巫和懋臺灣大學：國際企業學研究所朱戌文Chu, Hsu-WenHsu-WenChu2007-11-282018-06-292007-11-282018-06-292006http://ntur.lib.ntu.edu.tw//handle/246246/60188本文為考量資訊不完全下的結構式模型，認為市場上的投資人難以直接真實地觀察到公司的資產價值，頂多只能由公司的財務報表來推斷。資訊不完全是由於投資人所得到資產價值的資訊內容，為 之前的資產價值，資產價值的揭露時間與實際發生的時間之間有時間差 ，意即目前公司實際的資產價值為 ，但投資人只得到 的資訊以及僅知道公司在此段時間差中尚未違約，甚至有時公司經理人會選擇延後財務報表的發佈。然而，若模型只考慮時間差，則當現在實際的資產價值低於可觀察到的資產價值時，模擬出的信用利差卻和實證文獻呈現相反的結果。 因此本篇文章將選擇性揭露納入資訊不完全的模型，認為選擇性揭露亦為資訊不完全的來源，將公司揭露資訊的決策在債券評價模型中內生化。本文考量了三種類型的揭露成本，探討在不同類型的揭露成本的考量下，公司依據資訊品質產生不同的揭露門檻，依據此揭露門檻決定是否要揭露訊息，此訊息進而影響到債券價格。本文發現，當揭露成本與資訊精密度成正向關係時，本模型模擬出的信用利差與實證上的文獻相符合。此外，本文以公司選擇性揭露後發佈的資訊來模擬債券價格，將公司違約事件的發生視為是很難準確預測到的停止點（stopping time），雖然是結構式模型的延伸，也可推導出違約密度，和縮減式模型有一致的結果。 實證上常以報表資訊的延後作為選擇性揭露的結果，因此本文最後探討這兩年中GM和AIG延後會計報表的事件以及資本市場的反應，來觀察選擇性揭露的影響。In our structural credit model based on incomplete information, investors cannot observe a firm’s true asset value directly. Investors instead draw inference from the available accounting data. The incomplete information arises from a fixed time lag between the current value of assets and the asset value known to outsiders. In other words, when the true asset value is , the outsiders know only the value and that the asset value has not hit zero between time and time . Moreover, sometimes the manager chooses to withhold accounting data so that information asymmetry becomes worse. However, when the current value is substantially lower than lagged report and if we only consider time lag in the imperfect information model, the term structure of credit spreads is opposite to the one emerging from empirical reference. On this account, we introduce discretionary disclosure into the imperfect information model by making disclosure decision become an endogenous choice on the part of the firm with different types of disclosure costs. Then we show that the lack of precise knowledge of a firm’s value process and the awareness of disclosure threshold can lead to different predictions on the shape of the term structure of credit spreads. Our structural model is more consistent with the empirical result that a higher transparency is rationally associated with lower credit spreads for most of the term structure. Especially when the form of disclosure cost is quadratic and increasing in information precision, it seems to solve most conflicts between the imperfect information credit model and the empirical statistics. Default intensity, the critical feature of reduced-form models, is derived from this structural approach in models where bond investors have incomplete information about asset values of firms with discretionary disclosure rights. Last, I use GM and AIG cases to reveal the importance of delaying accounting reports to the capital market.CHAPTER 1 INTRODUCTION 1 1.1 MOTIVATION 1 1.2 OBJECTIVES 5 1.3 FRAMEWORK 6 CHAPTER 2 REVIEW OF THE LITERATURE 7 2.1 RESEARCH OF STRUCTURAL CREDIT MODELS 8 2.1.1 Classical Approach 8 2.1.2 First-Passage Approach 9 2.1.3 Stochastic Interest Rate 9 2.1.4 Endogenous Default 10 2.2 RESEARCH OF REDUCED-FORM CREDIT MODELS 12 2.2.1 Default Intensity 12 2.2.2 Rating Transition Intensity 