You are on page 1of 42

Contemporary Issues and Ideas in Social Sciences August 2007

A Behavioral Model For Stock Prices
N. Lalitha1 Department of Economics, Shyama Prasad Mukherjee College, University of Delhi, New Delhi 110026, India D. N. Rao Centre for Economic Studies and Planning, School of Social Sciences, Jawaharlal Nehru University, New Delhi 110067, India

Abstract Multi-factor models have been popularly used to explain asset market behavior. The Fama and French three-factor model fitted on the sample set of new economy stocks for the study period of late nineties to early 2000s, however, fails to give adequate explanation of the stock market behavior. A behavioral model built on the assumption of bounded rationality and biases in investor behavior seems to offer a better explanation of the stock price behavior. JEL Classification: G12

1

For Correspondence: Email: lalitha nat@yahoo.co.in

2

CIISS August 2007

INTRODUCTION Tracing stock price movements is a difficult exercise that involves detailed examination of firm specific and industry specific factors. It is now widely recognized that Sharpe (1964) and Lintner (1965) and Mossin’s (1966) Capital Asset Pricing Model (CAPM) and its other variants do not adequately describe the cross section of expected returns. Investors have been found to earn abnormal returns in excess of those predicted by the model. We therefore developed a simple model for stock price movement that incorporates investor behavior pattern. By examining the stock price data of a sample set of technology, media and telecom (TMT) companies, listed in the Bombay Stock Exchange, during the late 90’s and early 2000 period we identified inertia in investor behavior as an important factor in generating market fads. The hype associated with all sorts of dot-com and media companies that pushed stock markets worldwide during the later part of nineties and early two thousand poses an interesting question as to whether such boom in technology and media stocks was justified by fundamentals or was it a mere fad? Using the behavioral inertia approach, the present study tries to show that the market was indeed driven by the craze surrounding such stocks. The paper is divided into three broad sections. In section 1 we introduce asset pricing models. A brief description of Fama and French three factor model is followed by its empirical invalidation using the sample data under consideration. In section 2 the drawbacks of the neoclassical approach -the framework within which most asset pricing models are constructed– is highlighted. A behavioral stock price model that incorporates investor behavior pattern and is less restrictive in assumptions has been proposed. The result of an empirical test of the model is also examined in this section. In section 3 a comparative evaluation of the two models is done by looking at their forecasting power.

1

Asset Pricing Models

The Capital Asset Pricing Model (CAPM) is one of the most extensively studied models both theoretically and empirically for valuation of securities. The CAPM is a theory about the way stocks are priced in relation to their risk. The underlying logic of the theory is that assets with the same risk should earn the same expected returns. The CAPM puts forward the idea that in market equilibrium assets earn premia over the riskless rate that increases with their risk, where the determining influence on risk premia is the covariance between the asset and the so called market portfolio- the ‘β’ of the asset, rather than the own or intrinsic risk of an asset. Various versions of CAPM are available in the literature which were built by relaxing some of the stringent assumptions of the original model. However, starting in the late 1970s empirical work appeared that challenged even the robust version of the CAPM. Fama and French (1992) made a devastating blow to CAPM when they reported that there seemed to be no connection between beta and returns. Fama and French argued that the higher average return on small stocks and high book-to-market value stocks reflect unidentified state variables that produce undiversifiable risks (covariances) in returns that are not captured by market returns and are priced separately from market betas. Fama and French (1996) therefore formulated an extension of the CAPM model that describes asset

Lalitha & Rao: Stock Price Model

3

return as a function of three different sources of risk. This model explains many of the CAPM average return anomalies. In the three-factor model developed by Fama and French, size and ratio of book value of common equity to market equity (BE/ME) are used as proxy for sensitivity to common risk factors in returns. The model described expected return on a portfolio in excess of the risk-free rate as a function of: (i) excess return on a broad market portfolio Rm − Rf ; (ii) the difference between the return on a portfolio of small stocks and the return on a portfolio of large stocks (SMB); and (iii) the difference between the return on a portfolio of high BE/ME stocks and the return on a portfolio of low BE/ME stocks (HML). In other words, the expected return on portfolio i is

E(Ri ) − Rf = bi [E(RM ) − Rf ] + si E(SM B) + hi E(HM L)........(1)

where E(RM ) − Rf , E(SMB), E(HML) are expected premiums and the factor sensitivities and bi , si and hi are the slopes in the time series regression

Ri − Rf = ai + bi (RM − Rf ) + si SM B + hi HM L + εi ........(2)

The study showed that the three-factor model gives a good description of returns in portfolios formed on earnings/price (EPS/P), cash flow/price and sales growth in addition to explaining returns on portfolios formed on size, BE/ME and industry returns.

1.1

Empirical estimation of Fama and French three-factor model

The study period covered was April 1999 to November 2004. From the TMT stocks underlying the various BSE and NSE indices only 35 companies were found to have a continuous price data and information on economic fundamentals of the company for the period covered under the study. These, therefore, formed the sample for the present study. In general, stock market studies are done using portfolios rather than stock returns. Holding a diversified portfolio is assumed to take care of firm specific risk. In this study portfolios were constructed using various criteria (see Appendix 1). In the Fama and French model firms with high BE/ME value are treated as weak firms experiencing distress. The risk associated with such firms

gets priced and is reflected in high returns and positive hi . On the other hand, firms with low

BE/ME values are strong firms and have negative loading on hi . Similarly, small firms have low earnings on assets and are less profitable than big firms. So size is treated as a proxy for risk factor and therefore small firms will have higher expected returns and positive slope on

