Erasmus university rotterdam

CHAPTER 3. RESEARCH METHODOLOGY

Data selection

This study uses data collected from the Thomson Financial DataStream database. Data is collected for 739 companies from five emerging markets for the period January 2001 to September 2009 based on the availability and length of data sets maintained by DataStream. After analysis of the companies, 516 companies were used and 223 companies were excluded. The countries that are included are Egypt, Israel, Morocco and Turkey. US dollar denominated total returns are collected for the companies and the indices of the emerging markets. This is done in order to maintain uniformity of results. The return data used in this study are sampled at a weekly frequency. To calculate returns, DataStream adjusts weekly prices for dividends, splits and share issues. The lengths of the data samples are not uniform. The data on the returns for Egypt, Israel, Morocco and Turkey covers the period from January 12, 2001 to September 25, 2009. The dataset comes from the Datastream database. In order to avoid influence of survivorship bias, non survival shares are included in the tests as well.

Data for the market indices was also collected for the five emerging market indices using the DataStream database. The MSCI, S&P and FTSE market indices were collected. Analysis of the indices showed that the MSCI indices were most useful for inclusion in the analysis, due to the availability and length of the data. The weekly US dollar denominated returns on the Morgan Stanley Commodity Index (MSCI) World index are also collected from the DataStream Database for the period January 12, 2001 to September 11, 2009. The MSCI World index is used as a proxy for the returns on the world market portfolio.

When investigating the CAPM a proxy for the risk free rate is needed. Researchers often use the 10 year US Treasury-bill as a proxy for the risk free rate. This data is collected for each week from the website of the Federal Reserve Bank² for the period January 12, 2001 to September 25, 2009.

Since no data on the local risk free rate was available, Credit Default Swap (CDS) data was used. For Egypt, Israel, Morocco and Turkey as well as for the US, daily data on the mid rate spreads of 10 year Senior Credit Default Swaps were collected for the period April 20, 2006 to September 25, 2009. The country risk premium was calculated as the spread between the emerging market rate and the US rate. The global financial crisis influenced the country risk premiums, in some cases the risk premiums almost doubled. Therefore the CDS spreads that are used in the analysis are separately calculated for three periods, the total period (April 2006 to September 2009), before the onset of the financial crisis (April 2006 to August 2008) and the period during the financial crisis (September 2008 to September 2009).

To assess the robustness of the results, the sample of the historical returns are divided into three periods. The first period start January 2001 and ends September 2009 (total period). The second period starts January 2001 and ends August 2008 (before crisis) and the third period starts September 2008 and ends September 2009 (during crisis).