12 2.2.3 Affine Intensity Models 13 2.3 EMPIRICAL RESEARCH AND INCOMPLETE INFORMATION CREDIT MODELS 15 2.3.1 Empirical Research 15 2.3.2 Incomplete Information Credit Model 16 2.4 RESEARCH OF DISCRETIONARY DISCLOSURE 18 CHAPTER 3 DEFAULT BARRIER AND DISCLOSURE THRESHOLD 19 3.1 MODEL OF ENDOGENOUS OPTIMAL DEFAULT BARRIER 20 3.1.1 Model Assumptions 20 3.1.2 Optimal Default Barrier 24 3.1.3 Numerical Illustration of Base Case 27 3.2 MODEL OF DISCLOSURE THRESHOLD 29 3.2.1 Model Assumptions and Descriptions of the Market 30 3.2.2 A Description of Disclosure Threshold Equilibrium 34 3.2.3 Disclosure Threshold Solution and Static Comparisons Under Different Types of Disclosure Costs 36 CHAPTER 4 BOND PRICING WITH DEFAULT BARRIER AND 52 DISCLOSURE THRESHOLD 52 4.1 DESCRIPTION OF IMPERFECTION INFORMATION 53 4.2 CREDIT SPREADS WITH DISCLOSURE 57 4.2.1 Conditional Asset Density 57 4.2.2 Credit Spreads 61 4.3 CREDIT SPREADS WITH NON-DISCLOSURE 64 4.3.1 Conditional Asset Density 64 4.3.2 Credit Spreads 66 CHAPTER 5 DEFAULT INTENSITY AND COMPARISON WITH EMPIRICAL DATA 77 5.1 DEFAULT INTENSITY 77 5.1.1 Default Intensity With Disclosure 78 5.1.2 Default Intensity With Non-disclosure 79 5.2 EMPIRICAL DATA 82 5.2.1 Empirical Data of GM 82 5.2.2 Empirical Data of AIG 86 CHAPTER 6 CONCLUSION AND THOUGHTS FOR FURTHER STUDY 90 6.1 CONCLUSIONS 90 6.2 THOUGHTS FOR FURTHER STUDY 91 APPENDIX 1 92 APPENDIX 2 93 APPENDIX 3 94 REFERENCE 95 LIST OF TABLE 【TABLE 3-1】 THE THRESHOLD LEVEL OF DISCLOSURE VARYING INFORMATION NOISE LEVEL WHEN 39 【TABLE 3-2】 THE THRESHOLD LEVEL OF DISCLOSURE VARYING INFORMATION NOISE LEVEL WHEN 45 【TABLE 3-3】 THE THRESHOLD LEVEL OF DISCLOSURE VARYING INFORMATION NOISE LEVEL WHEN 49 【TABLE 4-1】 THE THRESHOLD LEVEL OF DISCLOSURE VARYING INFORMATION NOISE LEVEL WHEN 74 【TABLE 4-2】 PARAMETER SETS AND DISCLOSURE THRESHOLD LEVELS THAT GENERATE THE RESULT FIT IN WITH EMPIRICAL INFERENCE. 75 LIST OF FIGURE 【FIGURE 1-1】 TERM STRUCTURE OF CREDIT SPREADS WITH DIFFERENT ACCOUNTING PRECISIONS USING THE MODEL IN DUFFIE AND LANDO（2001）WHEN CURRENT REPORTED ASSET VALUE IS LOWER THAN PREVIOUS FIRM VALUE. 4 【FIGURE 3-1】 TIME LINE FOR MAJOR ACTORS’ ACTIONS IN EACH PERIOD 19 【FIGURE 3-2】 DISCLOSURE THRESHOLD VARYING INFORMATION NOISE LEVEL WHEN THE FORM OF DISCLOSE COST IS CONSTANT 40 【FIGURE 3-3】 DISCLOSURE THRESHOLD VARYING CONSTANT DISCLOSURE COST LEVEL WHEN THE FORM OF DISCLOSE COST IS CONSTANT 42 【FIGURE 3-4】 LOG-EXPECTED MARKET PRICE WITH VARYING ACTUAL ASSET LEVEL （WITHHOLD DISCLOSURE LEVEL）WHEN THE FORM OF DISCLOSURE COST IS A QUADRATIC FUNCTION OF THE REALIZATION 46 【FIGURE 3-5】 DISCLOSURE THRESHOLD VARYING INFORMATION NOISE LEVEL WHEN THE FORM OF DISCLOSURE COST IS A QUADRATIC FUNCTION OF THE REALIZATION 47 【FIGURE 3-6】 DISCLOSURE THRESHOLD VARYING INFORMATION NOISE LEVEL WHEN THE FORM OF DISCLOSURE COST IS QUADRATIC AND INCREASING IN INFORMATION PRECISION , =0.05 50 【FIGURE 4-1】CONDITIONAL DENSITIES OF FIRM VALUE WITH DIFFERENT ACCOUNTING PRECISIONS WHEN PREVIOUS FIRM VALUE AND CURRENT REPORTED ASSET VALUE ARE 90.7. 