Data Of Smes Rating Information Finance Essay - Free Essay Example

As discussed in the Chapter 3 on the methodology of the research, the data of SMEs rating information and the earnings information for the period has been computed. This chapter would analyze the obtained secondary data information and would provide the plan for hypothesis testing. 5.2. SECONDARY DATA ANALYSIS For the current study, secondary data on manufacturing and service SMEs was accessed from the CMIE-Prowess database. Initially by giving the search option in the Prowess data base adhering to the definitions of an SME as specified in chapter 4, the financial statement of manufacturing and service-based SMEs has been obtained. The lack of consistency of the data has been identified and it was mainly because of the following reasons: 1. Information on earnings and accruals were not available for the required time; 2. Many companies were new; and 3. Many SMEs rating information were not available. Unfortunately, this drastically reduced the sample available for secondary data analysis to 50. 5.2.1. DESCRIPTIVE STATISTICS The following table (Table 5.1.) depicts the descriptive statistics for variables in the regressions. It is computed from the secondary data of Manufacturing and Services SMEs. Table 5.1. Descriptive Statistics ÃÆ' ¢Ãƒâ€¹Ã¢â‚¬  Ãƒ ¢Ã¢â€š ¬Ã‚  Rating t ÃÆ' ¢Ãƒâ€¹Ã¢â‚¬  Ãƒ ¢Ã¢â€š ¬Ã‚  Rating t+1 Earn t ACC t CFO t Lev t Size t Cap Int t Liq t Mean 0.1200 0.2400 1.8971 1.4871 0.0184 0.3046 3.0378 0.0646 2.8593 Standard Error 0.1016 0.1264 0.1126 0.1423 0.0238 0.0335 0.0761 0.0101 0.3950 Median 0.0000 0.0000 1.9181 1.6457 0.0365 0.2670 2.9652 0.0320 1.8906 Standard Deviation 0.7183 0.8935 0.7963 1.0063 0.1682 0.2371 0.5379 0.0718 2.7934 Sample Variance 0.5159 0.7984 0.6341 1.0126 0.0283 0.0562 0.2893 0.0052 7.8032 Kurtosis 7.1147 2.1525 1.0811 -1.0021 14.5300 1.6161 0.3901 2.5296 7.3048 Skewness 1.5382 0.3917 -0.7036 -0.2597 -2.9950 1.0686 0.5198 1.6068 2.4717 Range 5.0000 5.0000 3.6621 3.4761 1.1197 1.0447 2.7613 0.3190 14.9807 Minimum -2.0000 -2.0000 0.0000 0.0000 -0.8588 0.0000 1.8904 0.0002 0.0328 Maximum 3.0000 3.0000 3.6621 3.4761 0.2610 1.0447 4.6517 0.3192 15.0134 Sum 6.0000 12.0000 94.8551 74.3542 0.9184 15.2292 151.8877 3.2296 142.9664 Note: Count= 50 Source: Self Computation Using Microsoft Excel INTERPRETATION The descriptive statistics figures for SMEs indicate that the median earnings for the firms stand at 1.91 and the mean for CFO and ACC are all found to be positive. The Kurtosis value of future change in ratings is found to be 2.15 which is close to the normality range of 3. The Skewness for the same is found to be 0.39 which is nearer to the normality range of 0.3. Hence to some extent the dependent value follows a normal distribution. The Skewness is found to be negative for all independent variables under discussion namely the earnings, cash flow and the accruals. The maximum value of earnings is found to be 3.6 and that of cash flow is 0.26. The standard deviation for all the variables are in the range less than 1 except for accruals which is found to be 1.0063. The future change in ratings has a maximum change of +3 and a minimum of -2 within the period under study. INFERENCE From the above descriptive statistics, it can be inferred that the change in ratings for the SMEs does not seem to have a greater upgrade or downgrade. However few companies have seen changes in their ratings. The earnings, accruals and cash flow measures are all varying with greater extent in the industry and thus could be a possible factor for the credit rating changes. CONCLUSION It can be concluded from the above table that the rating changes that the SMEs face today be it upgrade or downgrade depends on the accruals, cash flow and on the earnings as a change in these variables are found to be relatively high. The companies differ greatly in their size which has been taken as an independent variable for the study. 