動度與權益波動度所呈現的關係，試著提出適合台灣市場的Naïve Merton模型。綜言之，本計畫具有以下特色：首先，除了RV估計法之外，本計畫也擬採用具有統計意涵的Duan (1994,2000)資料轉換最大概似法估計模型參數，並搭配Merton模型與障礙選擇權模型，可以提供Naïve Merton模型較公平合理的比較基礎。其次，BS (2008)僅以美國市場檢驗Naïve Merton模型的績效，本計畫則採用台灣的上市、上櫃公司為測試樣本，實證結果有助於了解Naïve Merton模型在新興市場的表現績效。第三、本計畫擬提出適合台灣市場的Naïve Merton模型，後續研究者可以在本計畫的基礎上，進一步分析新興市場公司權益與負債的關係，擬定更適當的Naïve Merton模型。""Bharath and Shumway (2008) argue that Merton model’s forecasting performance in default risk comes mainly from its functional form, not its parameters estimation. Bharath and Shumway (2008) propose Naïve Merton model, of which parameters are arbitrary determined, and demonstrate that its default forecasting performance is superior to Merton model, of which model’s parameters are estimated through the Ronn and Verma (1986) approach (RV hereafter). The RV approach, however, is only a calculation process without statistical inference, and many papers have point out the disadvantages of the RV approach. Among them, Duan (1994，2000) states that the RV approach in essence assumes asset volatility be constant, which contradicts the assumption of Merton model. Thus, the conclusion of Bharath and Shumway (2008) is in doubt, because Merton model parameters from the RV approach are not an adequate benchmark. Besides, one obvious weakness of Merton model is that default only occurs at maturity of the debt. As a result, a barrier option pricing framework is proposed to model the firm’s equity value, and the default risk can be estimated from the barrier option pricing model. Due to the weakness of the Merton’s model and estimation approach by RV approach, the study plans to reexamine the default forecasting performances of Naïve Merton model compared with a barrier option pricing framework and the data-transformation maximum likelihood estimation approach. The market data of the listed firms and OTC firms over the period of 1999 to 2008 are used to test the performance of the default risk models. Besides, based on the data of firm’s equity value volatility and the asset value volatility, the study tries to construct the adequate Naïve Merton model for Taiwan’s listed firms and also OTC firms. After all, the study is difference from previous research in three dimensions: firstly, beside the RV approach, Duan’s data-transformation MLE approach with Merton model and also barrier option pricing framework are also utilized so as to provide more reasonable benchmarks for Naïve Merton model. Secondly, in Bharath and Shumway (2008), only American firms’ data are used to examine the performance of Naïve Merton model; in order to examine the performance of Naïve Merton model in the emerging markets, both listed and OTC firms’ data will be collected in the study. Finally, an adequate Naïve Merton model for Taiwan market will be proposed, and based on the study, subsequent studies can further analyze the relation between a firm’s equity and liability to determine more reasonable Naïve Merton model."

Keyword(s)

Merton 模型
障礙選擇權
違約風險
新興市場
Merton model
Barrier option pricing framework
Default risk
Emerging markets

DSpace CRIS

Re-Examining the Default Forecasting Performance of Naive Merton Model

Details

Description