4

CIISS August 2007

SMB as against big firms which have lower expected returns and negative slope SMB. After testing for stationarity (see Appendix 2) the model is estimated by running the regression equation (2) and validated by testing for Ho : ai = 0. We estimated this regression for the thirty-six constructed portfolios (see Table 1 and Table 2). A detailed analysis of Table 1 showed that: 1. For the six equal weighted size-BE/ME portfolios, the Ho : ai = 0 is accepted. The null, however, does not hold for value weighted size-BE/ME (except SMBEME) as also for single sort BE/ME (both equal and value weighted) portfolios evaluated at 5% level of significance. 2. As compared to value weighted portfolios, equal weighted portfolios seem to have both better explanatory power, as also greater sensitivity to size and BE/ME factors. 3. Slope on SMB is high and strong in the small category for both equal and value weighted portfolios. In the big group SMB is not only small in magnitude but also insignificant. Also, the slope coefficient is positively related to returns in some categories. Similarly when ranking is on the basis of BE/ME alone, SMB is significant for equal weighted category and not for value weighted category. 4. Loadings on HML on all the portfolios is strongly supported in all except SMBEME (both equal and value weighted) and SHBEME value weighted stocks. In all the regressions low BE/ME stocks have negative coefficients whereas high BE/ME stocks have positive coefficients. 5. The explanatory power of regression is good when the BE/ME criteria is used alone for constructing portfolios. Combined with size the regression works well for small as against big. Equal weighted portfolios seem to do better than value weighted. Thus, the high R2 associated with the above regression indicates that there is risk related to size and BE/ME which captures common variation in returns. However, whereas the model picks up the premium associated with small size when in the first place small size is already highlighted, it fails to establish a discount in returns for big sized firms. It however, prices the risk captured by BE/ME factor along the predicted lines of the model. Thus the size premium is weaker and less reliable than the value premium.

EPS/P Criteria The regression estimates of equation (2) using portfolios constructed on the basis of EPS/P ranking also shows poor performance of the multifactor model. In both equal and value weighted double sort category for only 3 out of 6 portfolios, the three-factor model fails to reject the Ho . In the single sort category again for only portfolios LEPS and LEPS weighted the Ho is not rejected again evaluated at 5% level of significance. The slope coefficients on SMB and HML also do not conform to the three-factor model. None of the low EPS/P portfolios have a negative loading on HML. Similarly in the big size-EPS/P

not eliminate mispricing. built in the neo-classical framework. brainpower and bounded rationality individuals are often not in a position to maximize their objective function and would instead settle for a ‘satisfactory’ level of the target variable. The Fama and French model. information is no longer a constraint. Thus. it will be incorrect to assume that errors cancel out in equilibrium and therefore that the estimated values will . And since individuals have similar biases. while the trading by rational traders traces the path towards equilibrium. Persistent mispricing might also occur because some relevant piece of public information is either ignored or misused by everyone leading to market prices being regularly at odds with fundamental values. The rise in stock prices not justified by fundamentals directs us into the realm of behavioral inertia. The neo-classical approach imposes stringent assumption on the behavior of economic agents.Lalitha & Rao: Stock Price Model 5 category none have significant SMB and only 2 out of 6 big-size portfolios have negative coefficient on SMB. It is now recognized that individuals at best have imperfect rationality. the weak correlation between BE/ME and EPS/P for sample companies (a meagre 0. While it might be argued that in the modern day computerized world. trading activities of irrational traders pushes the market away from the equilibrium point. Thus the validity of the Fama and French model is dependent upon the criteria used for constructing sample portfolios as also the risk mimicking portfolios. (which argues that past relative performance predicts future relative performance) of Jegadeesh and Titman (1993). A more serious problem facing the Fama and French model is its failure to explain the momentum effect. for instance. Arbitrage will. With these modified assumptions it is no longer certain that equilibrium values would be achieved. for example. All individuals have biases especially under conditions when information is slack. Moreover even the ‘optimum’ level so identified may not be chosen for altruistic reasons. The equilibrium prices are a weighted average of the beliefs of rational and irrational traders. 2 Drawbacks of neoclassical approach The results of Fama and French model challenge us with a need to develop an alternate model for stock market behavior in the late 1990s and early 2000. its proper interpretation and analyses still remains. Arbitraging does not work efficiently. has come in for a lot of criticism. This was not surprising considering the fact that the risk mimicking portfolios SMB and HML were themselves formed on the basis of size and BE/ME ranking. the empirical exercise showed that the model performed somewhat better when portfolios were constructed using BE/ME criteria. The assumption of perfect rationality. has great theoretical appeal but limited empirical relevance. and the influence of either group on prices depends upon their risk bearing capacity. Further. In the capital market. They would therefore be able to explain return on other portfolios formed on similar ranking. therefore. The insipid performance of the model when portfolios are formed using EPS/P criteria indicates that the earnings variable does not adequately capture information regarding returns in stock prices. the question of quality of information. since it is hard for an investor to know whether other investors have yet detected and acted upon it.19) implies that earnings variable also fails to pick-up the risk associated with high BE/ME firms. With limited time.

This however. Such ‘irrationality’ in behavior creates uncertainity in all economic phenomena. does not rule out the possibility of profit making by predicting changes in the conventional basis of valuation. Moreover. Thus. They downplay the fact that these valuations can be incorrect. Keynes (1936) argued that investors accept current valuations as a correct reflection of the market assessment of future prospects. Such a possibility generates speculative behavior amongst investors. Market participants will therefore find it more ‘rational’ and practical to change their decisions only slowly even when underlying economic conditions are constantly changing. They therefore. Variations that were successful in the most recent past will tend to remain successful in the near future as well. Thus. investors find that the best forecast of the future price is in fact the current price. Proponents of Efficient Market Hypothesis (weak form) treat current price as reflecting all the information that is contained in past prices. Caprice provides a mechanism of behavioral variation. By explicitly identifying inertia and caprice. a short time ahead of the rest of the investor population.1 Behavioral inertia model for stock prices Faced with information. i. Faced with uncertainty regarding factors that might affect stock prices market participants seek safety by conforming to the behavior of the majority on the average. In the literature on finance it is common to find lagged . However. 2. The idea that the current price is the best predictor of future price has been quite popular both with lay investors as well as academicians. which promotes advancement of the society. time and expertise.6 CIISS August 2007 approximate the true values. where information is revealed through a sequence of events. the time required to converge to an equilibrium strategy can be extremely long. It also exhibits inertia or pattern in variation. Since there are some opportunity costs to learning. even a completely ‘rational’ learner will chose not to experiment and remain in a non-optimal equilibrium if the cost of trying something else is too high. the Behavioral Inertia Model proposes a dynamic theory of economic phenomena. especially in a situation of changing environment. The behavioral inertia approach has been shown to be a combination of inertia and caprice. markets can be in a situation of perpetual non-convergence. time and resource constraints and also a stock price series that exhibits random walk. Inertia produces highly auto-correlated time series in which random events have lasting effects. highlight the futility of forming trading rules based on share price history. Individuals are often overconfident about their abilities and therefore may deviate from the tried and tested path. the cost of collecting and analyzing information are exorbitant in terms of money. Neoclassical economists arguing for the existence of equilibrium had immense faith in human learning capabilities and therefore believed that individuals will not systematically and consistently make the same mistake.e. experimentation literature has shown that there can be a complete lack of learning even in infinite horizons. implying positive autocorrelations among the innovations. In an uncertain world. they would find greater returns with inaction rather than optimizing action and behavioral inertia plays the important role of imparting stability in individual’s behavior. random change (Stanley 2000).

