Keywords: Average Returns, Standard Deviation, Fund Beta, Treynor, Sharpe, Jensen and Fama s Ratio, least square model, perception modeling

Transcription

1 MULTILAYER PERCEPTION MODELING AND PERFORMANCE MEASURES: MUTUAL FUND PATTERNS ON FDI WITH SPECIAL REFERENCE TO INDIAN EQUITY FUNDS Jothi Basu T.* and Dr. Kavitha Shanmugam** Centre for Research and Development, Bharathiar University, Coimbatore *SSN College of Engineering, Kalavakkam **Assistant Professor (Sl. Gr.), Department of Management Studies, Easwari Engineering College, Chennai ABSTRACT This study has been undertaken to evaluate the risk adjusted performance of the Indian Mutual Funds with special reference to Diversified equity mutual funds vis-à-vis the Indian stock market (i.e.) S&P CNX NIFTY on FDI. For the purpose of this study, 276 diversified equity mutual funds were selected as the sample. The data, which is the monthly NAV s of the funds and the closing of the S&P CNX NIFTY on FDI, were collected for a period of 8 years starting April 2009 to March Different statistical tools were used on the data obtained to get the average returns, standard deviation, Fund Beta, Risk Adjusted return. Mutual fund performance was also analyzed through performance measurement ratios (i.e.) Treynor, Sharpe, Jensen s Alpha, Fama s ratio and Sortino ratio to evaluate mutual funds. These risk adjusted performance variables of the funds were compared with the same variables of the market to assess performance against the stock market in India. All the mutual funds gave similar returns with respect to the market and there is a positive correlation with the returns of the market and the mutual funds over the period of time. The fund betas also show that there is significant correlation between the fund returns and the market returns. There is significant relationship between the 3 months returns of each schemes and the return from inception. Risk adjusted performance has interdependence with one another proved through perception modeling. Keywords: Average Returns, Standard Deviation, Fund Beta, Treynor, Sharpe, Jensen and Fama s Ratio, least square model, perception modeling 76

2 1. INTRODUCTION SAMZODHANA Journal of Management Research The mutual fund industry of India came into existence in 1963 with the establishment of Unit Trust of India, joint actions by the Government of India and the Reserve Bank of India. The next two decades from 1986 to 1993 can be termed at the time of public sector funds with entry of new public sector players into the mutual fund industry, Unit trust of India, Life Insurance Corporation of India, State Bank of India, IDBI, Etc. The year of 1993 marked the beginning of an Indian mutual fund industry with the entry of private players like Morgan Stanley, J.P Morgan, and Capital International. The rules and Regulation were framed for first time in Indian under the name of SEBI. AMFI (Association of Mutual Funds India) was created in year 1995 as a nonprofit organization. AMFI provided basement for mutual funds function as a professional and healthy manner to protect the interest of the Investors/Unit Holders. As of now, all the 44 India s operating Mutual fund companies are registered in AFMI. The asset under management by the AMC was billion on March 2016 with 1,884 schemes operated in India. The main objective of the AMFI is to define and maintain professional and ethical standards in all areas of operations and risk management of mutual fund industry and to protect the interest of investors/unit Holders. The main code of Ethics is ACE for short, to sets out the standards of good practices to be followed by the Asset Management Companies in their operations and in their dealings with investors, intermediaries and the public. SEBI (Mutual Funds) Regulation 1996 requires all Asset Management Companies and Trustees to abide by the Code of conduct as specified in the Fifth Schedule to the Regulation. 2. NEED AND SIGNIFICANCE OF THE STUDY After 1992 the liberalization of economic policies has led and created a rapid growth in financial sector. Efforts made by the Government on economic policies create financial sector, more vibrant and the birth of intense portfolio in mutual fund industry. It creates an important player in the capital market segment to increase portfolio and diversified funds towards investor s money and fund management. With various development in the capital market, the mutual fund industry gains confidence among investors/unit holder with the professionalism in managing money, entry and exit strategy, and performance appraisal. For last 20 years, Assets are 77