5.2.2. CORRELATION ANALYSIS Correlation is one of the most common and useful tools in statistics. Correlation describes the degree to which two variables are related. Correlation coefficient ranges between +1 and -1; +shows the variables are perfectly positively correlated and -1 shows the two variables are perfectly negatively correlated. The following table (Table 5.2.) depicts the correlation between the variables under study. Table 5.2. Correlation Matrix    ÃÆ' ¢Ãƒâ€¹Ã¢â‚¬  Ãƒ ¢Ã¢â€š ¬Ã‚  Rating t ÃÆ' ¢Ãƒâ€¹Ã¢â‚¬  Ãƒ ¢Ã¢â€š ¬Ã‚  Rating t+1 Earn t ACC t CFO t Lev t Size t Cap Int t Liq t ÃÆ' ¢Ãƒâ€¹Ã¢â‚¬  Ãƒ ¢Ã¢â€š ¬Ã‚  Rating t 1                         ÃÆ' ¢Ãƒâ€¹Ã¢â‚¬  Ãƒ ¢Ã¢â€š ¬Ã‚  Rating t+1 0.240 1                      Earn t 0.491 0.362 1                   Earn t+1 0.447 0.281 0.592                   ACC t 0.226 0.354 0.524 1                CFO t 0.385 0.065 0.351 0.099 1             Lev t -0.182 0.059 -0.175 -0.005 -0.502 1          Size t 0.015 0.120 0.573 0.373 0.157 -0.024 1       Cap Int t -0.076 0.004 -0.305 -0.199 0.082 0.197 -0.482 1    Liq t 0.107 0.076 0.164 0.163 -0.217 0.279 -0.078 -0.198 1 Source: Self Computation Using Microsoft Excel INTERPRETATION A correlation analysis was performed to verify possible association between and among the variables, in order to verify whether there is any linear correlation between and among the variables of interest of the study (see Table 5.2). The correlation matrix for the SMEs indicates that ÃÆ' ¢Ãƒâ€¹Ã¢â‚¬  Ãƒ ¢Ã¢â€š ¬Ã‚  Ratingt and ÃÆ' ¢Ãƒâ€¹Ã¢â‚¬  Ãƒ ¢Ã¢â€š ¬Ã‚  Ratingt+1 are positively correlated with Earnt, ACCt and CFOt. ÃÆ' ¢Ãƒâ€¹Ã¢â‚¬  Ãƒ ¢Ã¢â€š ¬Ã‚  Ratingt+1 is significantly related to Earnt and ACCt but it is not significantly related to CFO in the univariate analysis. ÃÆ' ¢Ãƒâ€¹Ã¢â‚¬  Ãƒ ¢Ã¢â€š ¬Ã‚  Ratingt is found to be negatively correlated with Lev and CapInt. However ÃÆ' ¢Ãƒâ€¹Ã¢â‚¬  Ãƒ ¢Ã¢â€š ¬Ã‚  Ratingt+1 is found to be positively associated with all the variables. INFERENCE From the above table it can be inferred that the rating information is more dependent on the earnings than the cash flow measures based on the Pearson coefficients obtained. More significant correlation exists among the earnings information than that of the cash flow and accrual measures. The accrual measure has shown a very little significance when compared to that of the cash flow and earnings measures. CONCLUSION The SMEs change in their future rating depends more on the current earnings information than that of their accrued earnings and cash flows. It could be concluded that rating agencies take more of current earnings information rather future predicted cash flows. 5.2.3. REGRESSION ANALYSIS Regression analysis is the process of constructing a mathematical model or function that is being used to predict or determine the value of one variable by another variable. In the regression model the variable to be predicted is called a dependent variable and the variables upon which the dependent variable depends are called independent variables. For the research study, two models are developed one for earnings and the other for the cash flow. The dependent variable is the ÃÆ' ¢Ãƒâ€¹Ã¢â‚¬  Ãƒ ¢Ã¢â€š ¬Ã‚  Ratingt+1 and the independent variables are Earnt, ÃÆ'Ã… ½Ãƒ ¢Ã¢â€š ¬Ã‚ RATINGt, LEVt, SIZEt, SUBORDt, CAPINTENt, LIQt for the first model and ACCt, ÃÆ'Ã… ½Ãƒ ¢Ã¢â€š ¬Ã‚ CFOt, ÃÆ'Ã… ½Ãƒ ¢Ã¢â€š ¬Ã‚ RATINGt, LEVt, SIZEt, SUBORDt, CAPINTENt , LIQt for the second model. A summary of the model of regression analysis is given in table 5.3. Table 5.3. Model Summary Model R Square Adjusted R Square Std. Error of the Estimate 1 0.562 0.603 0.8508 2 0.488 0.533 0.7982 Source: Self Computation using XLSTAT INTERPRETATION R Square represents the proportion of standard deviation in the dependent variable (Future rating changes) which could be explained by the independent variables (Earnings, Change in ratings for the current year, Leverage, size, Liquidity, Capital Intensity, Subordinated debt). However the variable subordinated debt has not been considered for modelling as the value doesnt change for the entire sample. This is an overall measurement of the strength of the association and hence the extent to which any particular variable (independent) associated with the other (dependent variable) is not reflected. The value of R square for the first model is 0.562 which shows a relatively acceptable association between the dependent and the independent variables. Adjusted R Square is an adjustment to the R Square that expresses the effect on the addition of any other external predictors to the model. The value of adjusted R Square for the first model is