Gan (2004) and Poon and Firth (2005) made use of econometrics to reach their conclusions. To achieve the purpose of this dissertation however, econometric techniques were not necessary - it only required the use of basic secondary data analysis (as will be explained later).
This study will be a quantitative study because some sort of scale is required to determine if there is a difference in the ratings given to the securities in the different sectors. Each letter of the Standard and Poor‟s rating scale described in Chapter 2 will be assigned a numerical value and the ratings given to the securities will be matched to that numerical value. Higher ratings (for example AAA) will be given a larger numerical value than lower ratings (for example C). The values assigned to the ratings given to securities in different sectors will be averaged to determine a
“sector” rating in order to determine the credit rating agency‟s behaviour towards different sectors. All calculations such as averages will be done in Microsoft Excel.
40 The research design will be a secondary data analysis. Existing data from reputable secondary sources is available to answer the research question. Standard and Poor‟s already has a list of defaulted securities for 2004, 2005, 2007 and 2008. The data includes information regarding the first rating of each security and when the security was rated for the first time. The data also includes the last rating each security had before default. There is thus no need to undertake a primary data search. Generally most papers regarding credit rating agencies make use of secondary data. Poon and Firth (2005) made use of secondary data obtained from Fitch and the Bankscope financial database. Gan (2004) collected data from the SEC database and created a dataset from this data.
The next section will detail exactly how the methods described above (quantitative data and secondary data) will be used to answer the research question. The data used will be described as well as the exact methodology that will be used.
3.3 Research methodology
3.3.1 Data
In order to answer the research question, data on all corporate publically rated defaulted securities will be collected from Standard and Poor‟s Annual Global Corporate Default Study and Ratings Transitions reports for 2004, 2005, 2007 and 2008. This period is chosen because the detailed Standard and Poor‟s data seems only to be publically available for those years. There is an Annual Global Corporate Default Study and Ratings Transitions for 2006, but it lacks the detail found in the other reports. Standard and Poor‟s also released the list of 2009 defaults, but the full report only became available at a very late stage of the writing of this paper.
The reason why only Standard and Poor‟s data is used is because they seem to be the only credit rating agency that publically releases their annual defaults in a very detailed form and it is very easy to obtain the data. The Standard and Poor‟s default data includes the name of defaulted company, what country the company is
41 registered in, the industry the company belongs to, the amount of debt on the company‟s books, the date of default, the rating the company had before default, the last time the company was rated before default, the company‟s very first rating and the date of that first rating.
It is important that the data is for securities that have already defaulted7 because that way one has what could be considered the entire rating history of a failed security.
Also as this study will examine why credit rating agencies have failed to warn investors of imminent crises, it is fitting to focus on defaulted securities as a default can be thought of as the ultimate crisis that can affect a security.
3.3.2 Methodology
The methodology must help provide results that could explain why credit rating agencies are reactive and not proactive to crises that affect the securities they rate.
Sector analysis will be used for this purpose.
The method used is one where the data collected on the different securities that will be examined is divided into sectors and then the credit rating agency‟s rating behaviour towards those groups of securities belonging to different sectors is noted and examined in light of events that affected various sectors in various years. For example Standard and Poor‟s downgraded a large number of transportation stocks in 2005 and 2006, in 2008 a large number of financial stocks were downgraded. In the years of the downgrades the sectors (transportation and financial) were in crisis.
This shows that sectors (that is a group of securities belonging to one sector) go into crisis and not a group of different securities belonging to different sectors. So in order to examine whether credit rating agencies can predict or merely react to crises the securities they rate should be divided into sectors, which this paper does.
7 Note that there is a difference between failed and defaulted banks. While this dissertation focuses on all securities, a failed bank, unlike a defaulted one, is a bank that defaulted and then later rescued by government influence.
42 Also the earlier instances where the activities of credit rating agencies were under suspicion were the dot-com bust, where the technology sector was the casualty.
There were also the accounting scandals of the 2000s and the latest crisis has to do with sub-prime mortgage loans. In each of these instances it is a group of securities belonging to one sector that experienced a problem.
Further support for the sector analysis can be found in the paper by Jiang, Koller and Williams (2009). In their paper they examine the financial performance of nine sectors8 during four different recessions.9 The results found different sectors enter and exit from downturns at different times. For example in the four recessions studied, the first sector to show strain was the consumer discretionary sector and in three of the downturns the last sector to show strain was the energy sector.
When it came to the magnitude of the strain that the sectors exhibited, this also varied from sector to sector. The sales and EBITA (earnings before interest, tax and amortisation) of the consumer discretionary sector, the materials sector, the energy sector and the industrial sectors showed the most severe drop and sectors such as healthcare and consumer staples showed relatively marginal declines in sales and EBITA. So credit rating agencies may for example downgrade securities in the consumer discretionary sector first thinking that those companies are bound to default, while failing to do the same for securities belonging to the consumer staples sector, which may also be heading for default.
The speeds of the declines and recoveries in sales and EBITA of the different sectors were also not the same. For example most sectors took over two years to recover to their peak EBITA levels after the 2001 recession, but industries such as the telecommunications sector never reached peak EBITA levels again.
