Performance Evaluation Model for Primary Agricultural Credit Societies

Transcription

1 Performance Evaluation Model for Primary Agricultural Credit Societies Yashavantha Dongre and M V Narayana Swamy This article by Dongre and Narayana Swamy is an attempt to evolve an appropriate statistical model to evaluate the financial performance of the primary agricultural credit societies. It is based on the empirical data gathered from 14 sample societies of two talukas in the Dakshina Kannada district of Karnataka state. Considering that fund management is the basic weak link in most primary agricultural credit societies, it is imperative that they be able to evaluate their performance and, based on the feedback, restructure their financial policies. It is hoped that the model suggested here comes handy to the management in periodic evaluation, feedback, and followup activities. Yashavantha Dongre is Reader in the Department of Post-Graduate Studies and Research in Commerce, Mangalore University, Mangalore and M V Narayana Swamy is Selection Grade Lecturer in the Department of Commerce, Bhandarkar's College, Kundapura. The authors are grateful to Dr K Aruna Rao, Department of Statistics, Mangalore University, for suggestions and help in preparing this article. The cooperative movement began in India as a move to provide credit to rele ase the rural community from the clutches of moneylenders. Accordingly, primary agricultural credit societies (PACSs) were assigned the role of providing agricultural credit to farmers, distributing critical agricultural inputs like seeds, fertilizers, pesticides, agricultural implements and machinery, providing facilities for marketing of agricultural produce, and inculcating the habit of thrift and saving. Over the years, the profitability of PACSs has been declining and their net worth has been eroded. Even though many factors have contributed to this state of affairs, the most significant factor has been the failure to introduce professionalism in functional management in general and financial management in particular. The primary task of any cooperative is to serve the economic needs of its members. If this is not done, there is no role for cooperatives. In order to serve the members well, a cooperative must ensure business efficiency which, in turn, closely depends on how well its finances are managed. Non-performance in the financial front will be reflected in the overall performance of the business. Therefore, cooperatives need to know where exactly they stand as far as their financial performance is concerned. However, they are handicapped by the lack of a tool to measure financial performance. Therefore, this paper aims at developing a statistical model to evaluate the financial performance of PACSs. Methodology For the purpose of this study, seven societies each in Puttur taluka and Kundapura taluka of Dakshina Kannada district (Karnataka) were selected. Financial data of the sample societies from to were analysed using statistical tools such as averages, correlation, standard normal distribution, Mann- Whitney U test, discriminant analysis, and cluster analysis. Review of Earlier Models A good number of models have been developed in the past to measure the efficiency of corporate and Vol. 24, No. 1, January - March

