The data to be used in this research will be obtained from the database of World Bank, at the website of http://databank.worldbank.org/ddp/home.do?Step=3&id=4. Accordingly, World Bank has a comprehensive database on the finance and economic related indicators for most of the countries in the world. As articulated in the web portal of World Bank, the database of World Bank, namely, Databank has more than one thousand indicators for 209 countries around the globe, from year 1960 to 2010.
Having access to the Databank, annual interest rates for the countries will be obtained and tabulated into Excel spreadsheet for further analysis. The countries in South East Asia that will be analyzed include: Indonesia, Singapore, Malaysia, Philippines, Thailand and Brunei Darussalam. The interest rates obtained from Databank is essentially the real interest rates in percentage for the respective countries. As stated clearly in http://data.worldbank.org/indicator/FR.INR.RINR, real interest rates are defined by World Bank as the lending interest rate adjusted for inflation as measured by the GDP deflator. Similarly, the market capitalization of the stock markets of these countries will also be obtained from Databank. All of the market capitalizations for these countries are converted to US Dollar for ease of comparisons. Indeed, US Dollar is used as the currency in recording the market capitalization of these countries in Databank. The data of market capitalization of listed companies in the respective countries can be obtained from http://data.worldbank.org/indicator/CM.MKT.LCAP.CD. Accordingly, World Bank defines market capitalization of listed companies (in current US Dollar) as the share prices times the number of shares outstanding. Then, the respective listed domestic companies are actually the domestically incorporated companies listed on the respective country’s stock exchanges at the end of the year. According to the data provided in the Databank, the following listed companies are excluded from the calculation of market capitalization of the respective countries: investment companies, mutual funds, and other collective investment vehicles.
Nonetheless, it should be acknowledged that not all data concerning economic indicators or variables for the respective countries can be readily obtained from Databank of World Bank. In that case, the research presented here will only analyze the relationships between interest rates to stock returns in the specific countries, subjecting to data availability. For example, it is found that market capitalization for Brunei Darussalam is missing from Databank throughout the years. As such, Brunei Darussalam will be dropped from the analysis due to absence of market capitalization data for the country. Other the other hand, data is available for Indonesia, Singapore, Malaysia, Philippines, and Thailand from year 1989 to year 2009. This means for each of the country, a total of 20 set of observation points between interest rates and market capitalization is available – and will be used in the subsequent analysis process in the following chapter. For the references of readers, the annual data of interest rates, changes in interest rates from year to year, market capitalization and changes of market capitalization from year to year (in percentage form) will be presented in Appendix (at the end of this report).
As discussed above, the relationships of interest rates to stock returns will be investigated. A total of five countries located in South East Asia will be included in the research design. These five countries are Indonesia, Singapore, Malaysia, Philippines and Thailand. Overall, this means that for each of the countries, the respective interest rates and market capitalization for each year under the research period will be obtained. Following this, a total of 10 research variables will be included in this research design. Apart from that, following the work of procedures and research design of the previous literature, the growth rates, or more specifically, the changes of these research variables from year to year will be computed. This is important because by investigating the changes of the independent variables and its implication to the changes in dependent variables are more economically meaningful. Ultimately, in the context of stock returns, scholars are interested in understanding how a particular changes in certain variables to the other variables. In this context, the research objective is to investigate if the changes of interest rates have any correlation relationships to changes in market capitalization. As such, the changes of interest rates and market capitalization each year, for the respective countries will be computed and employed in the research design. For clarification purpose, the shorthand as well as the description of the research variables used in this study will be reported in Table 3.1 below.
In this section, the relevant statistical models to be used in investigating the relationships between changes in interest rates to stock market returns will be discussed. In order to understand if there are any statistical relationships between stock market returns to the changes of interest rates, Pearson correlation between these two variables will be computed. With this, the degree of association between the two variables can be investigated. In order to analyze association relationships between the data, SPSS can be used (to generate Pearson correlation coefficients). In the following sections, the theoretical concepts and backgrounds of Pearson Correlation Coefficient will be articulated.