3 managed professionally and ethical to maintain tradeoff between risk and return of the schemes and maintain the high level of liquidity to capitalize the redemption pressure from the investors and unit holders. Around 1880 schemes, valued at various portfolio and diversification in managing money in entry and exit strategy. A net inflow received from the investor is over Rs. 1 trillion in equity funds helped the industry grow its average asset under management 14% as on March But equity fund average AUM first fall in 10 quarters owing to mark-to-market losses as the underlying asset class lost 2.62%. As the situation in worldwide, made mutual funds in India to tremble long term maturity debt funds fail to make a mark in AUM. Instead the FMP s log second consecutive growth in last two quarter. Mutual fund used various combinations in product innovation, distribution channel and technological advancement to stay in the competition. The schemes strategy made funds to set up market timing and persistence in managing money. The wide variety of knowledge pool among fund managers, lead to managing and perform in various ways. Innovation in the product diversification enabled mutual fund managers to increase their share in the capital market segment in India. In the year 2015, some 27 lakhs invested in equity funds. It is also notable that institutions investment is negligible but the retail investor invested in the funds. 3. OBJECTIVES OF THE STUDY The idea behind this project is to provide a better insight into the evaluation, financial performance and the returns provided by the various Diversified equity mutual fund schemes and the risk levels at which they are delivered in comparison with the market and the risk free rates during April 2009 to March 2016.It is also my aim to identify the out-performers. The specific objectives are:- To estimate the various returns (1 Week, 3 Weeks, 1 Month, 3 Months, 5 Months, 1 Year, 3 Years, 5 years, and since inception) Comparing the risk adjusted performance of diversified equity mutual funds vis-à-vis the BSE Sensex the benchmark index (Bombay Stock Exchange) using Sharpe, Treynor s ratio, FAMA, Jensen s measure and Sortino ratio. 78

4 SAMZODHANA Journal of Management Research To examine the risk adjusted performance of diversified equity schemes measurement through perception modeling and partial least squares 4. DATA COLLECTION Table 1 - Data Collection The Universe Data Type Data Source Sampling Frame Sampling unit Sampling size The Fund available in the mutual fund industry in India Quantitative data ACE Mutual fund database, AMFI Website, ACE Equity and RBI Web Portal Funds based on the diversified equity orientated Diversified equity funds 276 diversified equity funds in the Indian mutual fund market 4. SCOPE OF THE STUDY The primary scope of this project is to compare the performance of the selected diversified mutual funds for the period from 2009 to 2016, vies-a-vies the benchmark index, S&P CNX NIFTY of NSE (National Stock Exchange of India). It is descriptive in nature, drawing sources of information from secondary data. The performance of diversified mutual funds were analyzed using different statistical tools on the data obtained to get the average returns, standard deviation, Fund Beta, and Relative Performance Index. Mutual fund performance is also analyzed through performance measurement ratios (i.e.) Treynor, Sharpe, Jensen ratio and Fama s ratio to evaluate mutual funds and ranked accordingly. Portfolio performance measures flexibility of combining risk and return performance into a single value. While Treynor measures through systematic risk by beta, Sharpe ratio works on total risk of the mutual fund. Jensen alpha measures the performance as the excess return provided by the portfolio over the expected Capital Asset Pricing Model returns. FAMA is to measure fund performance through excess returns over expected returns based on premium for total risk equated with it. These performance measures of the funds were compared with the same variables of the market to assess how the different funds have performed against the market. 79