2 cooperative enterprises. Altman (1968), Gupta (1983), and Panigrahi and Mishra (1993) have developed models based on financial ratios to indicate the health of industrial concerns. But, as the behaviour of ratios varies from sector to sector and from industry to industry, the applicability of models thus arrived at cannot be generalized. Therefore, these models cannot be applied directly to the cooperative form of organizations. Sriram (1991) has developed a model by applying multiple discriminate analysis using ratio indicators as base to discriminate between sick and healthy cooperatives. The predictive power of the model is fairly accurate with cooperatives having trading activities and cooperatives with processing facilities, whereas it is not so accurate in predicting sickness in cooperatives dominated by credit activity. Sidhu and Sidhu (1990) measured success or failure of cooperatives on the basis of membership, loan per borrower member, owned capital per member, percentage of owned capital to total loan, percentage of overdues to loan outstanding, income per member, and profit/loss per member. The study is silent on the basis on which these variables were selected. There is, thus, a possibility that other variables may be more significant in discriminating between successful and unsuccessful cooperatives. Desai and Namboodiri (1991) have done an evaluation of farmers' service societies and PACSs at all-india level using seven variables. While these factors may be appropriate for a comparative study at the macro level, a different set of criteria may have to be used for a micro level study. A study on evaluation mechanism for impact of cooperatives on members was conducted by Reddy (1992). The variables used in this study were: membership and member-users, volume of business, classwise loan, recovery of loans, and cost of management. However, the study itself cautions that adequacy of these indicators varies from region to region based on socio-economic conditions in which the cooperatives function. Suresh and Vinaikumar (1993) studied the economic viability of cooperatives using positive net worth as the criterion. Loans, cash balance, deposits, investments, consumption loan, recovery of loans, borrowing members, and non-credit business are the significant variables which influence the performance of societies and the economically viable societies record highest performance in these variables. Any model used to evaluate the performance of cooperatives must include variables that indicate liquidity, profitability, and long-term solvency of business. But the evaluation methods developed in the studies quoted above are deficient in one or the other aspect of development. Therefore, these models may not be fully relevant in the context of PACSs. Hence, it is necessary to develop a method which can incorporate all the dimensions of development to measure efficiency. For PACSs whose activities are a combination of trading and banking, such a need is felt now more than ever before in view of their poor performance in recent years. Hence, an attempt has been made in this paper to develop a model based on scaling technique as suggested by Rao, Bhat, and Bhatta (1991). The scale is intended to measure the performance of PACSs by grouping them as good, satisfactory, and poor. This will be useful to management and administrators to evaluate and improve the performance of PACSs (Exihibit 1). Construction of the Scale The variables included in the construction of the scale are drawn from the audited annual accounts for five years from to of 14 societies. The fiveyear averages of the value of variables are taken here, since it has been established (Gupta, 1991) that fiveyear averages are more reliable than the value for any single year. Year to year figures can be easily manipulated by management but to do so continuously is difficult. Eighteen variables are selected in the first instance for the purpose of constructing the scale. Almost all the variables as shown in the profit and loss account and the balance sheet are included so that no important variable is left out. The scaling technique has a system to eliminate insignificant variables. Name of variables, mean annual values, and respective standard deviations are shown in Table 1. These standard scores are then converted into score values on a ten-point scale obtained from deciles of standard normal distribution. The scale values for operating expenses are given in opposite direction, because for operating expenses, low value represents efficiency and high value represents inefficiency. The scale values and total scores are shown in Table 2. To check the consistency of the variables included in the scale, i.e., whether the score value for a variable is in agreement with the total score value, the correlation coefficient of the score of each variable with the total score was calculated and is shown in Table 3. If the correlation coefficient is significant, we presume that the selected variables are appropriate. The 13 variables with correlation coefficient significant 46 Vikalpa

3 Table 1 : Mean and Standard Deviation of Variables Variables Mean Standard Deviation A Share Capital B Share capital Plus Reserves C Net Worth D Loans Borrowed (Outstanding) E Deposits F Total Debt (outstanding) G Net Fixed Assets H Investments I Loans to Members (Outstanding) J Loans Advanced (Current) K Quick Assets L Net Working Capital M Net Capital Employed N Total Income O Operating Expenses P Sales Q Turnover R Operating Cash Flow Source: Annual Accounts, to Let X r denote the value (lakh of rupees) of the jth variable for the ith" society (j=l to 18 and i = 1 to 14). In order to bring the variables into a comparable form, the variables are standardized by the formula, v j x ii- where )(, anc ' CT J are the mean and standard deviation of the jth variable. at significance level are retained for further analysis and the remaining variables are deleted at this stage. Total scores of all the sample units for these 13 variables are recalculated and a three-point scale using the standard normal distribution is constructed as follows: Score Value Scale Value Between < s.d s.d s.d. Table 2 : Scale Value of Variables in the Sample Units > s.d. 3 Here X and s.d. refer to mean and standard deviations of total scores for 14 societies. The total score, mean, and standard deviation of the total score and the classification of the sample units are shown in Table 4. The performance of those societies which get a scale value of 1 is poor whereas those that get a scale value of 3 is good; a scale value of 2 indicates satisfactory performance. Thus, we started with 18 variables whose values are measured in lakh of rupees. By adapting the scaling technique we have reduced them into scores which are countable. Therefore, it is appropriate to use a non-parametric test. In. order to study the discriminating ability of the variables included in the construction of the scale, Mann-Whitney U test is 1 2 Sample Units Scale Value of Variables A B C D E F G H I / K L M N 0 P Q R Total Byndoor Uppinangadi Khambadakone Shankaranarayana Panaje Vandse Kumbra Kavu Ullur Belenele Molahalli Nellyadi Narimogaru Kollur Vol. 24, No. 1, January - March

