DATABASE AND RESEARCH METHODOLOGY

Transcription

1 CHAPTER III DATABASE AND RESEARCH METHODOLOGY This chapter explains the methodology adopted by the researcher for the collection and analysis of data. The tools of data collection, sources of data and scientific methods of data analysis have been discussed in detail. Hypotheses have also been framed as per the objectives of the study. 3.1 STATEMENT AND OVERVIEW OF PURPOSE The main aim of this study is to understand the dynamics of performance of Foreign Multinational Companies operating in India during the Post Liberalization Period along with making a comparative analysis with that of domestic companies. In addition, an attempt has been made to analyze the policy related issues governing foreign MNCs in India and to suggest future measures to attract FDI in India will benefit both the host country (India) as well as MNCs operating herein. 3.2 UNIVERSE AND SAMPLE Universe of the Study The universe of the study includes both foreign multinational companies 1 as well as domestic Indian companies listed in Bombay Stock Exchange in the year The financial database relating to these companies has been compiled from the electronic database PROWESS developed by the Centre for Monitoring of Indian Economy (CMIE). PROWESS is the largest database providing financial information for large and medium domestic as well as foreign companies operating in India. The database contains detailed information for over 27,122 companies ( These comprise all companies traded on India's major stock exchanges and several 1 The word, corporation and company have been used as synonyms for the purpose of the present study throughout. 63

2 others including the central public sector enterprises (CMIE, 2012). The Prowess database is built from Annual Reports, quarterly financial statements, Stock Exchange feeds and other reliable sources. CMIE normalizes this database in order to enable inter-company as well as inter-temporal comparisons for the purposes of researchers and analysts, which further enhances the usefulness of this database. Following sectors are covered by PROWESS: Organized industrial sector; Banking; and Organized financial and other service sectors in India. PROWESS proves to be a comprehensive database for research as the companies covered by it account for: 65 per cent of all corporate tax assesses in India; Over 95 per cent of excise duty collected by the Government of India (CMIE, 2012); and Companies covered by Prowess companies account for more than half of India's external trade i.e. for about 62 per cent of India's exports and nearly 82 per cent of India's imports. The database is built on a sound understanding of accounting standards and disclosures practices currently prevailing in India. The data covered by the prowess includes most of the financial information required to be reported by incorporated companies in their annual reports, such as Profit and Loss account and a Balance Sheet along with information contained in the schedules and annexures (CMIE, 2012) Sample and Sampling Unit As the study aims to attain its objectives through firms operating at micro level, therefore, the most suitable unit of the analysis is company/firm/corporation as drawn from the review of literature. The sample of the study is based on both 64

3 foreign multinational and Indian domestic companies. Though these companies were operating in various sectors such as chemical sector, food sector, machinery sector, metal sector, non-metallic sector, textile sector, transport sector and miscellaneous sector etc. yet, it was decided to limit the present study only three sectors namely Chemical sector, Food sector, Machinery sector. This is so because, these sectors represent 60 per cent of data available on Indian manufacturing, whereas the data available for other sectors was not sufficient to form a reliable comparative base. Secondly, an attempt has been made to study those companies whose financial information for at least 5 consecutive years was available, whereas majority of the companies of the left-out sectors failed to meet this criteria. The sampled companies include all companies operating in these three sectors of Indian manufacturing. 3.3 PERIOD OF STUDY As per the objectives of the study, secondary data was considered to be the more suitable to draw the results. As the study relates to post-liberalization period, therefore, a comprehensive period of 18 years starting from 1992 to 2009 was chosen to attain the objectives was selected as the base year, as the reforms were initiated in the year 1991 in India. The period of the study finds its justification as it covers the post reforms period when large scale policy reforms had been introduced in the industrial and foreign trade policy. 3.4 NATURE OF DATA In order to study the dynamics related to performance of the companies, the data pertains to financial indicators comprising of both foreign multinational as well as domestic companies operating in India on the basis of Profit and Loss Accounts and Balance Sheets in the form of annual reports of the sampled companies. 65