their stock preferences and their behavioral biases jointly influence their diversification choices (Goetzmann and Kumar (2004).. More importantly. In this study. the performances of asset pricing models have been found to improve.. by working with prices rather than returns. Huberman(2001) also reported the tendency of household investments to be primarily concentrated in their employer’s stocks and in general in stocks of companies registered in their country as against foreign company stocks. Heaton(1995) used a simulated Method of Moments approach to evaluate a representative consumer asset pricing model and reported evidence for the local substitution of consumption with habit formation occurring over longer periods of time.. Therefore. Measurement choice between average monthly abnormal returns vis--vis buy and hold abnormal returns has severe effect on the outcome of the study. With returns. Kumar & Lim (2004)). Assuming inertial decay α1 Pt = α0 Pt−1 Ct . we developed a behavioral inertia model at the stock level rather than at the portfolio level.... Heaton(1989) and Sunderasan(1989) examined asset pricing on the basis of habit formation.. By incorporating such time non-separability in preferences. Polkovinchenko (2003). by working directly with prices rather than stock returns one can draw unambiguous conclusions.Lalitha & Rao: Stock Price Model 7 dependent variables when stock market analysis is in the returns framework (e. Further. our approach does not entail calculations of rationally expected returns. which in any case cannot be calculated in ex. These phenomena provide compelling evidence that people invest in the familiar stocks while often ignoring the principles of portfolio theory.g. Choice of a normal period to estimate a stock’s expected return is also problematic as stocks can show return continuation in the short run and mean reversion in the long-run. These. Detemple(1989). Investors’ personal characteristics. however. (3) If caprice also experiences the same type of exponential decay then . Constandinides(1988). The behavioral inertia model that we developed is not based on optimizing principles. we avoid the crucial question of unit of time for returns. are built in the neoclassical framework and incorporate habit persistence in consumption in the utility function. There are also studies on asset market that include time dependency in preferences in utility function. Goetzmann and Kumar (2004)). This is owing to the fact that investors rarely hold a well diversified portfolio (see Barber and Odean (2000).2 The Model We capture the inertia in stock prices and the market dynamism by formulating an asset pricing model as a function of lagged prices and Ct the caprice element.ante terms. 2. a choice between equal weighted versus value weighted returns has different implications for the behavior of the stock markets. Ours is a simple approach that allows for biases in individual behavior and treats them as incompletely rational. Fama and French (1988)).

(8) (obtained by decomposing Xit into its components) Thus.. ρ is the auto-regression coefficient for caprice. a measure of its persistence and t is the truly random irreducibly stochastic past... the testing of the model involves running the following regression ln Pit = (1 − ρi ) ln α0i + (α1i + ρi ) ln Pit−1 − ρi α1i ln Pit−2 + β1i ln X1it + β2i ln X2it + β3i ln X3it + β4i ln X4it + νt .....(7) Thus...e.. where νt = ln εit The presence of inertia is ascertained by testing for the (9) H0 : (α1i + ρi ) − ρi α1i = 1 i.... Re-writing the above equations in logarithmic form......... (5) ln Ct = β ln Xt + ρ ln Ct−1 + ln εt .. sum of AR(autoregressive) coefficients = 1 .. for each stock i...8 β ρ Ct = Xt Ct−1 t .. we get ln Pt = ln α0 + α1 ln Pt−1 + ln Ct . (4) CIISS August 2007 where Xt is 1 x k vector of explanatory variables... β is a k x 1 vector of regression coefficients.........(6) and further ln Ct−1 = ln Pt−1 − ln α0 − α1 ln Pt−2 .... ln Pit = ln α0i + α1i ln Pit−1 +β i ln Xit + ρi [ln Pit−1 − ln α0i − αi1 ln Pit−2 ] + ln εit or ln Pit = ln α0i + (α1i + ρi ) ln Pit−1 +β i ln Xit − ρi ln α0i − ρi α1i ln Pit−2 + ln εit or ln Pit = (1 − ρi ) ln α0i + (α1i + ρi ) ln Pit−1 − ρi α1i ln Pit−2 + β1i ln X1it + β2i ln X2it + β3i ln X3it + β4i ln X4it + ln εit ....

the following regression was run. For such periods with positive beta. However.e. The results of ADF test on all the variables used in the model (given in Appendix 3) show that only for the ln price series the null of Unit root with trend and intercept is not accepted at 5% (though not rejected at 1%) for a small number of companies. Thus.3 Empirical Testing of the Model In our exercise the explanatory variables included are natural log of book-to-market value (lnB/M). Similarly. we saw it reasonable to continue including beta in the set of explanatory variables. Similarly. To take care of the problem of finding natural log for negative betas we bifurcated the beta variable into lnβ positive and lnβ negative. TataElxsi and Vindhyas the absence of trend is accepted at 1% but not at the 5% level of significance. the presence of Unit root with intercept alone is not accepted by lnprice series of Eserve. considering the fact that for long. InfoSys. ln Pt = β0 + β1 ln Pt−1 + β2 ln Pt−2 + β3 ln BM + β4 ln M V + β5 ln beta + νt For the thirteen companies which had negative betas for part of the sample period the beta dummy of lnbeta negative was included in the above regression. Eserve. Many of the stocks reported negative beta values for part of the time period under consideration. Of the 35 stocks 3 stocks namely ASM Tech. Moreover. excluding negative sign) and this is treated as variable lnβ negative. DSQ. Moser.Lalitha & Rao: Stock Price Model 9 2. for negative betas. After ascertaining the independence of the error terms (see Appendix 4) we estimated the model by running the following regression on price series of nineteen companies which had positive beta values throughout the study period. For all other series the null hypothesis of Unit root is accepted with trend alone at 5% level of significance. Here the variable lnβ positive has elements zero for the corresponding time period. beta dominated the risk-return models. the ln value of |β|is estimated (i. MTNL. the variable lnbeta negative shows the value of zero. Crest. The p values of the likelihood ratio test of the coefficient on the one time lag of the dependent variable and the trend being jointly equal to zero shows that for Afte. Hinduja and Moser at 5% level of significance. The last column of the Table gives us the ‘t’ values calculated for the null hypothesis of β1 + β2 = 1 . Out of the thirty-two sample companies thirteen had two beta variables of lnbeta and lnbeta negative. ln Pt = β0 + β1 ln Pt−1 + β2 ln Pt−2 + β3 ln BM + β4 ln M V + β5 ln beta + β6 ln betaneg + νt The results of the above regression are reported in Table 3. natural log of market value (lnMV) and natural log of beta (lnbeta). BPL and Nelco had negative BE/ME value and were therefore dropped. Hinduja. For positive beta values natural log is estimated and is recorded as variable lnbeta positive. Penasoft. Orient. the coefficient of beta was found to be insignificant in a preliminary exercise of univariate regression using cross-section data.