4 Table 3: Correlation Coefficient of the Variables Variables Correlation Coefficient A Share Capital ** B Share Capital Plus Reserves ** C Net Worth * D Loans Borrowed (Outstanding) E Deposits ** F Total Debt (Outstanding) ** G Net Fixed Assets H Investments ** I Loans to Members (Outstanding) " J Loans Advanced (Current) ** K Quick Assets ** L Net Working Capital ** M Net Capital Employed ** N Total Income ** O Operating Expenses ** P Sales * Q Turnover *" R Operating Cash Flow Two-tailed significance: * = 0.01;» = applied to the extreme groups, i.e., societies classified as poor and good. If the result of the test shows a significant difference, we can conclude that the variable is consistent with the scale, i.e., the test discriminates between the societies in accordance with the scale. Table 5 gives results of the Mann-Whitney U test. All the 13 variables are highly significant and in the case of 12 variables, the p value is zero indicating complete agreement with the scale. From the utility point of view it is desirable if a scale, having highly discriminating ability, can be constructed using a minimum number of variables. Thus, four variables share capital, deposits, loans outstanding, and net capital employed are left out as they overlap with other variables in the set. A new scale is finally constructed with only nine variables: share capital plus reserves, total debt, investments, loans advanced-current, quick assets, net working capital, total income, operating expenses, and turnover. Table 6 summarizes the position of the sample units under this model. It is significant that neither the classification nor the rankings of sample units changed with the elimination of four variables. Validity of the Scale The validity of the scale is established by the known group method. For this purpose, we selected two known good societies and two known poor societies functioning in Udupi Taluka which are not included in the study. On the basis of the financial data for the year ending March 31, 1994, we calculated the total score of the four societies for nine variables and applied the scale. We found that the scale correctly classified the societies into good and poor. Therefore, we can safely conclude that the scale is perfectly valid. The scores of the variables of the four societies and the classification are shown in Table 7. Discriminant Analysis Discriminant analysis was carried out to find out how well the variables included in constructing the scale discriminate between good societies and poorly performing societies. The nine variables included in the scale were used for discriminant analysis. The standardized canonical discriminant coefficients were obtained to determine which variables contributed most to classify the sample units into good, satisfactory, and poor. The magnitudes of the standardized coefficients (ignoring the sign) are only considered. The larger the magnitude, the greater is the contribution of that Table 4: Total Score and Classification of the Sample Units Sample Units Total Score Classification Byndoor Uppinangadi Khambadakone 90 3 Shankaranarayana 86 3 Panaje 83 3 Vandse 76 2 Kumbra 57 2 Kavu 54 1 Ullur 44 1 Belenele 46 1 Molahalli 42 1 Nellyadi 40 1 Narimogaru 38 1 Kollur 36 1 Mean = 66.57; s.d. = Score 3-Point Scale < > 79 3 Scale Value 48 Vikalpa

5 Table 5: Results of the Mann-Whitney U Test Variables Share Capital Share Capital plus Reserves Deposits Total Debt Investments Loans to Members (Outstanding) Loans Advanced (Current) Quick Assets Net Working Capital Net Capital Employed Total Income Operating Expenses Turnover Critical values of the Mann-Whitney Wilcoxon test statistic: m = 5 n = 7 Significance Values U Score 34 Table 6: Classification of the Sample Units under Three-Point Scale Sample Units Total Score Classification Byndoor 81 3 Uppinangadi 79 3 Khambadakone 61 3 Shankaranarayana 61 3 Panaje 60 3 Vandse 54 2 Kumbra 40 2 Kavu 39 1 Ullur 32 1 Belenele 31 1 Molahalli 30 1 Nellyadi 30 1 Narimogaru 28 1 Kollur 26 1 Mean = 46.57; s.d. = Score 3-Point Scale Scale Value < 39 1 Poor Satisfactory > 55 3 Good Table 7: Scores of Variables of Societies (Rs in lakh) Societies Share Capital Total Investments plus Reserves Debt Loans Advanced Variables Quick Assets Net Working Capital Total Income Operating Expenses Turnover Brahmavar (10) (10) (10) (10) (10) (10) (10) (1) (10) 81 Good Ganapathi Good (7) (8) (10) (10) (10) (10) (10) (10) (10) Shirva (2) (2) (2) (3) (2) (4) (8) (3) (4) 30 Poor Belle (1) (2) (2) (2) (2) (2) (3) (8) (2) 24 Poor Total Score Note: Figures in parenthesis indicate the score of the variables. Source: Annual Accounts variable. Table 8 gives the values of the standardized canonical discriminant function coefficients. Turnover, loans advanced, total debt, total income, investments, and quick assets contributed most to classify the societies into good, satisfactory, and poor. The discriminant analysis also shows that the classification of sample units made by the scale is very accurate. Canonical discriminant functions evaluated at group means and the classification results are shown in Table 9. Table 9 indicates that the societies classified as good, satisfactory, and poor by the scale are also classified as good, satisfactory, and poor by the discriminant analysis. Cluster Analysis Cluster analysis was done to find groups of similar societies on the basis of a set of variables. The analysis was made with 18 variables selected in the first instance and finally with nine variables used in the Vol. 24, No. 1, January - March