59 【FIGURE 4-2】 DEFAULT PROBABILITIES WITH DIFFERENT ACCOUNTING PRECISIONS WHEN PREVIOUS FIRM VALUE AND CURRENT REPORTED ASSET VALUE ARE 90.7. 60 【FIGURE 4-3】 TERM STRUCTURE OF CREDIT SPREADS WITH DIFFERENT ACCOUNTING PRECISIONS WHEN PREVIOUS FIRM VALUE AND CURRENT REPORTED ASSET VALUE ARE 90.7. 62 【FIGURE 4-4】 CONDITIONAL DENSITIES OF FIRM VALUE AND TERM STRUCTURE OF CREDIT SPREADS WITH DIFFERENT ACCOUNTING PRECISIONS AND DISCLOSE COST IS CONSTANT WHEN CURRENT REPORTED ASSET VALUE IS 86.7, LOWER THAN PREVIOUS FIRM VALUE 90.7. 68 【FIGURE 4-5】 CONDITIONAL DENSITIES OF FIRM VALUE AND TERM STRUCTURE OF CREDIT SPREADS WITH DIFFERENT ACCOUNTING PRECISIONS AND DISCLOSE COST IS WHEN CURRENT REPORTED ASSET VALUE IS 86.7, LOWER THAN PREVIOUS FIRM VALUE 90.7. 69 【FIGURE 4-6】 CONDITIONAL DENSITIES OF FIRM VALUE AND TERM STRUCTURE OF CREDIT SPREADS WITH DIFFERENT ACCOUNTING PRECISIONS AND DISCLOSE COST IS WHEN CURRENT REPORTED ASSET VALUE IS 86.7, LOWER THAN PREVIOUS FIRM VALUE 90.7. 71 【FIGURE 4-7】 CONDITIONAL DENSITIES OF FIRM VALUE AND TERM STRUCTURE OF CREDIT SPREADS WITH DIFFERENT ACCOUNTING PRECISIONS AND DISCLOSE COST IS WHEN CURRENT REPORTED ASSET VALUE IS 86.7, LOWER THAN PREVIOUS FIRM VALUE 90.7. 72 【FIGURE 4-8】 CONDITIONAL DENSITIES OF FIRM VALUE AND TERM STRUCTURE OF CREDIT SPREADS WITH DIFFERENT ACCOUNTING PRECISIONS AND DISCLOSE COST IS WHEN CURRENT REPORTED ASSET VALUE IS 86.7, LOWER THAN PREVIOUS FIRM VALUE 90.7. 73 【FIGURE 4-9】 DISCLOSURE THRESHOLD VARYING INFORMATION NOISE LEVEL WHEN THE FORM OF DISCLOSURE COST IS QUADRATIC AND INCREASING IN INFORMATION PRECISION , =0.06 75 【FIGURE 4-10】CONDITIONAL DENSITIES OF FIRM VALUE AND TERM STRUCTURE OF CREDIT SPREADS WITH DIFFERENT ACCOUNTING PRECISIONS AND DISCLOSE COST IS WHEN CURRENT REPORTED ASSET VALUE IS 86.7, LOWER THAN PREVIOUS FIRM VALUE 90.7. 76 【FIGURE 5-1】 DEFAULT INTENSITIES WITH DIFFERENT ACCOUNTING PRECISIONS WHEN PREVIOUS FIRM VALUE IS 90.7 AND INVESTORS HAVE CURRENT REPORTED ASSET VALUE. 80 【FIGURE 5-2】 DEFAULT INTENSITIES WITH DIFFERENT ACCOUNTING PRECISIONS WHEN PREVIOUS FIRM VALUE IS 90.7 AND INVESTORS DON’T HAVE CURRENT REPORTED ASSET VALUE 81 【FIGURE 5-3】 STOCK PRICE OF GENERAL MOTORS FROM 2005/ 10 ~ 2006/ 05 84 【FIGURE 5-4】 THE CREDIT SPREADS OF AN GENERAL MOTORS BOND WITH MATURITY DATE 1 MAY 2008. THE COUPON RATE IS 6.375%. 84 【FIGURE 5-5】 THE CREDIT SPREADS OF AN GENERAL MOTORS BOND WITH MATURITY DATE 15 APRIL 2016. THE COUPON RATE IS 7.7%. 85 【FIGURE 5-6】 THE CREDIT SPREADS OF AN GENERAL MOTORS BOND WITH MATURITY DATE 15 MARCH 2036. THE COUPON RATE WAS 7.75%. 85 【FIGURE 5-7】 STOCK PRICE OF AMERICAN INTERNATIONAL GROUP FROM 2005/ 01 ~ 2006/ 05 88 【FIGURE 5-8】 1 YEAR CREDIT DEFAULT SWAP SPREADS OF AMERICAN INTERNATIONAL GROUP 88 【FIGURE 5-9】 3 YEAR CREDIT DEFAULT SWAP SPREADS OF AMERICAN INTERNATIONAL GROUP 89 【FIGURE 5-10】 5 YEAR CREDIT DEFAULT SWAP SPREADS OF AMERICAN INTERNATIONAL GROUP 89749270 bytesapplication/pdfen-US選擇性揭露揭露成本專利資訊Discretionary disclosureDisclosure costProprietary information[SDGs]SDG16thesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/60188/1/ntu-95-R93724015-1.pdf