In order to construct the database, we used the Mergent Fixed Income Securities Database (FISD) as a starting point.19 Most researches (e.g. Reisel, 2004, and Billet et.al., 2007) that have done empirical work on the subject, uses this database. Thus, making this as our starting point, we can better compare our results to previous work and the situation in the US. However, the database we have built is what we would like to call an adjustment to the Norwegian market. We have included some additional categories not found in the FISD, which we believe is important in the Norwegian market. Generally speaking, most of them are a result of special conditions in some of the most important businesses in Norway such as oil and shipping. We describe these on how they are used in more detail in chapter 6.1 and 6.5.20 Some of the FISD categories of covenants are also modified slightly, in order to better describe the prevailing conditions in Norway. Other modifications is a result of the difficulty we had obtaining extensive information about the FISD categories. We also looked at the main categories that NTM uses in order to get a better understanding of the Norwegian distinctive characteristics21. Thus, some covenants are either broader or narrower defined in our database than in the FISD. The analysis in chapter 6 provides the reader with an understanding of all the covenants, but for replication purposes we will direct the reader to appendix 1 and the definitions of all the covenants included in our database. For readers unfamiliar with covenants in general, a better understanding can also be obtained by looking at the appendix.
The first step in constructing our database was to create a list of all the issues in the period and their respective ISIN numbers.22 To our surprise and to the best of our knowledge, no institution in Norway have created such a list (i.e. sorting the bonds by issue year and containing the ISIN numbers or other information), and we could only obtain unstructured “data files”. We created such a list by searching through the Stamdata database. Based on this list we collected bond
19
Interpretation of FISD covenants and what they include is based on “Corporate bond securities: Database Dictionary", by Mergent.
20
See appendix 1 for definitions 21
At the meeting 04/11/2008 22
International Securities Identification Number (ISIN) - an alphanumeric code that uniquely identifies a security. It consists of a two character country code, a nine digit number and one check digit.
Page | 26
specific information from the Stamdata by looking up each individual ISIN number. We typed inn all the available information about the different issues and downloaded the respective loan agreements whenever available.23 After completing this work, we sent NTM a list of the loan agreements not found at Stamdata. They were so helpful and returned the ones they were able to obtain.24 The private placements proved impossible to obtain. The exact results of this work can be seen in table 4.1.
After completing the list of issues and collecting the loan agreements, we started the work of reading through them. Based on the FISD classifications, the process described in the next chapter, we created a large Excel file containing all the bond information and the covenant categories. We used this to register the covenants in each bond contract. Upon completion of this work, we collected information on each issuing firms by using the AMADEUS database. We used this database because of the relatively high number of foreign issues in our data and because of the possibility of exporting information to Excel. In addition to the financial statements we collected data on the year of incorporation and other firm specifics. The bond issues were linked with the financial data as of the first of January the issuing year. 25 For some issuers we had problems finding the corresponding financial statement. In the cases this was a problem, we used the financial statements for the previous year or at the end of the bond issuing year. We did not find the relevant information for all the issues, reducing our second sample to and 323 issues (and 162 issuing firms), see table 4.2.
23
Appendix 2 shows how the bond specific information is presented at Stamdata 24
The remaining contracts were tried to obtain through www.oslobors.no, but they could, of course, only provide us with loan agreements for listed bonds.
25
The latest financial statement is our best estimate of both the bond issuers’ financial situation and investors’ knowledge about the company on the bond issuing date. We have disregarded that some firms in our sample publish quarterly reports.
Page | 27
Table 4.2: “Sample 2”
In other words, we operate with two data samples. Sample 1 includes all loan agreements and information about the bonds. In Sample 2 we have excluded the issues that we could not find company specific data for. In the descriptive analysis in the next two chapters we primarily use Sample 1, and the analysis can be viewed as representing all the issues in the period. We will not clearly distinguish between these two samples in the following parts, but bear in mind that if we only discuss covenants, Sample 1 is used. If our discussions both can be related to covenants and firm specific information, Sample 2 is used.
Finally, the macroeconomic factors used in the analysis are based on figures from SSB26 and Norges Bank. We have included all the relevant interest rates; the Sight deposit rate, different maturities of NIBOR, and several Government bond interest rates.