4 3.5 COLLECTION OF SECONDARY DATA Since 1960, the research on multinational corporations has gained enormous worldwide attention. In past few decades, the liberalization and economic reform process initiated by many developing countries has further broadened the area of research in this field. As a result, a comprehensive literature covering various aspects of multinational corporations is available in various books, journals, reports, dissertations and articles etc. Therefore, to collect this diverse literature both offline as well as online sources have been accessed. Among these online sources of data, databases such as CiteSeer, Elsevier, Emerald, Inflibnet, Ingenta, JSTOR, National Bureau of Economic Research (NBER), Proquest, RePEc (Research Papers in Economics), Science Direct, Scopus, Social Science and Research Network(SSRN), Springer, Taylor and Francis and Wiley Interscience are a few to mention. Among offline data sources, many national as well as regional libraries such as National Council of Applied Economic Research; New Delhi, Indian Council for Social Science Research; New Delhi, Jawaharlal National University; New Delhi, IIM Udaipur, Punjabi University; Patiala, Guru Nanak Dev University; Amritsar and Panjab University; Chandigarh have been consulted from time to time for the collection of secondary data. Data Arrangement Used of pooled data has been made for the purpose of the present study. The data set, considers observations related to n individuals (companies) over a t period of time. In simple words, each individual carries observations ranging from first year to t years. Hence, the total number of observations in pooled data set should be nt. The data arrangement has been presented in table 3.1. The illustrated data shown hereunder shows various variables for two set of companies i.e. foreign and Indian for a 5 year period for 5 companies i.e. I T C Ltd.; V S T Industries Ltd.; Nestle India Ltd.; Glaxo-Smithkline Healthcare Ltd.; and Agro Tech Foods Ltd. 66

5 TABLE 3.1 DATA ARRANGEMENT OF POOLED DATA S. No. COMPANY YEAR STATUS INDUSTRY SALES PROFIT 1. I T C Ltd Foreign Machinery V S T Industries Ltd Foreign Chemical Nestle India Ltd Foreign Food Glaxosmithkline Healthcare Ltd Indian Food Agro Tech Foods Ltd Indian Food I T C Ltd Foreign Machinery V S T Industries Ltd Foreign Chemical Nestle India Ltd Foreign Food Glaxosmithkline Healthcare Ltd Indian Food Agro Tech Foods Ltd Indian Food I T C Ltd Foreign Machinery V S T Industries Ltd Foreign Chemical Nestle India Ltd Foreign Food Glaxosmithkline Healthcare Ltd Indian Food Agro Tech Foods Ltd Indian Food I T C Ltd Foreign Machinery V S T Industries Ltd Foreign Chemical Nestle India Ltd Foreign Food Glaxosmithkline Healthcare Ltd Indian Food Agro Tech Foods Ltd Indian Food I T C Ltd Foreign Machinery V S T Industries Ltd Foreign Chemical Nestle India Ltd Foreign Food Glaxosmithkline Healthcare Ltd Indian Food Agro Tech Foods Ltd Indian Food

6 Following the same pattern, the data for all the Indian and multinational corporations has been arranged. The data set was framed in excel file format which was further used for determining statistical output in E-views, a statistical software used for statistical, forecasting, and modeling tools through an innovative, easy-to-use object-oriented interface. Furthermore, Statistical Package for Social Sciences (SPSS 16.0) has also been used to analyse the data. 3.6 HYPOTHSES OF THE STUDY To attain the objectives of the study, the following hypotheses have been designed: H 1 = There is no significant difference in mean of profitability performance of Foreign Multinational Corporations and Domestic companies operating in India in Chemical sector. H 2 = There exists a significant difference between mean of profitability performance of Foreign Multinational Corporations and Domestic companies operating in India in Chemical sector. H 3 = There is no significant difference in mean of profitability performance of Foreign Multinational Corporations and Domestic companies operating in India in Food sector. H 4 = There exists a significant difference between mean of profitability performance of Foreign Multinational Corporations and Domestic companies operating in India in Food sector. H 5 = There is no significant difference in mean of profitability performance of Foreign Multinational Corporations and Domestic companies operating in India in Machinery sector. H 6 = There exists a significant difference between mean of profitability performance of Foreign Multinational Corporations and Domestic companies operating in India in Machinery sector. 68