found to be 0.603 and thus it indicates if an ad dition of external predictor to the model will add significant predictability to the dependent variable. The dependent variable (Future rating changes) which could be explained by the independent variables (Accruals, Cash Flow, Change in ratings for the current year, Leverage, size, Liquidity, Capital Intensity, Subordinated debt). However the variable subordinated debt has not been considered for this modelling too as the value doesnt change for the entire sample. The value of R square for the first model is 0.488 which shows a relatively acceptable association between the dependent and the independent variables. The value of R square is found to be lesser as the rating information does not necessarily consider the cash flow information and it is dependent on many other industry factors which are not included for the study. The value of adjusted R Square for the first model is 0.533. Std. error of the Estimate, also referred as the root mean square error represents the standard deviation of the error term and the square root of the mean square for the residual in the ANOVA Table 5.4 presented below. Table 5.4. ANOVA Table Model DF Sum of squares Mean squares F P value 1 Regression 6 8.563 1.427 2.008 0.0557 Residual 43 30.557 0.711       Corrected Total 49 39.120          2 Regression 7 6.974 0.996 1.302 0.0273 Residual 42 32.146 0.765 Corrected Total 49 39.120 Note: H2a0. Credit rating agencies do not fully incorporate information in earnings about future performance into their ratings. H2b0. Credit rating agencies do not fully incorporate information in accruals and cash flows about future performance into their ratings. Source: Self Computation using XLSTAT INTERPRETATION F test is used to test whether the model is statistically significant. The p-value of the F-test is looked at so as to see the overall model is significant. The P value is found to be 0.0557 and 0.0273, proving that the models are statistically significant. For the Model 1, the p value is greater than the significance level at 5 per cent. So the null hypothesis is accepted however the null hypothesis is rejected in the model 2 and it can be concluded that all mean values are not equal. It implies that there is a significant difference among all the independent variables. A summary of coefficients of Model 1 and Model 2 are given in the table 5.5 5.6. Model 1 Source Value Standard error t P Value Intercept 0.282 0.959 0.294 0.770 Earn t 0.546 0.205 2.658 0.011 ÃÆ' ¢Ãƒâ€¹Ã¢â‚¬  Ãƒ ¢Ã¢â€š ¬Ã‚  Rating t 0.305 0.178 1.717 0.039 Lev t 0.339 0.599 0.901 0.373 Size t -0.223 0.335 -0.667 0.509 Cap Int t 0.268 2.157 0.291 0.772 Liq t -0.014 0.052 -0.272 0.787 Table 5.5. Coefficients for Model 1 Source: Secondary Data Table 5.6. Coefficients for Model 2 Model 2 Source Value Standard error t P Value Intercept 0.266 0.991 0.471 0.640 ÃÆ' ¢Ãƒâ€¹Ã¢â‚¬  Ãƒ ¢Ã¢â€š ¬Ã‚  Rating t 0.217 0.149 1.450 0.154 ACC t 0.312 0.139 2.243 0.030 CFO t 0.340 0.942 0.361 0.250 Lev t 0.273 0.700 0.391 0.698 Size t -0.025 0.311 -0.079 0.937 Cap Int t 0.273 2.337 0.354 0.725 Liq t 0.008 0.052 0.162 0.872 Source: Secondary Data INTERPRETATION The first variable (Intercept) represents the constant, is the predicted change in the future credit ratings when all other variables are not considered. From the Table 5.5, Coefficients for Model 1, the coefficient value tells us about the relationship of each variable with the independent variable. For example, let us look at these variables: 1. Earnings: The coefficient for earnings is 0.546. So for every unit increase in earnings, a 0.546 unit increase in future rating changes is predicted, holding all other variables constant. 2. Size: The coefficient for size is -0.223. So for every unit increase in growth, a 0.223 decrease in change in future credit ratings is predicted, holding all other variables constant. 3. Current Rating Changes: The coefficient for Current Rating Changes is 0.305. So for every unit increase in Current Rating Changes, a 0.305 unit increase in change in future credit ratings is predicted, holding all other variables constant. The relationship in the case of other variables also follows the same pattern. Similarly, From the Table 5.6, Coefficients for Model 2, the coefficient value tells us about the following relationship. 