Finolex.10 CIISS August 2007 From Table 3 we see that the one time lagged price for all the series is strongly significant and has a positive feedback effect on the current price level whereas the series lagged by two time periods has a low slope coefficient and is also highly insignificant. From the t statistic reported in the last column of Table 3 it is clear that the null of inertia evaluated at 1% level is not accepted in the following series. major national and international collaborations and high rating of their performance by various agencies. The results of the Wald test reported in Table 4 also confirms the presence of inertia in twenty two of the thirty two sample companies. MTNL. Therefore. the two economic fundamental variables along with the beta dummies and the two lagged price variables do a very good job in modeling the price behavior. NIIT. we also evaluated the model using the Wald test. Fama (1998). In fact the null hypothesis was rejected for the same set of companies which failed the t test. This can be seen from the R2 statistic reported. Wipro and Zensar. 3 Forecasting power of the two models In the debate on the superiority of rational-behavioral approach researchers in the rational paradigm argue for their approach on the grounds of the testability and predictive power (e. characteristics tend to measure total risk. the researchers working within the rational paradigm seem to have abandoned testability and prediction in favor of a scheme of ex-post rationalizations of observed price behavior. lnbeta and lnbeta negative are significantly different from zero only for five companies. Moreover. whether Pt−2 pushes the current price in the same direction or in the opposite direction is not clear from the mixed results obtained. Contantinides (2002)). The crucial difference between the two approaches of rational and behavioral school lies in the type of risk that is quantified to describe the risk-return relationships. An examination of the company history revealed that this period was marked by issue of ADRs. But taken together. The market therefore viewed these stocks favorably and their market prices were driven away from their existing valuations. Jain. However. While factor loadings are proxies of systematic risk. . These conveyed information on the future prospects of the firm which were orthogonal to those obtained from the past performance of the firm. Again lnBM and lnMV are individually significant only for seven companies. In all the 32 cases the ‘F’ test of all the coefficients being equal to zero is rejected.Bluestar. an asymptotic test which does not require the assumption that the error term is normally distributed. Since the rational model has enormous flexibility to generate such rationalization. it is nearly always possible for the rational school to explain seemingly anomalous results even when behavioral explanations enjoy at least equal plausibility. Mphasis. TataElxsi. As the estimated β’s are measured with error and as measures of behavioral characteristics use more up to date information the latter promise to be better proxies of true unobservable betas. whereas the two beta variables. as Brav and Heaton(2002) pointed out. In the behavioral inertia approach no distributional assumptions are made about the error term except them being identically and independently distributed.g. Satyam.

The value of U lies between 0 and 1. Researchers choose estimation techniques by looking at the diagnostic statistics of the model. The Tables report two statistics. whereas a value of U=1 indicates that the model has no predictive value. characteristics could have information independent of the covariance structure of returns that helps explain expected portfolio returns. Acting on the belief that prices will be what they were. A look at mean absolute percentage error and Theil’s inequality shows that the behavioral inertia model gives a better description of price behavior in the Indian stock market. especially share price as opposed to accounting data given in the company financial reports. Agents assume the current market valuation to be a reflection of the markets assessment of future prospects. Thus the forecasting exercise clearly suggests the need to incorporate investor’s inertia in modeling stock prices. To understand the implications of these two approaches we undertook a comparative evaluation of the multifactor model and the behavioral model. significant serial correlation in estimated residuals may be due to the presence of ARCH effect in residuals or the omission of lagged dependent and explanatory . Though the two models differ in terms of the dependent variable. in-sample forecast using forecasted values for lagged dependent variables was done for the two models viz. The behavioral inertia approach involves the use of a simple regression exercise. Popular stock investment strategies are often fads based on market generated data. investors use market valuation to pick stocks. For example. the better is the forecasting ability of the model.Mean Absolute Percentage Error (MAPE) and Theil’s inequality coefficient. they can still be compared on the basis of their predictive power. Such behavior is self fulfilling and inertia becomes the basis for the market action. 4 CONCLUSIONS The focus of this study has been on developing a behavioral model for understanding asset market behavior. Complex econometric techniques though highly popular generally do not serve much purpose. MAPE is defined as 1 T T Y t − Yt t=1 Yt ∧ and Theil’s inequality coefficient U is given as 1 T T t=1 Yt − Yt 1 T T 2 t=1 Yt ∧ 2 U= 1 T ∧ ∧2 T Yt t=1 + Here. Fama and French model and the behavioral model (see Tables 5 and 6).Lalitha & Rao: Stock Price Model 11 Moreover as suggested by Daniel and Titman(1997). A dynamic. The smaller the value of MAPE. A value of U=0 implies perfect fit. It is however quite possible that what might apparently seem to be the problem may in fact not be so. where. These two measures of forecast error are invariant of the scale of the dependent variable. Yt and Yt are the simulated and the actual values of dependent variables of the two models.