7 H 7 = There is no difference in mean of export intensity of Foreign Multinational Corporations and Domestic companies operating in India in Chemical sector. H 8 = There exists a difference between mean of export intensity of Foreign Multinational Corporations and Domestic companies operating in India in Chemical sector. H 9 = There is no difference in mean of export intensity of Foreign Multinational Corporations and Domestic companies operating in India in Food sector. H 10 = There exists a difference between mean of export intensity of Foreign Multinational Corporations and Domestic companies operating in India in Food sector. H 11 = There is no difference in mean of export intensity of Foreign Multinational Corporations and Domestic companies operating in India in Machinery sector. H 12 = There exists a difference between mean of export intensity of Foreign Multinational Corporations and Domestic companies operating in India in Machinery sector. H 13 = There is no difference in mean of Research and Development intensity of Foreign Multinational Corporations and Domestic companies operating in India in Chemical sector. H 14 = There exists a difference between mean of Research and Development intensity of Foreign Multinational Corporations and Domestic companies operating in India in Chemical sector. H 15 = There is no difference in mean of Research and Development intensity of Foreign Multinational Corporations and Domestic companies operating in India in Food sector. 69

8 H 16 = There exists a difference mean of Research and Development intensity of Foreign Multinational Corporations and Domestic companies operating in India in Food sector. H 17 = There is no difference in mean of Research and Development intensity of Foreign Multinational Corporations and Domestic companies operating in India in Machinery sector. H 18 = There exists a difference mean of Research and Development intensity of Foreign Multinational Corporations and Domestic companies operating in India in Machinery sector. 3.7 TECHNIQUES APPLIED FOR EMPIRICAL ANALYSIS The data collected from different sources have been analyzed with various statistical, econometric as well as accounting tools and techniques. These techniques have been analyzed in the following direction: Student s t-test The application of t-test is based on the assumption that the samples have been randomly drawn and are normally distributed over the population, with unknown variances 2. This assumption states that the variables to be considered should be of such a nature whose values should change randomly. Furthermore, the value of one variable should be independent of the value of other variables. T-test further assumes random sampling without any selection bias. Therefore, if any research, knowingly selects some samples with properties that best suits the requirements of the study and compares these values with other samples, then the conclusions drawn from non-random sampling will neither be reliable nor generalized. However, according to the type of data considered in the study, the researcher has to select the appropriate method of t-test. The methods of t-test can generally be studied under Independent one-sample t-test, Independent two-sample t- test, and Dependent t-test for paired samples. 2 If population variances are known then z-test with σ2 can be determined and there is no need of determining variances. 70

9 In statistical terminology, t-test is a statistical hypothesis test in which the test statistic follows a Student's t-distribution if the null hypothesis is true (Wikipedia, 2009). The Student's t-test is used for determining the statistical significance of the difference between two sample means, and for confidence intervals for the difference between two population means. In probability and statistics, Student's t-distribution (t-distribution) is a probability distribution that arises in the problem of estimating the mean of a normally distributed population. The Student's t-distribution is a special case of the generalized hyperbolic distribution. The general formation of data under t-distribution can be depicted with the help of following diagram. The above diagram shows the normal distribution of the data. Like other probability distributions, the total area under the curve of t-distribution is equal to one. As the number of degrees of freedom increases, the shape of the t-distribution converges to that of the standard normal distribution. In the above diagram the student's-t distribution has been depicted with the help of blue hyperbola whereas, the normal distribution has been depicted through red hyperbola. 71

10 The researchers frequently use one-sample t-test or two sample t-test. Where, one-sample t-test determines if the mean of a normally distributed population has a value specified in a null hypothesis or the population mean is same as the hypothesized value or not, two sample t-test attempts to test the null hypothesis that the means of two normally distributed populations are equal. As the data in the present study was related to two group of companies i.e. Indian and multinational, therefore, independent two-sample t- test was found to be most suitable Independent Samples t test The independent samples t-test is used to compare the statistical significance of a possible difference between the means of two groups on some independent variable i.e. t- test helps to determine if the samples have been drawn from populations having different mean values or not. Hence, under t-test application the two samples are desired to be independent of each other. Therefore, observations considered in group one will not be linked with the observations considered in group two and vice-a-versa. The suitability of t-test is based on the satisfaction of following assumptions that: Each of the two populations should follow a normal distribution; Both the populations should have same variances. In case the sample sizes of both the groups are roughly equal and their variances are also equal then the application of Student's original t-test is highly recommended. However, Welch's t-test is recommended where the variances are not equal irrespective of the size of the samples (Elliott and Woodward, 2006); and The data used to apply the test must be sampled independently from the two populations. However, this assumption is not testable but in case the data are known to be collected from mutually dependent sources, then the results derived may not be conclusive. Hypothesis for Independent Sample t-test The null hypothesis for the independent samples t test is as follows: H 0 : µ 1 = µ 2 72

