4. Method
4.3 Data Collection and Source Criticism
4.3.1 Data Collection
This study consists of a secondary analysis, as the data is collected by other institutions and for other purposes than for this specific study. There are several
advantages of using this type of data, particularly if the quality of the data can be secured (Bryman & Bell, 2005). Firstly, it is a cost- and time effective approach to gathering data. Secondly, the quality of the data is often very high, meaning that sampling has been rigorous, the organisations responsible are accredited and reliable, the researchers responsible are highly experienced, and the data has gone through several quality checks and similar procedures. However, it is important to remember the absence of control over the quality checking, and hence to the extent possible, some caution should be taken when using and analysing the data (ibid).
To enable the examination of this study’s problem statement and answer the chosen research question, quantitative data in the form of daily stock prices for all companies in the population, i.e. all listed Swedish companies within the event- and estimation windows between May 10th 2006 and November 21st 2013, as well as daily index prices for OMX Nordic Stockholm, are retrieved from Thomson Reuters Datastream.
Thomson Reuters Datastream is also used for extraction of data on market capital, total assets, turnover, and number of employees for 2013. The chosen database is a widely used and acknowledged source, and is hence considered a reliable and authentic mean for data gathering. An alternative source could for example be Bloomberg, but as they have similar reputation as Thomson Reuters Datastream, and both are widely used, the choice between these would not make a difference for the study and its results (ThomsonReuters, 2015:2; Bloomberg, 2015). As for the market index, used for example to calculate normal and abnormal returns, OMX Nordic Stockholm Price Index (OMXSPI) is chosen. OMXSPI includes all stocks that are listed on OMX Nordic Exchange Stockholm and is cleared from dividends (NASDAQ, 2015:1-2). An alternative could be to use OMXS 30, i.e. the index of the thirty most sold stocks on the Stockholm Stock Exchange. As can be seen in Figure 4.2 below, these two indexes correlate highly, which indicates that the choice of index would most likely not have any significant effects on the results of the study.
However, as mentioned, OMXSPI includes all listed stocks, and hence the use of the first index is justified.
Figure 4.2. Correlation between OMX StockholmPI and OMX Stockholm30Index 1995-2015 (DN Ekonomi, 2015).
Further, organisational documents are used to approach the event of CSR-rankings in the Swedish stock market. The chosen ranking report is Folksam’s CSR ranking report, as described in section 4.1. As mentioned, the report is based on the UN Global Compact guidelines, and would hence be applicable on a global basis, which simplifies the analysis of applicability of this study’s results on other local markets.
4.3.2 Data Criticism
To the greatest possible extent, only well-cited articles from well-reputed journals are used as a theoretical base. In this study, well cited implies both the number of citations, but especially the quality of them. As for the journals used, they are considered being of high quality and high reputation, which is also shown in rankings made by e.g. the Financial Times. The books chosen have either been used during courses at Copenhagen Business School, or are other well-cited books by famous researchers. The chosen articles are often also cross-cited, which implies a connection and a clear platform for existing research. However, the previous literature has had different views on, and results from, looking at the relationship between CSR and financial performance, and therefore the arguments and results from both sides have been highlighted. Throughout the thesis, a critical approach is taken, and the intention and goal is to be as objective as possible.
As for the event study methodology, theoretical and methodological books, as well as previous event study based research articles, are used to assess how the best possible event study is executed for the purpose of this study’s research question. As the majority of the articles found focus on other types of events, and hence have different
research questions than this study, it is rather the choices of models within the event studies that are considered interesting to examine to ensure a stable ground for the methodology. Since the methods used in these various event studies are similar and follow similar patterns based on renowned theory and methodology, it is considered reliable to base this study’s models on the same.
4.3.3 Data Adjustment
During the execution of the event study, some data adjustment is made to ensure accurate results. When estimating the normal returns of the stocks, the returns included in the estimation window described below are used in a linear regression to estimate what the normal return would be in the event window. These are then compared to the actual returns during the event window to estimate the potential abnormal return during the period, as further described in section 4.4.3 Normal returns.
To ensure that the normal return estimation is as unbiased and accurate as possible, bigger events such as mergers and acquisitions are adjusted for. Hence, companies that have had these types of events within the estimation period will be removed from the analysis. A list of these adjustments can be found in Appendix 9.2.
4.3.4 Analysis of Missing/Excluded Data
As mentioned in the previous section, the aim is to generate results that are as unbiased as possible. Hence, any events that may affect the results of the study considerably need to be removed in order to avoid any bias. This is especially relevant during the period of estimation of normal returns, as these numbers are an important component when finding abnormal returns.
In total, 4 companies have been excluded as a result of having a substantial event during the period that may have a significant influence on the study results. However, any potential name change during the period is not considered to be a large event and is hence not excluded, as this is not considered to have any significant effect on the value of the company.
In addition, all companies that do not have complete data during the whole estimation period are excluded, as these companies do not meet the criteria of being listed 126 days prior to the event. This can for example occur when companies go from listed to unlisted and back to listed. More on the criteria is presented in section 4.4.2 below. In addition, the companies that are no longer listed on the day of the event or any of the event days are also excluded as that indicate that no abnormal returns can be calculated, which would lead to an error. As Folksam ranks all companies that are listed on the 31st of May for each year, there is an inherent risk that some of the companies merge or acquire other companies, or become acquired, stop their business, or leave the stock exchange during the estimation or event period and hence do not have complete data for the period. The number of companies excluded for this reason amounts to 4.