PYXIDA Institutional Repository
and Digital Library
Collections :

Title :An empirical analysis of bond ratings in shipping industry
Creator :Σπυροπούλου, Αθηνά
Contributor :Καβουσανός, Εμμανουήλ (Επιβλέπων καθηγητής)
Athens University of Economics and Business, Department of Accounting and Finance (Degree granting institution)
Type :Text
Extent :208p.
Language :en
Abstract :We have employed the Cohort and Hazard approach for generating transition matrices for shipping bonds issued by companies around the world. Both approaches calculate transition matrices but the cohort approach does not make full use of the available data, instead it gives us an estimation about how the probabilities will change in the short term horizon. On the other hand, the implementation of the Hazard approach has calculated transition probabilities by taking into account all the information as it is counting all the changes in the transition ratings that occur within a year and not the start and end rating as cohort does. The analysis was divided into two main parts. In the first part we calculated transition matrices by separating the observations according to where the companies have their headquarters and by employing eight, twenty two and sixty four classes rating schemes. The classification of the companies into four different geographical areas (North America, Europe, Asia & Pacific and Scandinavia) was made according to Lloyd’s list. The different rating class schemes were used in order to examine the transitions from one rating category to the other in detail. The sixty four class scheme was employed in order to incorporate in our study all the available information as S&P, except from the general rating symbols, readjusts the rating by giving a positive, stable or negative watch list. Thus, in the sixty four class scheme we calculated extra probabilities for the changes in the watchlist status. In the second part of the study, we calculated transition matrices according to which stock exchange the bonds negotiate (three categories: 1) New York Stock Exchange, 2) Luxembourg Stock Exchange and 3) Other Stock Exchanges). This separation was performed by using the classification of Data stream database. The purpose of this segmentation was to examine whether there are differences inthe transition probabilities and especially Probabilities of Defaults (PD’s) that can been attributed to the Stock exchange. Finally we have calculated confidence intervals for the Hazard approach and contrast the estimated matrix for shipping bonds with a matrix calculated by Moodys with all the bonds issued between 1920-1996.This empirical study has given us some very interesting results. Some general results that are valid for all the transition matrices but obviously also for the shipping bonds matrixes calculated are the following: On diagonal entries are the highest, meaning that the system is relatively stable and PD’s for the best rating classes are zero but in the Hazard approach arenon-zero even for the best grades. Also, the probabilities of downgrades are bigger for the lower rating classes and the opposite is true for the upgrades. Additionally, we have revealed characteristics that are specific for the shipping bonds. The Cohort and Hazard approaches give us different estimates and we conclude that in the eight classes scheme Cohort underestimates the probabilities of downgrades and overestimates the probabilities of sustaining the same rating grade. By applying the Cohort approach to the 22-class scheme we observe that the probabilities of sustaining the same rating in the majority of the classes for all the samples, with exception of Scandinavia, are lower than those calculated with the Hazardmethod. We must note that in the Hazard approach the highest probabilities of upgrade for the non-investment bonds are for North America sample, then follows Scandinavia, Asia &Pacific. Another interesting result stemming from the 64 class is that probabilities for sustaining the same rating are quite high for the North America sample and that Asia &Pacific bonds demonstrate a tendency to upgrade rather than remain in their initial rating.Moving to the non-investment grades we conclude that bonds with a positive Watchlist tend to be upgraded or to maintain their initial rating rather than be downgraded, meaning that they move in the direction of their Watchlist status, bonds with negative Watchlist tend to downgrade or to default.By applying the hazard approach for the 8-class scheme using stock exchange criteria it is observed that the probabilities of sustaining the same rating are higher for NYSE than Luxembourg and “Other”. Moreover, the probabilities of upgrade are smaller than those calculated with the Cohort approach and the PD’s are higher for the NYSE sample than“Other” sample. The results when applying the stock exchange criteria for the 22 class scheme non-investment grades Luxembourg bonds seem on average to be more stable than the NYSE bonds and “Other” bonds as they have grater probabilities of maintain the same rating and being upgraded. Furthermore, we notice that for rating grades in which we observe downgrades NYSE bonds have smaller probabilities of a downgrade than “Other” and Luxembourg bonds and that upgrades do not follow a specific pattern and PD’s are higher initial bonds than in “Other” bonds.Also, by analyzing the confidence intervals for PD’s we observe that these are wider for the North America bonds than Europe bonds. Finally, by comparing the total sample transition matrix for shipping bonds with a matrix calculated for bonds issued from different industries,we conclude that non investment shipping bonds tend to be riskier than other similar bonds as they have grater probabilities of downgrades and default and their risk is significantly higher for the 6th and 7th rating grade.
Subject :Shipping bonds
Transition matrices
Date :16-07-2009
Licence :

File: spyropoulou_2009.pdf

Type: application/pdf