11 or µ 1 - µ 2 = 0 S x1x2 is an estimator of the common standard deviation of the two samples. The objective of defining S x1x2 is to assure that its square is an unbiased estimator of the common variance whether or not the population means are the same. In these formulae, n = number of participants, 1 = group one, 2 = group two. n 1 is the number of degrees of freedom for either group, and the total sample size minus two (that is, n 1 + n 2 2) is the total number of degrees of freedom, which is used in significance testing. However, to test and satisfy the second assumption of homogeneity of variances, Levene test for homogeneity of variances was applied WELCH T-TEST One of the assumption before application of independent sample t-test is homogeneity of the variances. Therefore, the application of the test will not provide reliable results when this assumption is violated. In such a case, another form of t-test is applied namely Welch s t-test (Welch, 1947). This form of t-test is particularly for the situations where the samples share unequal variances. This is because it has an in-built application called the Welch correction designed with the objective of applying a valid t-test when the population variances are not equal. In Welch s t-test: The degrees of freedom are modified to account for the unequal sample sizes and the unequal variances as well as small sample sizes. The Standard Error does not pool the sample variances to estimate a common population variance. 73

12 Welch s t-test uses following equation to derive conclusive results from the data with unequal variances. Welch s t = Sample Mean 1 Sample Mean 2 Variance 1 + Variance 2 Sample Size 1 Sample Size 2 Under Welch s t-test, a corrected number of degrees of freedom are utilized to assess the significance of the t-statistic computed as usual. This number of degree of freedom is determined by applying the formula given hereunder: Variance 1 + Variance 2 Sample Size 1 Sample Size 2 2 Welch s d.f. = 2 2 Variance 1 Variance 2 Sample Size 1 Sample Size 2 + Sample Size 1 1 Sample Size 2 1 However, while calculating the degree of freedom under Welch t-test, the researcher must keep in mind that it cannot be larger than n 1 +n 2-2 and it cannot be smaller than n 1-1 and n 2-2. The application of Welch s t-test is based on two assumptions which states that: the observations are independent from each other; and both the samples have been drawn from normal populations. However, there may arise certain cases where the assumptions of Welch s t-test are not satisfied. In those particular cases, the researcher can overlook the second assumption of drawing of samples from normal populations but in no case the test can be applied where the samples turn out to be dependent on each other. 74

13 3.7.3 Levene's Test for Homogeniety of Variances Levene's test of Homogeneity of Variances (Levene, 1960) is an inferential statistic used to assess the equality of variances in different samples. Equal variances across samples are called homogeneity of variance. Some common statistical procedures such as T-test and Analysis of Variance (ANOVA) assume that variances of the populations from which different samples or groups have been drawn are equal. Levene's test is used to verify this assumption. The tests such as F test or the Bartlett test may be applied to test the differences in variances. However, these tests tend to be highly sensitive towards the assumption that the population is normally distributed. Therefore, application of Levene s test for measuring the differences in variances has been widely suggested. The suitability of Levene s test over other test is due to the reason that Levene (1960) proposed to compare the mean values of absolute deviations from sample means rather than variances. Schultz (1983) proposed Levene test to be among the best of the tests for determining the differences in variation. Also, Hines and O Hara Hines (2000) termed it as a widely used and robust test. Furthermore, Milliken and Johnson (1984) also recommended the use of Levene test, subject to the condition that there is confidence that the data are nearly normal or the data set is very large. However, Levene (1960) further recommended the application of t test as he regarded the test to be quite robust but further emphasized on the suitability of t-test in the cases where observations were drawn from a normal distribution. As Levene's test is used for comparing the means, it tests the null hypothesis that the population variances are equal. If the resulting p-value of Levene's test is less than some critical value (typically.05), the obtained differences in sample variances are unlikely to have occurred based on random sampling. Thus, the null hypothesis of equal variances is rejected and one may conclude that there is a difference between the variances in the population. Given a variable Y with sample of size N divided into k subgroups, where N i is the sample size of the ith subgroup, the Levene test statistic is defined as: 75