yielding statistics that may have virtually unknown sampling properties. Bluestar. Finolex.12 CIISS August 2007 variables. A similar number of portfolios in the EPS/P ranking also accept the null hypothesis. By assuming inertial decay we derive a log-linear model to describe stock price behavior. Such prescreening of the model may however change the distributional properties of subsequent statistical analysis. Wipro and Zensar. Any problem in empirical testing either in the form of misspecified error terms or coefficients with wrong sign is treated as evidence against the theory under examination. formulated using rational and behavioral approaches respectively. Jain. The model that we have developed is guided by theory rather than an outcome of sample data. Satyam. The results of both the ‘t’ test and ‘Wald’ test show that the null is not accepted in 10 out of 32 sample companies. NIIT. In particular. Under the condition when no distributional assumptions are made about the returns series. The alternate model proposed in this paper based on investor behavior pattern seems to offer a better explanation of stock market behavior. in the BE/ME ranking the null hypothesis of the three-factor portfolios being minimum variance efficient is accepted only for the six size-BE/ME equal weighted portfolios. Hence a model with autocorrelation in disturbances should be developed only after testing for the above mentioned possibilities. and the small and medium BE/ME value weighted portfolios. . an asymptotic test like Wald test would be more appropriate. there is no pretest bias. The coefficients on SMB and HML also do not unambiguously reflect the pricing of risk associated with small size and high book-to-market value firms. An in-sample forecasting of the two models viz. Mphasis. Fama and French and the behavioral inertia model. We. The Fama and French model is tested using portfolios constructed on the basis of BE/ME and EPS/P ranking. namely. TataElxsi. This is in contrast to the usual approach of modifying a model to arrive at a better fit to the data in hand. Since no inductive searches for suitable econometric models are made. This latter approach is not a desirable method of model building as corrections made by looking at apparent deviations from the assumptions may or may not eliminate the root cause of the problem. also supported our argument for incorporating behavioral biases in investor’s behavior. The presence of inertia in the stock price series is ascertained by undertaking a ‘t’ test on the coefficients of the two lagged variables adding up to 1. therefore. MTNL. undertook a Wald test of the null hypothesis of presence of inertia. The results show poor performance of the model in explaining average returns on the various constructed portfolios.

1927 8.8269 1.38 7.07 14.32 0.2710 9.0087 0.1971 9.3352 0.99 0.5834 1.9427 5.68 1.0555 4.9443 HML R2 0.65 0.70 0.34 0.91 0.32 1.16 2.1438 0.14 2.1356 0.00 1.90 0.72 2.50 0.45 0.08 0.0684 0.1199 6.0128 0.55 0.3078 -0.6478 1.5893 0.1310 0.77 t-Stat SLBEME Coeff t-Stat SMBEME Coeff t-Stat SHBEME Coeff t-Stat BLBEME Coeff t-Stat BMBEME Coeff t-Stat 0.60 0.90 0.9127 4.0703 4.52 0.67 0.41 0.6397 0.73 AdjR2 F DW stat stat 0.38 22.9606 4.7965 0.8059 1.95 17.7271 1.1488 1.0196 3.25 2.63 0.77 204.34 2.2832 EXRET SMB BSE 1.61 201.15 1.20 0.77 0.9534 0.0066 0.14 0.78 0.6804 7.1125 0.0410 2.61 1.42 0.3222 0.4029 0.15 .77 74.8525 0.0192 0.0817 7.45 2.3833 0.6146 4.6610 2.68 1.82 0.63 0.7781 0.86 0.7744 0.1045 0.77 0.Lalitha & Rao: Stock Price Model 13 Table 1: Regression of excess returns to portfolios formed on the basis of BE/ME criteria on the Fama and French factor portfolios Portfolio LBEME Coeff t-Stat MBEME HBEME LBEME wtd Coeff t-Stat Coeff t-Stat Coeff t-Stat MBEME wtd Coeff t-Stat HBEME wtd Coeff C 0.06 0.8026 1.23 4.72 56.11 1.58 31.26 4.0582 4.6656 0.9225 1.0001 -0.0649 2.0191 1.29 0.0564 2.3762 1.4063 0.39 15.0202 0.24 0.28 3.3300 1.1212 5.8917 0.2469 7.23 35.26 6.0512 9.53 25.10 3.0449 8.7699 1.0697 0.71 0.72 17.91 0.87 0.6594 0.0636 5.

89 0.3710 0.0551 0.59 0.58 0.2069 6.1221 6.16 0.4951 0.96 1.61 0.1756 6.2247 1.9560 0.91 1.96 1.53 0.17 2.43 0.0551 0.51 23.40 15.2503 6.0908 0.0908 0.87 .3710 0.33 0.39 0.0802 3.1221 0.4951 0.00 1.49 20.0652 2.4573 0.04 0.04 0.7639 0.1544 0.35 5.4316 0.2132 4.33 0.1367 1.0977 1.1425 8.14 Table 1: Contd.30 4.4311 3.86 0.0745 1. Portfolio BHBEME Coeff t-Stat SHBEME Coeff wtd SLBEME Coeff wtd t-Stat SMBEME Coeff wtd t-Stat t-Stat SHBEME Coeff wtd t-Stat BLBEME Coeff wtd t-Stat BLBEME Coeff wtd t-Stat BMBEME Coeff wtd t-Stat BHBEME Coeff wtd t-Stat C 0.0062 0.60 0.53 0.2197 0.85 2.0380 0.9560 0.12 1.1905 3.50 0.2069 6.73 CIISS August 2007 AdjR2 F DW stat stat 0.0498 2.15 4.8289 2.48 0.36 13.53 0.0751 HML R2 0.16 0.48 20.5320 0.86 35.3289 EXRET SMB BSE 1.38 0.0878 0.0745 0.0939 6.72 58.0684 1.63 0.8519 5.05 0.1575 1.0578 3.68 1.7453 3.4424 2.0878 2.53 0.10 0.1575 2.97 1.9110 1.57 1.0802 3.7359 1.19 0.35 5.45 10.51 23.7963 1.50 0.7963 3.59 0.91 1.77 0.