6 Table 8: Standardized Canonical Discriminant Function Coefficients Variables Function 1 Function 2 construction of the scale. Cluster membership of societies is shown in Table 10. Turnover Loans Advanced Total Debt Total Income Investments Quick Assets Share Capital plus Reserves Net Working Capital Group Function 1 Function Table 10 indicates that groupings of societies with 18 and 9 variables are almost same. We can, therefore, conclude that the reduction of variables from 18 to 9 has not altered significantly the classification of societies. When we compare these classif ication with that made on the basis of the scores under scaling technique (Table 6), we find that cluster analysis also endorses the validity of the scale. Table 9: Canonical Discriminant Functions Evaluated at Group Means (Group Centroids) Highest Probability Group 2nd Discriminant Sample Units Actual Group P(D/G) P(G/D) Group ghest P(G/D) Scores * » Note: Sample units are serially arranged as in Table 8. Classification Results Actual Group No. of Cases Predicted Group Membership Group 1 Poor Group 2 Satisfactory % 0.0% 2.0% 0 Group 3 Good 7.0% % 0.0% 5.0%.0% 100.0% Per cent of "grouped" cases correctly classified: % 50 Vikalpa

7 Table 10: Cluster Membership of the Sample Units Using Average Linkage (B etween Groups) Sample Units Number of Clusters with 18 Variables Number of Clusters with 9 Variables Byndoor Uppinangadi Khambadakone Shankaranarayana Panaje Vandse Kumbra Kavu Ullur Belenele Molahalli Nellyadi Narimogaru Kollur Conclusions The model developed here is powerful enough to measure the efficiency of PACSs, because it incorporates variables that indicate different dimensions of development of a business. Therefore, the total score of a society represents its overall financial performance. There is a possibility in this model that a society may be found to be good according to the scale but incurring a loss as is the case with our own sample units Khambadakone and Panaje. The inference we can draw is that the society is potential enough to be highly profitable but the problem lies with the deployment of funds profitably on different assets. As a result, the yield on assets is not sufficient to cover the fixed cost. This calls for some adjustments in the funds deployed on different assets so that the situation can be improved (Dongre and Swamy, 1997). Performance evaluation that the societies make every year with the help of this scale will enable them to know the direction in which they are moving. Considering the low level of education and technical skill of the staff and board members, this scale is simple enough to understand and apply when compared to evaluation models based on ratios. Moreover, all balance sheet ratios (liquidity and solvency ratios) are less reliable as indicators of health of a business unit (Gupta, 1983, 1991). One limitation of this model is that the reasons for poor performance of societies will not be known automatically. For this purpose, services of financial experts may have to be used. This scale can be used in other districts where the nature and volume of business are almost similar to those prevailing in Dakshina Kannada. As a precaution, the scale can be first applied to a known good and a known poor soc iety in a region. If the scale classifies the good and poor societies as they are, the scale can be applied to other societies in the region. If necessary, a new scale can be constructed easily by using the methodology suggested in this paper. This scale is meant for evaluating the PACSs for the next five years after which the scale may have to be revised in view of the changing business conditions. Vol. 24, No. 1, January - March