Technically speaking, one of the most basic and yet highly important statistical attributes of two variables in the context of forecasting is the degree of association between two different variables. As discussed by DeLurgio (1998), some of the statistical measures on the degree of association between two or more variables include the following: covariance coefficient, correlation coefficient, auto-covariance coefficient, and autocorrelation coefficient. These variables are useful, as predictability (i.e., using predicting variables to estimate the predicted variables in the context of forecasting) is possible when the values of the two variables are found to move together. In other words, variables that have statistical association relationship may offer predictive contents for interested researchers. In contrast, if the two variables have no degree of association between each other, then one variable cannot be used to predict the other variable. Both of the variables are considered statistically independence.
A common statistical measure to investigate degree of association between two or more variables in the context of finance and economics is Pearson Correlation Coefficient. For example, Ahmed and Cardinale (2005) had performed a study to investigate relationships between stock returns to macroeconomic variables using the concept of Pearson correlation coefficient. From statistical perspective, Pearson Correlation Coefficient is useful as the concept measure the proportion of the covariance of two variables to the product of their standard deviation (DeLurgio, 1998). To explain, correlation coefficient can be interpreted as the average standard deviation change in one of the variable associated with one standard deviation change in another variable. The statistical concept can be expressed mathematically as follow:
In simple English, correlation coefficient inform researcher about how a particular variables behave in relative to the other variable (over the time period being studied). A correlation coefficient value close to 1 indicates existence of positive correlation between two variables. For example, when variable A and Variable B have correlation coefficient of 1, it suggest that as variable A goes up by x unit, variable B will be going up by x unit as well. In contrast, a correlation of -1 indicates negative correlation. In this situation, if both variable A and Variable B have correlation coefficient of -1, it suggests that when variable A zig; variable B zag by the similar amount. Both variable A and variable B move in an opposite direction. Then, a correlation of zero suggests that there is no correlation between the variables. The movement of variable A simply is not related (even from a purely statistical observation perspective) to variable B.
Due to time and resource limitations, the research design employed in this study has several limitations. Firstly, the study can only serve as the preliminary study on if there are indeed any statistical relationships between interest rates to stock returns. The existence of statistical significant correlation coefficient between variables may not imply that one of the variables can be used to predict the other variables in the future. As the financial market is dynamic, and situations or the context in the countries being investigated in this research may change accordingly, the historical relationships may no longer stay valid in the future. Secondly, correlation does not imply causation. In order to understand or investigate if the predicting variable is indeed causing changes in the other variable, it is researcher imperative to have solid understandings on the economic rationale behind the cause and effect linking the two variables.
Besides, the research method, namely, application of correlation coefficient measures in the study of stock returns is perhaps too simplistic. Stock market is a highly complicated invention of human beings, and the inability of even the many professional to beat the market suggest that the drivers of stock returns can be highly dynamic, intricate and complex. As such, it is highly possible that correlation coefficient may not be sufficient to draw conclusive conclusion on the relationships between interest rates and stock returns in the countries being studied. Other statistical concepts may be required to investigate the changes of interest rates to changes of stock returns in the past.
Then, it should be acknowledged that the research design presented in this study explicitly ignore the possibility of structural changes in the period under investigation. This can affect the conclusion of the statistical results generated from this study. There are reasons to believe that in different era, for example, during booming and recessionary period, the relationships between interest rates and stock returns may differ. However, structural change issues are ignored in this study due to time constraints.
Apart from that, the use of annual data may not be sufficient to truly understand the relationships of interest rates to stock returns. There are possibilities that when monthly data is used, different relationships may be discovered. As financial market change in a very fast manner, it is possible that relationships between interest rates to stock returns may be more easily observed in shorter time period. Nonetheless, monthly data is not employed due to data availability issues.