3217 0.0486 2.26 DW stat 1.1241 0.1392 0.75 0.0109 0.0122 0.8421 1.0119 9.2907 0.52 0.54 0.0750 4.02 0.0933 4.01 1.1915 1.0736 1.8956 1.0263 1.1735 0.39 0.4333 0.3011 8.0829 3.3984 0.2144 0.65 13.74 2.29 0.43 0.0550 0. Regression of excess Returns on portfolios formed on the basis of EPS/P ranking on the Fama and French factor portfolios.0620 0.1030 6.00 1.Lalitha & Rao: Stock Price Model 15 Table 2.1305 6.2765 2.53 0.1105 1.67 0.10 0.0545 3.1485 1.8709 0.0543 0.74 .9383 3.4471 0.6818 2. Portfolio LEPS wtd MEPS wtd Coeff t-Stat Coeff t-Stat HEPS wtd LEPS wtd MEPS wtd Coeff t-Stat Coeff t-Stat Coeff t-Stat HEPS wtd SLEPS wtd SMEPS wtd SHEPS wtd Coeff t-Stat Coeff t-Stat Coeff t-Stat Coeff t-Stat BLEPS wtd Coeff t-Stat BMEPS wtd Coeff t-Stat BHEPS wtd Coeff t-Stat C 0.49 1.2634 1.34 8.1500 2.3538 0.01 Fstat 0.52 15.2821 7.67 0.63 38.1573 1.4357 0.4544 0.2194 0.9792 SMB 0.1982 1.2796 EXRET BSE 0.3501 1.5517 1.0881 0.0294 0.4147 0.6012 0.4615 0.68 Rˆ2 0.6181 1.3180 3.3384 0.00 1.1147 0.74 0.0873 1.1258 0.2011 5.5568 0.1362 2.6395 0.2859 0.73 0.3019 HML 0.9213 0.03 0.96 1.06 Adj Rˆ2 -0.7691 0.1076 0.4189 0.5227 6.2416 7.06 0.76 1.28 41.20 2.1232 0.9548 0.1668 0.76 1.1805 0.9919 5.1259 2.72 57.36 0.1154 0.65 0.0196 4.0279 0.25 0.2730 0.52 25.3969 5.3390 0.40 1.43 1.51 0.33 1.57 2.02 0.2425 4.6173 1.54 0.9372 5.0859 5.3748 0.1535 2.1161 5.4972 0.49 22.0485 2.2865 1.1250 0.26 64.8716 7.6828 0.3918 0.

Portfolio SLEPS wtd SMEPS wtd Coeff t-Stat Coeff t-Stat SHEPS wtd Coeff t-Stat BLEPS wtd BMEPS wtd Coeff t-Stat Coeff t-Stat BHEPS wtd Coeff t-Stat C 0.0901 5.20 0.71 CIISS August 2007 Fstat 6.41 0.39 5.4857 0.3178 0.0050 0.11 1.6125 0.17 14.0560 0.9308 1.0546 3.08 1.4305 SMB 0.1213 0.16 0.2230 1.5673 1.4830 0.7540 2.1057 1.24 0.97 0.20 0.1455 2.60 0.48 .38 14.1017 0.8596 0.65 2.9338 HML 0.3309 6.65 0.0021 0.9918 6.3305 0.4376 10.0511 0.9849 5.1161 0.91 2.0060 0.1721 0.1205 2.8191 1.1472 1.6194 0.0907 4.8133 5.72 Adj Rˆ2 0.8574 1.2366 3.8278 3.08 DW stat 2.0844 0.6746 0.42 0.62 55.82 31.7240 EXRET BSE 0.8993 0.5269 0.58 Rˆ2 0.1659 0.44 1.16 Table 2: Contd.2996 0.3583 2.0285 0.9350 0.

4613 6.8945 -0. Coeff S.7135 -0.0000 0. ACE ln Pt−1 ln Pt−2 ln BM ln M V ln β 17 Coeff S. Regression of ln Pt on ln Pt−1 .836702 DW = 1.032666 Prob(F) = 0 t = -3.0841 0.1410 -0.2883 R-sq = 0.1635 0.2236 R-sq = 0.0383 -0.E t-Stat Prob.1148 0.6536 0.0439 1.4928 0.0010 0.8964 -0.1836 0.1264 0.0761 1. ln M V .5055 0.1600 0.2702 1.98 BLUESTAR 0.0000 0.1367 7.9667 0.0008 R-sq =0.1321 0.1628 0.0000 0.7568 0.5440 3.1312 0.2170 0.5031 0.4692 0.1122 0.2940 0.1666 2.3939 0.5980 0.747653 DW = 1.0693 -3. Coeff S.0000 0. 1.0000 0.032666 Prob(F) = 0 t = -1.1321 0.4510 -2.7284 6.E t-Stat Prob. Coeff S.0905 0.7692 0.2578 0.0714 0.2089 0.5739 0.0000 .0947 1.663 AFTE 3.0498 0.0770 0.8106 0.0026 0.5792 0.0000 0.0000 0.8352 -0.2708 -1.2725 -0.096961 Prob(F) = 0 t = -1.2086 0.6698 0.910739 AdjR-sq = 0.0062 0.0000 0.2196 4.920026 AdjR-sq = 0.1182 0.903301 DW = 2.0032 0.4247 0.6900 0.0918 0.8489 -0.0435 0.0388 0.Lalitha & Rao: Stock Price Model Table 3.0868 3.3374 -0.E t-Stat Prob.849264 AdjR-sq = 0.7697 3.9879 -0.8073 -1.872346 DW = 2. Coeff S.8420 0.8988 -0.E t-Stat Prob.E t-Stat Prob.0823 0.0744 0.3672 0.4064 -1.5711 0.1934 0.0000 0.99 ln β Negative 0.1433 0.1237 0.3038 0.1224 0.9782 0.0019 0.1476 0.E t-Stat Prob.1324 0.1196 0.0324 0.0345 -0.1636 -0.0002 0.9830 3.0607 -0.0000 0.1416 0.0635 -0.897341 AdjR-sq = 0.0000 0.2452 13. ln BM .0966 0.9829 -1.8040 -0.1187 -0.0000 0.1195 0.0000 0.7147 R-sq = 0.0000 0.35 COSMO 1.0354 -0.1067 3.8396 -3.0379 2.1312 0.4168 0. ln β and ln βnegative C Coeff S.4230 6.0008 0.6859 0.0000 0.0000 0.0000 0.1429 R-sq = 0.0000 0.2407 -1.1347 0.4712 -0.992401 Prob(F) = 0 t = -2.8177 4.61 CREST 10.0373 3.1297 0.888786 DW = 2.5491 -0.911893 DW = 2.1227 0.2297 0.1028 0.2412 R-sq = 0.767065 AdjR-sq = 0.0000 0.0000 0.1216 0. ln Pt−2 .9377 3.0000 0.053098 Prob(F) = 0 t = -1.5352 0.49 CMC -0.0810 0.884129 AdjR-sq = 0.6672 0.0676 -0.4847 0.6535 2.1018 0.7875 0.0419 7.915063 Prob(F) = 0 t = -1.3522 0.1121 1.6456 0.9274 -0.5654 -1.1713 0.0414 -0.5275 0.0757 0.