8 Exhibit 1: Application of the Scale Application of the scale developed for this study involves the following stages: 1. Finding out the rupee value (in lakh of rupees rounded off to two decimal places) of the following variables drawn from the audited accounts: Share capital plus reserves (accumulated losses not to be deducted) Total debts outstanding (loans borrowed and de posits) Investments (long-and short-term) Loans advanced to members current (long-and short-term) Net working capital (current assets minus current liabilities) Quick assets (current assets minus inventory minus overdue Total income (sales plus interest plus other income) Operating expenses (cost of goods sold plus employees cost plus administration expenses plus interest paid) Turnover (sales plus loans recovered during the year) 2. Finding out the standardized score for all the above variables with the following formula: Mean and standard deviations of the above variables References Altman, E I (1968). "Financial Ratios, Discriminant Analysis and the Prediction of Corporate Bankruptcy," Journal of Finance, September, pp Desai, B M and Namboodiri, N V (1991). "Comparative Performance of Farmers' Service Societies and Primary Agricultural Cooperative Credit Societies," Vikalpa, Vol 16, No 4, October-December, pp Dongre, Yashavantha and Narayana Swamy, M V (1997). "Primary Agricultural Credit Societies: A Profitability Analysis," Maharashtra Cooperative Quarterly, Vol 80, No 3, January-March, pp Gupta, L C (1983). "Financial Ratios for Signalling Corporate FeUure," Chartered Accountant, Vol 31, No 10, April, pp Gupta, L C (1991). Corporate Financial Health: Building Reliable Ratio Indicators, New Delhi: Manas Publications. are shown in Table 1. Converting the standard scores into 10-point scale values using standard normal distribution as follows: Standardized Score Scale Value < = (-0.84) (-0.52) (-0.25) > = Give the scale value of operating expenses in the reverse direction. Finding out the total score and classifying as follows: Under 3-Point Scale Under 5-Point Scale Score Scale Score Scale < = 39 1 Poor < = 31 1 Very Poor Satisfactory Poor > = 55 3 Good Satisfactory Good > = 63 5 Very Good Panigrahi, D and Mishra, D P (1993). "Predicting Corporate Sickness Using Cash Flow Analysis," Vikalpa, Vol 18, No 3, July-September, pp Rao, Aruna K; Bhat, A R S and Bhatta, Ramachandra (1991). "A Methodology to Measure the Efficiency of Business Organizations," Indian Cooperative Review, Vol 28, No 3, July, pp Reddy, C R (1992). "Monitoring and Evaluation of Cooperatives," Indian Cooperative Review, Vol 29, No 3, January, pp Rohatgi, V K (1986). An Introduction to Probability Theory and Mathematical Statistics, New Delhi: Wiley Eastern. Sidhu, J S and Sidhu, R S (1990). "Case Studies of Successful and Unsuccessful Primary Cooperative Service Society and Milk Producers' Cooperative Society in Punjab," Indian Journal of Agricultural 52 Vikalpa

9 Economics, Vol 45, No 3, July-September, pp Sriram, M S (1991). "A Study of the Indicators of Sickness in Rural Cooperatives," unpublished dissertation,, Bangalore. Suresh, K A and Vinaikumar, E (1993)."Economic Viability of Cooperatives: A Case Study of Primary 'Agricultural Credit Cooperatives in Kerala," Cooperative Perspective, Vol 28, No 1, April-June, pp STATEMENT OF OWNERSHIP OF VIKALPA The following statement about ownership and other particulars of Vikalpa: The Journal for Decision Makers is published in accordance with Rule 8 of Newspapers (Central) Rules, FORM IV 1. Place of publication 2. Periodicity 3. Name of Printer Nationality Address 4. Name of Publisher Nationality Address 5. Name of Editor Nationality Address 6. Names and addresses of individuals who own the newspaper and partners or shareholders holding more than one per cent of the total capital Ahmedabad Quarterly P S Seshadri Indian Ahmedabad P S Seshadri Indian Ahmedabad K V Ramani Indian Ahmedabad Ahmedabad Campus, Ahmedabad I, P S Seshadri, hereby declare that the particulars given above are true to the best of my knowledge and belief. March 1, 1999 (sd.) P S Seshadri Publisher Vol. 24, No. 1, January - March