5 5. PREVIOUS STUDIES RELATED TO THE RESEARCH D.R Joshi in his book entitled, Mutual Funds and Hedge Funds, explained the concept of Risk management through market with various schemes. Therefore it creates difficulty for investors to choose the suitable scheme providing higher earnings. In this context, evaluation of mutual funds has become essential and Ramesh chander in his book entitled, Performance Appraisal of Mutual Funds in India, studied the performance of selected mutual funds in terms of risk and return on the basis of fund characteristics. Besides, the study examined the portfolio construction, portfolio management, portfolio evaluation, disclosure practices and investor services. The researcher concluded that in terms of average returns, majority of the sample mutual fund schemes have recorded superior performance. Bhattacharyya, T. (2016) suggested mutual fund is an investment vehicle which well known for diversification of risk. The core of mutual fund lies in the basket of securities in which the corpus of a fund is invested. Professionals are employed to minimise the risk at an expected level of return. The level of risk of a scheme depends on the securities in which the corpus is invested. The present study attempted to access the risk associated with 25 selected equity diversified mutual fund schemes from five different fund houses. These fund houses are the leading players of the Indian mutual fund industry in terms of Asset Under Management. Dr.Rao.D.N (2015) had one more study that talked about a 4-step model for selecting the right equity fund and illustrated the same in the context of equity mutual funds in Saudi Arabia. The 4 step model was as follows: 1. Compare returns across funds within the same category. 2. Compare fund returns with the returns of benchmark index. 3. Compare against the fund s own performance. 4. Risk-related parameters: as indicated by the Standard Deviation (SD) and risk-adjusted returns as calculated by the Sharpe Ratio (SR). Suraiya Jabin (2014) International Journal of Computer Applications ( ) Volume 99- No.9, August This paper presents computational approach for stock market prediction. 80

6 Artificial Neural Network (ANN) forms a useful tool in predicting price movement of a particular stock. In the short term, the pricing relationship between the elements of a sector holds firmly. An ANN can learn this pricing relationship to high degree of accuracy and be deployed to generate profits with sufficiently large amounts of data, preferably in times of low volatility and over a short time period. Experimental results are presented to demonstrate the performance of the proposed system. The paper also aims to suggest about training algorithms and training parameters that must be chosen in order to fit time series kind of complicated data to a neural network model. The proposed model succeeded in prediction of the trends of stock market with 100% prediction accuracy. Dhanda, S. K., Batra, G. S., & Anjum, B. (2012) explained mmutual Fund is a vehicle for small investors to enter into Blue-chip companies. Mutual fund companies collect the savings of the investors and make a big corpus of these savings and invested in a well diversified portfolio of different companies. It is generally believed that mutual funds are able to diversify the risk. Mutual fund industry has just four decades old in India. During this short span of time it has made tremendous growth. So considering these points this paper is an attempt to study the performance evaluation of selected open ended schemes in terms of risk and return relationship. For this rate of return method, Beta, Standard Deviation, Sharpe ratio and Treynor ratio has been used.bse-30 has been used as a benchmark to study the performance of mutual funds in India. The study period has been taken from 1st April 2009 to 31st March 2011.The findings of the study reveal that only three schemes have performed better than benchmark. Kumar, R. (2012) study is based on the monthly data of 28 equity diversified Indian fund schemes for the period from January 2007 to June The selected equity diversified fund schemes show mixed performance. About 60% of the fund schemes were able to beat the benchmark markets. Better performing fund schemes were exposed to higher risk but were less afflicted to market risks. All the schemes under study were relatively exposed to less risk than the market, however with high degree of volatility. A majority of the funds were reasonably diversified and reduced the unique risk. Consequently, unique risks and the returns were negatively associated. The study also exposes that about 58% of fund schemes were capable of beating the market by stock selection skills. So far as market timing is concerned, the fund 81

7 managers almost failed both to book the profits in the up market and accumulate the stock in the down market. Mohamed M. Mostafa (2010) Expert systems with application, Volume 37, Issue 9, September 2010, pages , Financial time series are very complex and dynamic as they are characterized by extreme volatility. The major aim of this research is to forecast the Kuwait stock exchange (KSE) closing price movements using data for the period Two neural network architectures: multi-layer perceptron (MLP) neural networks and generalized regression neural networks are used to predict the KSE closing price movements. The results of this study show that neuro-computational models are useful tools in forecasting stock exchange movements in emerging markets. These results also indicate that the quasi-newton training algorithm produces less forecasting errors compared to other training algorithms. 6. METHODS Descriptive and Analytical research is used in this study because it will ensure the minimization of bias and maximization of reliability of data collected. The researcher had to use fact and information already available through mutual fund financial statements of earlier years and analyze these to make critical evaluation of the available material. For the purpose of this study, out of 60 fund houses available in India, 20 Funds across 5 fund houses have been selected. On the basis of the highest AUM (assets under management); these 5 fund houses were selected. All the funds selected for the study are open-ended equity funds under the growth option. The Net Asset Values (NAV) for all the 21 funds are from April 2009 to March 2016, which is the period of this study. Since, all these are equity funds, the S&P CNX NIFTY (National Stock Exchange of India Sensitive Index); which is the most widely and commonly used benchmark index in India; has been considered as the benchmark index. 7. TOOLS USED The performance evaluation of selected mutual funds was evaluated by Treynor Ratio, Sharpe Ratio, Jensen Measure, Fama s Ratio, and Sortino Ratio. Different statistical tools were used on the data to obtain the average returns, standard deviation, Fund Beta, Normality, neural 82