5680 -2.718707 DW = 2.0711 0.8616 0.1078 0. Coeff S.5153 0.0000 0.1305 0.0379 R-sq = 0.919182 AdjR-sq =0.1472 0.0649 -0.9429 0.27 HCL 47.4964 1.2783 0.0591 0.3663 0.E t-Stat Prob.3113 0. CYBERSYS -0.0068 0.0411 0.6056 0.5453 0.941186 AdjR-sq = 0.9789 0.7619 -0.1305 0.4492 0.7413 5.1426 0.4614 0.0000 0.9163 2.6326 0.9940 R-sq = 0.8389 0.00 FINOLEX 0.1246 -0.0000 0.1481 0.2462 0.4656 4.0106 R-sq = 0.3051 -2.6236 2.4861 0.2606 0.1260 0.0746 -0.2590 -0.6422 89.0000 0.1731 0.024837 Prob(F) = 0 t = -2.5490 0.1682 -1.3300 -0.4734 3.1235 0.0142 -0. Contd.0328 0.0037 3.960987 Prob(F) = 0 t = -0.3702 R-sq = 0.1222 0.6672 -0. Coeff S.7659 0.4631 0.0871 -2.0000 0.9247 1.1277 0.E t-Stat Prob.0295 0.8022 -2.6297 1.7386 -0.9641 0.7748 -0.2753 5.1018 -0.4899 0.0126 0.0048 4.1041 -0.0000 0.946233 DW = 1.0000 0.909747 DW = 2.2858 0.2244 0.7761 0.E t-Stat Prob.7841 0.0038 0. ln Pt−1 ln Pt−2 ln BM ln M V CIISS August 2007 ln β Coeff S.1886 0.1303 3.45 DSQ 7.6988 9.1346 0.8419 -0.0817 -1.0449 -0.1288 0.1525 0.2533 -0.4526 1.1806 0.4987 1.3059 0.3139 1.935205 DW = 2.0000 .E t-Stat Prob.9408 0.0000 0.5525 6.9834 0.3023 0.8966 0. Coeff S.109003 Prob(F) = 0 t = -1.0051 63.14 ESERVE 0.1115 -2.4941 0.1355 0.912448 DW = 1.1280 0.97 HINDUJA -2.961691 DW = 2.3008 0.0058 0.1433 0.E t-Stat Prob.1257 0.918078 AdjR-sq = 0.4832 1.0000 0.1192 0.7953 R-sq = 0.950369 AdjR-sq = 0.6390 0.0101 -0.5441 0.9028 0.2991 0.0000 0.5755 -0. C Coeff S.501 ln β Negative 0.6198 0.7646 R-sq = 0.0000 0.6754 0.1225 0.7081 0.0209 7.0133 0.0000 0.1217 0.1734 16.744673 AdjR-sq = 0. Coeff S.0000 0.5072 0.9219 24.7635 -0.0009 -0.0000 0.964638 AdjR-sq = 0.105351 Prob(F) = 0 t = -3.8765 0.1264 2.4874 0.6217 11.3656 -0.0000 0.0026 -0.0000 0.1060 0.916596 Prob(F) = 0 t = 0.1393 -0.0496 -3.0076 0.0000 3.3191 0.00396 Prob(F) = 0 t = -2.1226 11.7613 0.7094 0.0004 0.E t-Stat Prob.1214 0.3472 2.0988 0.0266 -0.8987 1.9790 0.18 Table 3.9989 -2.

0684 1.0237 -0.2520 0. Contd.8469 0. MOSER Coeff S.9738 0.6225 0.8909 0.6093 0. C Coeff S.2299 0.3814 0.7472 0.1252 0.0006 0.0407 0.2155 0.2019 0.4682 0.904903 Prob(F) = 0 t = -2.0867 -0.8226 0.1321 0.0895 1.74 1.4847 0.7462 5.82 30.9098 0.7178 0.1138 1.5673 0.6277 0.888267 AdjR-sq = 0.4114 0.6700 2.2905 5.917345 DW = 1.0000 0.4579 0.0238 0.4480 0.0103 -1.3309 0.E t-Stat Prob.4103 1.749793 DW = 2.3900 2. MTNL 1.0859 0.0490 0.5900 0.0738 8.E t-Stat Prob.0694 3.0329 -0.0000 0.2924 R-sq = 0.8844 R-sq = 0.5738 -0.3446 0.923703 AdjR-sq = 0.1285 0.E t-Stat Prob.4784 0.1335 0.2548 0.0000 0.0027 0.0467 -0.46 9.1757 -0.8614 0.1147 0.804946 DW = 1.4809 0.1076 2.0000 0.0440 0.5069 6.1072 0.1296 0.6141 0.745948 AdjR-sq = 0.6414 R-sq = 0.7638 -0.0071 0.9542 0.5447 R-sq = 0.29 ln β Negative 0.0662 0.915473 Prob(F) = 0 t = -0.9210 0.8466 0.2506 0.1270 0.2929 0.0874 0.7308 -0.3315 0.5665 -1.0154 0.0000 0.1060 0.7043 0.0911 R-sq = 0.0000 0.0043 -0.1286 0.0000 0.0044 -0.5418 -0.02991 Prob(F) = 0 t = -4.0159 0.2035 R-sq = 0.9791 0.9839 0.0858 0.1499 -2.0659 -0.9790 7.772889 AdjR-sq = 0.962494 Prob(F) = 0 t = -3.2857 0.724777 DW = 1.8672 -0.4300 -0.1284 0.1439 0. INFOSYS ln Pt−1 ln Pt−2 ln BM ln M V ln β 19 Coeff S. JAIN Coeff S.48 5.1346 0.E t-Stat Prob.0886 -0.3315 0.0510 -0.0000 0.0000 0.5171 0.1347 0.0008 0.1801 -1.1516 -0.0000 -0.2427 0.0544 0.876904 DW = 1.E t-Stat Prob.2054 17.057018 Prob(F) = 0 t = -2.5248 4.1282 0. MPHASIS Coeff S.0624 0.4288 0.0000 0.3371 0.4669 -0.822951 AdjR-sq = 0.797689 AdjR-sq = 0.977954 Prob(F) = 0 t = -2.E t-Stat Prob.1135 0.3691 1.2697 0.1217 1.0491 0.9595 0.0000 0.0000 0.0206 1.4834 6.0070 -0.1239 0.7031 -0.0258 0.9744 0.1367 0.8248 -0.0322 -1. MASTEK Coeff S.1461 0.6309 -3.28 2.1940 -3.8730 0.9957 -0.1224 .0013 0.1082 0.9100 -1.8429 -0.1223 0.7773 -0.1605 0.0000 0.6331 -0.7552 0.1054 1.6423 0.777115 DW = 2.1730 0.Lalitha & Rao: Stock Price Model Table 3.3421 0.1151 0.