14 H 0 : H a : for at least one pair (i,j). The Levene test statistic W is defined as: Where, W is the result of the test; k is the number of different groups to which the samples belong, N is the total number of samples, N i is the number of samples in the i th group, Y ij is the value of the j th sample from the i th group, where Z ij can have one of the following three definitions: 1. Where, is the mean of the ith subgroup. 2. Where, is the median of the ith subgroup. 3. Where, is the 10% trimmed mean of the ith subgroup. 76

15 The significance of W is tested against F (α, k 1, N k) where F is a quantile of the F test distribution, with (k 1) and (N k) are its degrees of freedom, and α is the chosen level of significance (usually 0.05 or 0.01). The rejection of null hypothesis will tend to the inability of Student s t-test to derive accurate results. In such cases, Welch s t-test is highly recommended which ignores the differences in variances and provides direction to apply valid t-test Accounting Ratios The simple meaning of ratios is to express one number in terms of another. A ratio is regarded as a statistical yardstick which attempts to compare and measure relationship between two or more variables. In finance, the term accounting ratios is used to describe relationship between the figures shown in financial statements i.e. Balance Sheet and Profit and Loss Account. In the present study various ratios have been used wherever required to determine the relationship between the variables considered under different sectors Trend Analysis The term "trend analysis" refers to the concept of collection of data and attempts to spot a pattern, or trend in the data. As the data related to financial indicators has been processed for a period of 18 years, therefore, the trend analysis has been carried out in order to know the change over this period while making a comparative analysis of the financial indicators Logistic Regression Logistic regression analysis (LRA) extends the techniques of multiple regression analysis to research situations in which the outcome variable is categorical. The model for logistic regression analysis assumes that the outcome variable, Y, is categorical (e.g., dichotomous), but LRA does not model this outcome variable directly. Rather, 77

16 LRA is based on probabilities associated with the values of Y. For simplicity, and because it is the case most commonly encountered in practice, we assume that Y is dichotomous, taking on values of 1 (i.e., the positive outcome, or success) and 0 (i.e., the negative outcome, or failure). For theoretical, mathematical reasons, LRA is based on a linear model for the natural logarithm of the odds (i.e., the log-odds) in favor of Y = 1 (Dayton, 1992) An explanation of logistic regression begins with an explanation of the logistic function, which, like probabilities, always takes on values between zero and one: The input is z and the output is ƒ(z). The logistic function is useful because it can take as an input any value from negative infinity to positive infinity, whereas the output is confined to values between 0 and 1. The variable z represents the exposure to some set of independent variables, while ƒ(z) represents the probability of a particular outcome, given that set of explanatory variables. The variable z is a measure of the total contribution of all the independent variables used in the model and is known as the logit. The variable z is usually defined as where is called the "intercept" and,,, and so on, are called the "regression coefficients" of,, respectively. The intercept is the value of z when the value of all independent variables are zero (e.g. the value of z in someone with no risk factors). Each of the regression coefficients describes the size of the contribution of that risk factor. A positive regression coefficient means that the explanatory variable increases the probability of the outcome, while a negative regression coefficient means that the variable decreases the probability of that outcome; a large regression coefficient means 78

17 that the risk factor strongly influences the probability of that outcome, while a nearzero regression coefficient means that that risk factor has little influence on the probability of that outcome. Coefficient of Logit Regression and Mode Fit Instead of finding the best fitting line by minimizing the squared residuals, as we did with OLS regression, we use a different approach with logistic Maximum Likelihood (ML). ML is a way of finding the smallest possible deviance between the observed and predicted values (kind of like finding the best fitting line) using calculus (derivatives specifically). With ML, the computer uses different "iterations" in which it tries different solutions until it gets the smallest possible deviance or best fit. The regression coefficient of the model represented by McFadden R-squared sometimes called the Likelihood Ratio Index [LRI]): 2 LL( B) 1 LL(0) McFadden s R square tends to be smaller than R-square. This is because the Likelihood Ratio Index (LRI) depends on the ratio of the beginning and ending log-likelihood functions, it is very difficult to "maximize the R 2 " in logistic regression. and the values between 0.2 to 0.4 are considered to be highly satisfactory (McFadden, 1979) Testing for Assumptions of Regression Model Heteroscedasticity Most of the basic forms of models make use of the assumption that the errors or disturbances u i have the same variance across all observation points. However, when the variance of errors differs at different values of the independent variables, the presence of heteroscedasticity is indicated. Heteroscedasticity is reflected in the residuals estimated from a fitted model. To deal with this problem, heteroscedasticity- 79