1. Accruals: The coefficient for Accruals is 0.312. So for every unit increase in accruals, a 0.312 unit increase in future rating changes is predicted, holding all other variables constant. 2. Cash Flow: The coefficient for cash flow is 0.340. So for every unit increase in cash flow, a 0.340 increase in change in future credit ratings is predicted, holding all other variables constant. 3. Current Rating Changes: The coefficient for Current Rating Changes is 0.217. So for every unit increase in Current Rating Changes, a 0.217 unit increase in change in future credit ratings is predicted, holding all other variables constant. The Beta values have an associated standard error indicating the extent to which these values would vary across different samples, and these standard errors are used to determine whether or not the B value differs significantly from zero. By standardizing the variables prior to regression, all the variables are put on the same scale and the magnitude of the coefficients are compared so as to see which has more effect. The t and p value are the statistical tools that are used to test for the significance in of a given coefficient. If the t test associated with the B value is significant then that predictor is making significant contribution to the model. Using an alpha of 0.05 for the first model, we may explain these variables as follows: 1. The coefficient for earnings (0.546) is significantly different because its p-value is 0.011, which is lesser than 0.05. 2. The coefficient for change in current ratings (0) is also significantly different and its p-value is 0.039, which is lesser than 0.05. 3. The coefficient for capital intensity is 0.268 is not significantly different as its p-value is 0.772, which is greater than 0.05 The relationship in the case of other variables also follows the same pattern. The regression equation for the model 1 is ÃÆ' ¢Ãƒâ€¹Ã¢â‚¬  Ãƒ ¢Ã¢â€š ¬Ã‚  Ratingt+1=0.282+0.305*ÃÆ' ¢Ãƒâ€¹Ã¢â‚¬  Ãƒ ¢Ã¢â€š ¬Ã‚  Ratingt+0.546*Earnt+0.339*Levt-0.223*Sizet+0.2683*Cap Intt-0.014255*Liqt The regression equation for the model 2 is ÃÆ' ¢Ãƒâ€¹Ã¢â‚¬  Ãƒ ¢Ã¢â€š ¬Ã‚  Ratingt+1=0.2664+0.217*ÃÆ' ¢Ãƒâ€¹Ã¢â‚¬  Ãƒ ¢Ã¢â€š ¬Ã‚  Ratingt-0.31239*ACCt+0.34039*CFOt+0.27331*Levt-0.0246*Sizet+0.27285*Cap Intt+0.00841237*Liqt 5.2.4. HYPOTHESIS TESTING To test the first major hypothesis the relation between accrual-based earnings and credit ratings and the relation between cash flows and credit ratings, the level of credit rating (Ratingt) is regressed on total earnings (Earnt) and several cash flow measures. The coefficient on Earnt is significantly positive with the value being 0.469 and the adjusted R square is 0.213 and thus indicating earnings being an important input in the determination of credit rating. In another regression, the coefficient of CFOt is 0.278 is also positive and significant and indicating the rating agencies reliability on the cash flow for credit ratings. The adjusted R square is found to be 0.145. Thus it could be found that the adjusted R square in cash flow model is just 68% of the adjusted R square of the earnings model as indicated by the ratio of RSquare cash flows / R Square earnings (0.145/0.213) This indicate that earnings are more related to credit ratings than the cash flow measures and earnings could mitigate the timing problems. Thus we are rejecting the null hypothesis of the relation between accrual-based earnings and credit ratings is not stronger than the relation between cash flows and credit ratings. 5.3. CONCLUSION From the secondary data analysis it is concluded that the level of credit rating is dependent on both the earnings accruals and the cash flow measures. However, it was proved statistically that the earnings information are more significant in credit rating than the cash flows. The two major model equations that has been designed to test the other hypotheses indicate that credit rating agencies does not fully incorporate earnings information in their future credit rating changes and that of accruals and cash flows information into credit rating process. The other independent variables and industry factors are to be included while arriving at the conclusion that there exists an inefficiency in credit rating for SMEs that are under consideration.