8 networking perception model and method of partial least squares. These variables of the funds were compared with the same variables of the market to assess how the different funds have performed against the market. 8. DATA VALIDATION (H0): Fitting of a model is good for the data. (H1): Fitting of a model is not good for the data. Figure - 1: Partial Least Squares - Performance Calculation of Mutual Fund Patterns with Special Reference to Indian Diversified Equity Funds (Source: Secondary Data/ ACE Mutual Fund& AMFI/ Software Used: SmartPLS 3.0) Step 1:- Variables for multilayer perception modeling and performance calculation of mutual fund patterns are listed; in our research work. Step 2:- From the variables identified in step 1, contextual relationship among the variables with respect to different pairs of variables are examined. Step 3:- The explained and assumed variables are theoretically examined and fitted with a model. Step 4:- A Structural Equation diagram is developed for variables, which indicates pair wise relationship among variables of the system under consideration. 83

9 Step 5:- Model fit is developed from the Structural Equation diagram and the Default model summary is checked for multilayer perception modeling and performance calculation of mutual fund patterns. The independent variables are very much necessity for multilayer perception modeling and performance calculation of mutual fund patterns. Step 6:- Based on the contextual relationships in the reach ability matrix, a directed graph is drawn and the transitive links are removed. Step 7:- The resultant diagraph is converted into an Interpretive Structural Model by replacing variable nodes with statements Table 2 - Discriminate Validity Treynor Sharpe Jensen FAMA Sortino Treynor 1 Sharpe Jensen FAMA Sortino From the above table, it s inferred that all arrived P value is greater than 0.05 (greater that 5 % level significant). So the result of the structural equation modeling for the above independent and dependent variables are indicates that fitting of the variable model is good for the data. 9. ANALYSIS AND INTERPRETATIONS The Neural Network architecture, used in this study, is a multilayer feed forward network using SPSS 20.0 the architecture which provides the best fit for the data is the network with three hidden layers and an output layer. The learning and momentum parameters are 0.6 and 0.9 respectively and error convergence falls below 0.01 Percent. Tan sigmoid is the activation function chosen for the hidden layers, and the pure linear function is used to get the output layer which is the real time values. The architecture which provides the best fit for the data is the network with eight input layers, nineteen covariate variables and one hidden layers and one 84

10 output layer, as shown.the neural network model stems from the studies on the working of human brain systems, and serves as an associative memory between the input and output patterns. These models contain many densely interconnected elements called Neurons or Nodes. The neuron has a set of n inputs x j, where the subscript j takes a value from 1 to n and indicates the source of the input signal. Each input x j is weighted before reaching the main body of the processing elements, by the connection strength or weight factor wj. (Multiplied by wj ). In addition, it has a bias term w 0, a threshold value that has to be reached or exceeded for the neuron to produce a signal, a non-linearity function F that acts on the produced signal (or activation) R, and an output O. The non-linearity function used in this network is the sigmoid. The sigmoid is very popular because it is monotonic, is bounded, and has a derivative: f (s) = kf (s) [1-f(s)]. The model used in this work is the Feed Forward Multilayer perception, using the Back Propagation Algorithm. Where (4-3-1) 8 - Input layers : 19 - Covariates layers 1 - Hidden layers : 1 - Output layer All inputs are analyzed in the experimental validation part, with appropriate output results by the illustration of graphs so that the influences of the parameters of tensile strength are taken into consideration. The network information presented in the table, the validation of the estimated NN and Experimental value illustrations is shown in Figure. 85

