This thesis deals with various issues regarding determinants of US corporate credit spreads. These spreads are estimated as the difference between yields to maturity for corporate bonds and default-free instruments (Treasury bonds) of the same maturity. Corporate credit spreads are considered as measures of default risk. However, the premium required by investors for holding risky rather than risk-free bonds will incorporate a compensation not only for the default risk but also for other factors related to corporate bonds such as market liquidity or tax differential between corporate and Treasury bonds. In this study we firstly examine the relationship between bond ratings and credit spreads given that bond rating changes are expected to carry some informational value for debt investors. The findings indicate that bond ratings generally carry some informational value for corporate bond investors. The Granger causal relationship is more evident for negative watch lists and during periods of uncertainty in financial markets. In line with previous studies, our results suggest that changes in credit spreads are significantly related to interest rate levels, systematic risk factors (Fama and French) factors and equity returns.

This work empirically examines six structural models of the term structure of credit risk spreads: Merton (1974), Longstaff & Schwartz (1995) (with and without stochastic interest rates), Leland & Toft (1996), Collin-Dufresne & Goldstein (2001), and a constant elasticity of variance model. The conventional approach to testing structural models has involved the use of observable data to proxy the latent capital structure process, which may introduce additional specification error. This study extends Jones, Mason & Rosenfeld (1983) and Eom, Helwege & Huang (2004) by using implicit estimation of key model parameters resulting in an improved level of model fit. Unlike prior studies, the models are fitted from the observed dynamic term structure of firm-specific credit spreads, thereby providing a pure test of model specification. The models are implemented by adapting the method of Duffee (1999) to structural credit models, thereby treating the capital structure process is truly latent, and simultaneously enforcing cross-sectional and time-series model constraints. Quasi-maximum likelihood parameter estimates of the capital structure process are obtained via the extended Kalman filter applied to actual market trade prices on 32 firms and 200 bonds for the period 1994 to 2000. / We find that including an allowance for time-variation in the market liquidity premium improves model specification. A simple extension of the Merton (1974) model is found to have the greatest prediction accuracy, although all models performed with similar prediction errors. At between 28.8 to 34.4 percent, the root mean squared error of the credit spread prediction is comparable with reduced-form models. Unlike Eom, Helwege & Huang (2004) we do not find a wide dispersion in model prediction errors, as evidenced by an across model average mean absolute percentage error of 22 percent. However, in support of prior studies we find an overall tendency for slight underprediction, with the mean percentage prediction error of between -6.2 and -8.7 percent. Underprediction is greatest with short remaining bond tenor and low rating. Credit spread prediction errors across all models are non-normal, and fatter tailed than expected, with autocorrelation evident in their time series. / More complex models did not outperform the extended Merton (1974) model; in particular stochastic interest-rate and early default accompanied by an exogenous write-down rate appear to add little to model accuracy. However, the inclusion of solvency ratio mean-reversion in the Collin-Dufresne & Goldstein (2001) model results in the most realistic latent solvency dynamics as measured by its implied levels of asset volatility, default boundary level, and mean-reversion rate. The extended Merton (1974) is found to imply asset volatility levels that are too high on average when compared to observed firm equity volatility. / We find that the extended Merton (1974) and the Collin-Dufresne & Goldstein (2001) models account for approximately 43 percent of the credit spread on average. For BB rated trades, the explained proportion rises to 55 to 60 percent. For investment grade trades, our results suggest that the amount of the credit spread that is default related is approximately double the previous estimate of Huang & Huang (2003). / Finally, we find evidence that the prediction errors are related to market-wide factors exogenous to the models. The percentage prediction errors are positively related to the VIX and change in GDP, and negatively related to the Refcorp-Treasury spread.