18 consistent standard errors are used to allow the fitting of a model containing heteroscedastic residuals. One of such approaches is White's (1980) estimator, which explicitly tests forms of heteroscedasticity i.e. the relation of u 2 with all independent variables (X i ), squares of independent variables (X 2 i ) and all cross products (X i X j for i=j). The present study makes use of White Heteroscedasticity Consistent Covariances to deal with problem of existence of any heteroscedasticity in the data as this test is also particularly suitable for large sample sizes. Multicollinearity Multicollinearity is a statistical phenomenon in which two or more predictor variables in a multiple regression model are highly correlated. In this situation, the coefficient estimates may change erratically in response to small changes in the model or the data (Donald and Glauber, 1967). To detect the presence of multicollinearity in data, two statistical measures have been employed. These are: Variance Inflation Factor (VIF): The Variance Inflation Factor (VIF) quantifies the severity of multicollinearity in an ordinary least squares regression analysis. It provides an index that measures how much the variance of an estimated regression coefficient is inflated because of collinearity. Usually, the VIF values ranging from 4 to 10 indicate the presence of higher multicollinearity between the predictors (Rogerson, 2001; and Pan & Jackson, 2008). In order to check the presence of multicollinearity in the data, VIF s have been used in all regression models. Tolerance: Tolerance is a measure of collinearity reported by most statistical programs such as SPSS. A small tolerance value indicates that the variable under consideration is almost a perfect linear combination of the independent variables already in the equation and that it should not be added to the regression equation (Cohen et al., 2003). All variables involved in the linear relationship will have a small tolerance. If a low tolerance value is accompanied by large standard errors 80

19 and non significance, multicollinearity may be an issue (Fox, 1991). The measure of tolerance is given as Where, is the coefficient of determination of a regression of explanatory j on all the other predictors. Tolerance values of less than 0.20 or 0.10 and indicates a multicollinearity problem (O'Brien 2007). In other words, a tolerance close to 1 means existence of modest multicollinearity in the data, whereas a value close to 0 suggests that multicollinearity may be a severe threat. 3.8 SCOPE AND LIMITATIONS OF THE STUDY This study has universal applicability as the data for the study consists of all the Indian and foreign corporations operating in India which have been listed with Bombay Stock Exchange. Moreover, the study has attempted to measure the performance of multinationals in India and also attempts to compare their performance with Indian counterparts, the study provides a base for the policy makers to estimate the impact of foreign companies on domestic companies. This will help them in formulating the policies keeping in mind the interest of domestic companies. This study will further help the Indian entrepreneurs to know the areas where foreign companies are not performing well or where less competition is posed by foreign companies and hence guide them to invest in such areas and to earn profits. Limitations have always been a part and parcel of any analytical research work. This study is also not free from the ambit of the same. Some of the limitations are listed below: 81

20 1. The study is based on secondary data; therefore, the study suffers from all limitations suffered by a research based on secondary data. 2. As the topic of the research is too comprehensive to cover all the units as well as sectors in the universe in the given time frame, however, the study remained confined to three main corporate sectors of India i.e. chemicals, food and machinery. Therefore, the findings of the study may not be generalized to excluded sectors and a separate research is required to be conducted for these sectors. 3. Due to limitation of time and resources, the study excluded tertiary sector which constitutes a significant share of Indian GDP. As policy guidelines are favoring increasing share of foreign participation in Indian service sector, therefore, research in this area can be pursued in future. 4. Limitations concerning to time, also denied a possibility of collection of primary data to know the manager s perception of Indian policy framework concerning smooth growth of MNCs for mutual benefit. 82