Source of Data
The data relating to the different economic variables of companies have been collected from various issues of the Bombay Stock Exchange Official Directory.
The source of data for the fixed investment policy of Chemical industry is the data relating to the individual sample companies in Chemical industry. The industry, for the purpose of the study, means the aggregate of sample units in the industry. Thus the cross section data of micro level economic variables is added to make up the industry data.
Period of Study
The present study covers a period of 10 years from 2000 to 2009. Since the fixed investment policy is a long-term policy, a period of 10 years is considered to be long enough to study the Fixed Investment policy of companies/Industries.
The Sample Selection
The selection criteria of the companies for inclusion in the sample of the study have been that
1. Companies must have been incorporated on or before 1975, i.e., 25 years before the period for which analysis has been started here so that a minimum period of at least 25 years must have been elapsed for them to establish themselves and invest in fixed assets;
2. Companies must have had a paid-up capital of more than Rs 10 lakhs in 1975 so that only medium and large companies as per the classification of the Reserve Bank of India are included in the sample; and
3. Companies must be continuously profit making companies in all 10 years (which is the study period here) so as to ensure that only which made profits on consistent basis are included.
Based upon the above selection criteria a total of the following 20 firms constitute the size of the sample for the purpose of this study.
1. BASF India Ltd.
2. Bayer (India) Ltd.
3. Cheminar drugs Ltd.
4. Cipla Ltd.
5. Clariant (India) Ltd.
6. Colour Chem Ltd.
7. Excel Industries Ltd.
8. Glaxo India Ltd.
9. Gujarat Narmada Valley Fert. Co. Ltd.
10. Gujarat State Fert & Chemicals Ltd.
11. Indian Oil Corporation Ltd.
12. Indian Petro Chemicals Corporation Ltd.
13. Madras Refineries Ltd.
14. Monsanto Chemicals of India Ltd.
15. Nicholas Piramal India Ltd.
16. Parke Davis (India) Ltd.
17. Pidilite Industries Ltd.
18. Rhone-Poulenc (India) Ltd.
19. Tata Chemicals Ltd.
20. Zandu Pharmaceutical Works Ltd Variables
A list of the variables – both dependent and independent – that are used in this study is presented.
1. GBt- (t-1) = Change in Gross Block
2. Pm t- (t-1) = Change in Plant & Machinery Independent Variables
1. S t- (t-1) = Change in sales
2. GIFt = Gross Internal Funds
3. NL t = Stock of Net Liquidity
4. Dt =Dividends
5. ECt- (t-1) =Growth of equity capital
6. DETOUTt =Debt outstanding
7. Tt = Provision for taxes
8. It =Interest on borrowed funds
Step Wise Regression
The present study is mainly based on stepwise multiple regression analysis. This technique begins with the simple correlation matrix and enters into regression of the independent variables most highly correlated with the dependent variable. Using the partial coefficients generated with respect to the other variables, the computer programme then selects the next variable to enter the model.
Stepwise regression permits the analyst to start with a large number of variables that might have predictive values and then use the model to select the particular variables that appear to provide the prediction.
The data used in this study was processed by using computer packages, they are Statistica and Limdep. The multiple linear stepwise regression was run in order of importance in terms of explanatory powers of different variables influencing the dependent variable in the study. In other words, which independent variable has the greatest effect in determination of the dependent variable?
How sensitive is dependent variable to fluctuations in independent variables? This technique is adopted in order to obtain a realistic picture of the importance of the various independent variables, which influence financing investment in the Chemical industry in India.
This study is conducted on the basis of three models. These three models have been tested in the case of each company. They are
1. Adding Model
3. Elimination Model.
The above three models have been tested in each case with the intercept term. Thus altogether 15+ equations are estimated in each case.
It may be noted that in this model, an independent variable has been entered into the model at an earlier step, and then another independent variable is added to the first one and then another variable etc. So ultimately all the independent variables are added and tested under this model.
The following are the equations, which are estimated under this model.
In this model the first two independent variables (change in sales and gross internal funds) are kept as constant variables because these two are very closely related to the dependent variables, and the third variable is changed in each model.
The following are the equations, which are estimated under this model.
In elimination model, the estimated equations are not constant but the number of equations depends on the significance of the variables which proved to be significant.
The following procedure is adopted while estimating the equations. Initially, all the independent variables are included in the model. Based upon the significance of‘t’ values, the variable with the least‘t’ value is dropped and then again the equation is estimated with the remaining independent variables. Again the variable with the least‘t’ value is dropped and the equation is again estimated. This process is continued till all the independent variables in the equation have proved to be significant either at 5% or at 10% level.So the number of equations varies depending upon the significance of variables in each case of companies.
The above 15+ equations are estimated for all the 20 companies and industry aggregate. The total numbers of estimated equations are as follows:
For 20 companies & industry aggregates in two cases (both gross block and plant & machinery):
In Adding Model …….21x8x2 = 336
In Constant Model ……..21x6x2 = 252
In Elimination Model ……………….. = 149
Total = 737
Thus altogether 737 equations have been estimated with all the necessary tests, using the data for 10 years in each case.To find out the effect of different independent economic variables on the fixed investment of the companies during the period of this study, the Multiple Linear Regression Analysis is used with all its limitations.
Selection of the Best Model
The following procedure is adopted to select the best model in each case from out of the 15+ estimated equations.
Step – I Out of the 15+ estimated equations in each case, all those equations, whose Multiple Correlation Coefficients are found to be significant at 5% level based on their calculated ‘F’ values are picked up for further analysis.
Step – II The equations thus picked up according to step-I above are further screened in the following way:
1) The values of intercept term (b0) and other regression coefficients (b1, b2, b3) are tested at 5% level of significance based on their calculated‘t’ values. If only one equation is found in which all the explanatory variables are significant at 5% level, then that equation is taken as the best model to explain the fixed investment behavior of the company. If, on the other hand, there are two or more equations in which all the explanatory variables are found significant at 5% level, the procedure explained in step III is followed.
2) But if, in a company, there is not even a single equation in which all the independent variables show significant effect at 5% level, the significance level is relaxed and the impact of the variable is tested at 10% level wherever necessary. That is, the variables, which are not significant at 5% level, are tested at 10% level of significance. However, this has happened in a very few cases in this study. If only one equation is found in which the explanatory variables are significant at 5% level or 10% level, then that model is selected as the best model to describe the fixed investment behavior of the company. On the other hand, if there are two or more than two equations in which the independent variables are significant at 5% or 10% level, the procedure explained that in step III is followed to decide the best model.
Step – III As stated in step II, if there are two or more equations in which all the explanatory variables are significant that particular equation whose R2 is the highest is chosen as the best equation to explain the fixed investment behavior of the company.
Limitations of the Study
This study has the following limitations.
1) The accounting years of the sample companies are not common and the closing of the accounting years is spread over all the 12 months of the year. So for the industry aggregate data the accounting year is not uniform.
2) The Industry data, for the purpose of the study, comprise the aggregate of the data of the micro level sample units that are selected for this study. As there is difference in the classification of industries between Reserve Bank of India and the Bombay Stock Exchange, the RBI data could not be relied upon for the industry aggregate data and the Bombay Stock Exchange Directory does not provide the Industry aggregate data. Since it is highly difficult to collect the data of all the firms which appear on the Bombay Stock Exchange Directory the aggregate data of the sample micro level units is taken to represent the industry data for this study.
a) The data for the study are taken in absolute values as given in the Bombay Stock Exchange Directory and no price deflator is used to adjust for the inflationary trends.
b) This study is only exploratory in its objectives and does not aim at recommending any policy measures either for the companies or for the government.