Issuing environmentally aligned green bonds has become an increasingly popular way to raise capital for green investments during the last decade. This thesis explores potential pricing differences between green and conventional bonds, known as the green premium, on the Swedish secondary bond market. Prior green bond research is inconclusive regarding the direction, size and even existence of such a premium. By creating a sample of 50 matched pairs of green and conventional bonds, we show an average positive green premium of 10 bps on the Swedish market, indicating that Swedish green bonds trade at higher yields than their conventional counterparts. We also study whether the size of the green premium is affected by credit ratings and third-party green certification but find no evidence of such effects. Overall, the results from this thesis add to current green bond research by showing a positive green premium, but the lack of shown effects from credit ratings and green certification indicate that further study is needed to fully understand the pricing mechanisms of green bonds.

神經網路已經被成功地應用於解決各種分類及函數近似的問題，尤其因為神經網路是個萬能的近似器(universal approximator)，所以對於函數近似的問題效果更為顯著。以往對於此類問題雖然多數以線性的分析工具為主，但是實際上多數問題本質上是非線性的，所以對於非線性分析工具的需求其實是很大的。自1986年起，神經網路本身的運作一直被視為一個黑箱作業，難以判斷網路學習結果的合理性，更無法有效地幫助使用者增進其知識，因此提供一套合理及有效的神經網路分析方法是重要。 本文提出一套分析神網路系統的方法；利用線性規劃的技巧萃取及分析網路中的規則(rule)，而不需要對任何資料集做分析；進而利用統計無母數方法－符號檢定－歸納出網路中的知識。以債券評價為例，驗證此方法的可行性，實證結果亦顯示此方法所萃取出來的規則是合理的，且由這些萃取出的規則中，所歸納出來有關債券評價的知識多數是合理的。 / Neural networks have been successfully applied to solve a variety of application problems including classification and function approximation. They are especially useful for function approximation problems because they have been shown to be uni-versal approximators. In the past, for function approximation problems, they were mainly analyzed via tools of linear analyses. However, most of the function approxi-mation problems needed tools of nonlinear analyses in fact. Thus, there is the much demand for tools of nonlinear analyses. Since 1986, the neural network is considered a black box. It is hard to determine if the learning result of a neural network is rea-sonable, and the network can not effectively help users to develop the domain knowl-edge. Thus, it is important to supply a reasonable and effective analytic method of the neural network. Here, we propose an analytic method of the neural network. It can extract rules from the neural network and analyze them via the Linear Programming and does not depend on any data analysis. Then we can generalize domain knowledge from these rules via the sign test, a statistical non-parameter method. We take the bond-pricing as an instance to examine the feasibility of our proposed method. The result shows that these extracted rules are reasonable by our method and that these generalized domain knowledge from these rules is also reasonable.

This thesis contain three chapters. Default Risk in Corporate Bond Pricing. This chapter provides a model for how the corporate bond default risk influences the systematic risk and an empirical analysis of the systematic and idiosyncratic parts of U.S. corporate bond returns during 2001-2005. The average corporate bond beta is low and positive (0.06). Investment grade bonds have negative betas (between - 0.01 and -0.13) and non-investment grade bonds have positive betas (between 0.11 and 1.48), but both groups have similar within groups systematic risks. When controls for interest rate and liquidity risks are introduced there are still remaining default probabilities, implying that the default risk is in part systematic and in part idiosyncratic.   Returns to Defaulted Corporate Bonds.   In the second chapter short term excess returns in a sample of 279 defaulted US corporate bonds are tested for using multiple regression analysis. There are robust excess returns after controlling for market and liquidity risk. The expected recovery rate during 2001-2006 is estimated to be, on average, four percentage points lower the first month after default than the present value of the recovery rate after nine months. Capital Structure Choices.   The trade-off and pecking order theories are tested using both established tests from the literature and new tests. The main contributions of this chapter are the new tests of financing of operating net assets (for the pecking order theory), the mean reversion tests (for the trade-off theory) and the test of mean reversion and trends. These tests allow for extended conclusions on the validity of the pecking order versus the tradeoff theory. / <p>Diss. Stockholm : Handelshögskolan, 2008 Sammanfattning jämte 3 uppsatser</p>