11 Table 3 - Multilayer Perception Modeling and Performance Calculation Training Testing Relative Stopping Rule Used Training Time Relative Dependent Variable: FAMA Training Testing Relative Stopping Rule Used Training Time Relative Dependent Variable: Sharpe Training Testing Relative Stopping Rule Used Training Time Relative Dependent Variable: Jenson Training Testing Relative Stopping Rule Used Training Time Training Testing Relative Stopping Rule Used Training Time

12 Relative Dependent Variable: Sortino SAMZODHANA Journal of Management Research Relative Dependent Variable: Treynor Hypothesis (H0): There is no significant relationship between 3 years diversified mutual fund returns and returns for the till inspection for Indian mutual funds (H1): There is a significant relationship between 3 years diversified mutual fund returns and returns for the till inspection for Indian mutual funds Table 4 Normality Analysis Name of Fund Mean Std. Deviation Skewness Kurtosis 3 Years Till Inception ** Level Significant at 5 percent Pearson's R Value Sig ** Figure 2-3 years Diversified Mutual fund returns and returns for the till inspection for Indian Mutual funds 3 Years Till Inception Mean Std. Deviation 87

13 10. CONCLUSIONS SAMZODHANA Journal of Management Research The objective of this study was to evaluate the performance of Indian diversified Mutual Fund Schemes through Treynor s ratio, Sharpe s ratio, Jensen s ratio, Sortino Ratio and Fama s ratio. The conclusions are as follows: Based on the result of normality analysis, there is a significant relationship between 3 years diversified mutual fund returns and returns for the till inspection for Indian mutual funds, entire structural equation modeling for all the independent and dependent variables are indicates that fitting of the variable model is good for the data. Hence, the Indian Mutual Fund Industry pricing mechanism with empirical studies on its valuation. It also analyzes data at both the fund-manager and fund-investor levels. The study revealed that the performance is affected by the saving and investment habits of the people and the second side the confidence and loyalty of the fund Manager and rewards affects the performance of the mutual fund industry in India. REFERENCES Agrawal, D. (2011). Measuring performance of Indian mutual funds. Finance India, June. Berk, J. B., & Van Binsbergen, J. H. (2015). Measuring skill in the mutual fund industry. Journal of Financial Economics, 118(1), Bhattacharyya, T. (2016). Risk and Return Profile Analysis of Selected Mutual Fund Product of Indian Mutual Fund Industry. Available at SSRN Deb, S. G., Banerjee, A., &Chakrabarti, B. B. (2007, December). Performance of Indian equity mutual funds vis-a vis their style benchmarks: an empirical exploration. In 10th Capital Markets Conference, Indian Institute of Capital Markets. Dhanda, S. K., Batra, G. S., & Anjum, B. (2012). Performance evaluation of selected open ended mutual funds in India. International Journal of Marketing, Financial Services & Management Research, 1(1). Duggimpudi, R., Abdou, H. A. H., & Zaki, M. (2010). An evaluation of equity diversified mutual funds: the case of the Indian market. Investment Management and Financial Innovations, 7(4), Kostovetsky, L. (2016). Whom Do You Trust?: Investor-Advisor Relationships and Mutual Fund Flows. Review of Financial Studies, 29(4),

14 SAMZODHANA Journal of Management Research Kumar, R. (2012). Market timing, selectivity and mutual fund performance: an empirical investigation of selective equity diversified schemes in India.IUP Journal of Financial Economics, 10(1), 62. Ma, L., Tang, Y., & Gómez, J. P. (2016). Portfolio manager compensation in the US mutual fund industry. Finance Down Under 2014 Building on the Best from the Cellars of Finance. Panwar, S., &Madhumathi, R. (2006, February). Characteristics and performance evaluation of selected mutual funds in India. In Indian Institute of Capital Markets 9th Capital Markets Conference Paper. Pool, V. K., Sialm, C., & Stefanescu, I. (2016). It Pays to Set the Menu: Mutual Fund Investment Options in 401 (k) Plans. The Journal of Finance. Prajapati, K. P., & Patel, M. K. (2012). Comparative study on performance evaluation of mutual fund schemes of Indian companies. Researchers World,3(3),