The above suggests that sectors are unique and have characteristics unique to them which may cause credit rating agencies to treat each sector differently. Therefore dividing the data up into sectors is a valid method to use in determining credit rating
8 The nine sectors were consumer discretionary, consumer staples, energy, financial, health care, industrial, information technology, materials, telecommunication services and utilities.
9 The recessions were those of November 1973-March 1975, January 1980-November 1982, July 1990-March 1991 and March 2001-November 2001.
43 agency‟s behaviour during crises. Dividing the data into sectors focuses the analysis on the securities (which make up the different sectors) which were in crisis and so helps to control for the effect of variables that may affect all sectors (such as economic growth.) Once it is known which sectors were in crisis, the ability of credit rating agencies to predict crises can be examined.
The first step in the methodology explained above would be to show that credit rating agencies consider sectors (or put another way, take fundamentals into account) in their ratings. The data will be sorted into sectors and various questions will be answered about the collected data. The various questions will be answered to determine whether credit rating agencies rate different sectors differently; which if they did would suggest that they take fundamentals into account in their rating decisions. This method of looking at sectoral fundamentals is useful because as was described earlier in the literature review, the paper by Jiang et al (2009) finds that different fundamentals are at work in different sectors. If it can be seen that credit rating agencies do take fundamentals into account then this will help to ascertain the timing of when they took the fundamentals into account in order to determine whether credit rating agencies are reactive or proactive. So the questions below are the first phase in testing the hypothesis (the first phase is to test whether credit rating agencies take fundamentals into account). The sub-questions that will be answered in this first phase include:
1. Is the very first average rating received by the securities in one sector on average higher than the ratings received by securities in another sector/s?
2. Is the average next to last rating of defaulting securities in one sector on average higher than that of other sectors?
3. When comparing the number of days on average between the first time securities in a sector were rated and the average default date of securities in that sector, do securities in one particular sector take longer to default than other sectors?
4. Are the average number of days between the last date of rating of securities in a sector and the average default dates of those securities on average longer for a particular sector/s?
44 In order to answer the sub-questions, the Standard and Poor‟s rating data will have to be converted to numerical values (the reason for this is given shortly). The Standard and Poor‟s data collected gives ratings at three points in time. These points are: the very first rating the security obtained, the rating the security obtained before default and the date of default (the default is a rating in itself as well). Each of these ratings will be assigned a numerical value.
The reason why each of these ratings is given a numerical value is to convert the Standard and Poor‟s letter-based rating system into quantitative data in order to be able to calculate averages. For example a AAA rating will be worth 21 points and a C rating will be worth 1 point, the difference from one rating level to another will differ by one (Appendix A has a list of the points system). So the higher the rating, the higher the numerical value attached to it.
An important step in the whole methodology is dividing the securities up into different sectors. The Annual Global Corporate Default Study and Ratings Transitions data on the Standard and Poor‟s website gives the sector that each particular defaulted security belongs to. The sectors are:
Telecommunications
Utilities
Capital goods
Automotive
Transportation
Brokers
Healthcare
Aero/auto/CG/metal
Real estate
Metals and mining
Forest products and building materials
Chemicals, packaging and environmental services
Consumer products
Retail/restaurants
45
Energy and natural resources
Media and entertainment
Information technology
Finance companies
Customer service
Insurance
Leisure time/media
Oil and gas exploration and production
After showing that Standard and Poor‟s considers sectors (takes fundamentals into account), then the timing of when Standard and Poor‟s took those fundamentals into account should be determined to test whether the credit rating agency is reactive or proactive when it comes to rating securities that eventually default. To do this the sectors that Standard and Poor‟s examined at different points in time will be compared with news events at the time of the ratings change (for example the subprime crisis is categorised as such an event). This will be achieved by using articles collected from different sources to show what was happening in the world at the time.
This section will match up Standard and Poor‟s actions towards securities in the different sectors with those events that were occurring during the period of this study and were reported by the media. For example a table will be assembled which marks out when the various crises (for example the auto crisis and the subprime crisis) occurred and over these dates will be placed those sectors that Standard and Poor‟s was examining closely at the time (those are sectors that for example had the most securities downgraded in that period). With both the “popular financial events”
and sectors that were closely examined in one table, it will be determined whether Standard and Poor‟s was reactive or proactive when it came to events that affected a sector and the ratings of its securities.
46 3.4 Shortcomings of the methodology
The major shortcoming of this study is that it only concentrates on Standard and Poor‟s data, so it is very specific and probably cannot be strictly generalised to all credit rating agencies. The literature review though contains research that shows that credit rating agencies piggyback on each other. Güttler et al. (2006:753) in their paper examine the interaction between rating agencies. They conclude that there is in fact a fair amount of piggybacking between different rating agencies which rate the same securities. For example a downgrade or upgrade by one credit rating agency is followed by a more severe action in the same direction by another credit rating agency. Given this piggybacking, the results from this study could apply to other credit rating agencies, but caution must be applied when doing so.
The next chapter will discuss the results of the calculations done and the implications of the findings. It will also show whether in fact credit rating agencies are reactive or proactive when they rate securities in crisis.
47 Chapter 4