PENTA SOFT 0.1812 3.973686 Prob(F) = 0 t = -1.0338 0.9998 -2.2841 -1. PENTA MEDIA 37.0000 0.8103 0.7546 0.79 ln β Negative 0.0025 0.2701 R-sq = 0.9719 -0.7539 0.7213 -1.971087 DW = 1.1238 0.0004 0.3141 7.0191 0.0000 0.1549 5. Contd.6551 -0.5149 0.7937 0.5879 0.1361 1.7904 0.2988 0.7398 -0.3608 -0.0754 -2.1260 0.2847 0.7572 0. SATYAM 0.1320 0.0305 -3.1132 0.8495 0.8460 0.1647 2.0000 0.0000 0.1246 0.0000 0.0478 0.E t-Stat Prob.0000 0.1548 17.5635 -0.E t-Stat Prob.0398 0.0004 0.973311 AdjR-sq = 0.0000 0.88 1.932581 AdjR-sq = 0.0917 6.0000 0.9572 0.20 Table 3.4213 -0.0000 0.2023 0.1369 0.9722 -1.1736 5.0095 -0.6255 0.8692 -0.3214 0.2714 0.939241 DW = 2.1393 0.8385 -0.0706 -0.842621 AdjR-sq = 0.4907 -2.9787 Prob(F) = 0 t = -2.1301 0.2841 0.0000 . C Coeff S.1810 R-sq =0.6851 0.3535 0.0000 0.3706 0. NIIT ln Pt−1 ln Pt−2 ln BM ln M V CIISS August 2007 ln β Coeff S.0765 -3.0000 0.1100 -1.0000 0.829506 DW = 2.0000 0.2587 -0.94485 AdjR-sq = 0.0000 0.1393 -1.0000 Coeff S.5824 0.932581 DW = 2.7269 -0.9117 -0.2793 0.0000 0.8900 0.9393 0.043714 Prob(F) = 0 t = -2.8323 0.0381 0.880299 DW = 1.0000 0.2450 0.0000 0.1557 -3.4403 0.2572 0.8670 -0.1439 0.1951 1.0195 0.0574 0.4011 1.4848 -2.5597 0.0911 -0.4607 -0.E t-Stat Prob.1176 R-sq = 0. ORIENT Coeff S.3694 74.2126 -1.0807 -1.0864 2.46 11.0158 0.1604 -0.1436 3.6679 0.E t-Stat Prob.0637 0.8814 -3.2877 3.5752 0.4089 7.1144 0.3424 R-sq = 0.4967 3.E t-Stat Prob.1390 0.041689 Prob(F) = 0 t = -2.12 80.1022 0.1576 17.0198 0.0423 0.1274 0.0000 Coeff S.7416 6.1131 0.889507 AdjR-sq = 0.1576 -0.0000 0.0820 0.6667 -0.81 42.3318 1.0784 R-sq = 0.061302 Prob(F) = 0 t = -2.

6288 0.954265 AdjR-sq = 0.110483 TRIGYN 5.862401 VINDHYA 10.7600 0.1263 -1.8740 0.948448 AdjR-sq = 0.966808 AdjR-sq = 0.0000 0.0000 0.94 0. TATA ELXSI 3.3931 0.0000 0.0019 0.0000 .14 1.0000 0.6597 0.1310 0.E t-Stat Prob.918284 DW = Prob(F) = 0 t = -2.4772 -0.0776 0.950454 DW = Prob(F) = 0 t = -3.51 -0.7108 -3.2855 0.0000 0.0007 0. Contd.5873 0.1258 0.0000 0.944152 DW = Prob(F) = 0 t = -2.950218 AdjR-sq = 0.E t-Stat Prob.964042 DW = Prob(F) = 0 t = -1.0397 7.8370 0.0000 0.1695 -3.5290 -0.0078 1. Coeff S.0317 0.7422 0.0000 0.8659 0.0757 0.2191 0.8106 0.0204 0.4263 21 ln β Negative 0.4727 0.E t-Stat Prob.6891 0.1513 -0.0452 0.0158 6.0000 0.0027 0.0000 0.0853 3.5622 -0.2134 16.4430 0.9523 0.7364 0.1057 ln β -0.0006 R-sq = 1.1240 0.0776 -0.4058 3.1286 0.1014 1.0182 0.6190 0.0865 SOFT 0.1272 0.6957 R-sq = 2.E t-Stat Prob.1097 R-sq = 2.0672 8.2729 1.4552 -0.1335 0. C Coeff S.0002 -0.0380 -0.998641 VISUAL -0.0000 0.8659 -0.4020 3.5420 0.0000 0.1207 ln Pt−1 ln Pt−2 0.3069 -0.925827 AdjR-sq = 0.0045 0.0005 0.0000 0.0000 0.1558 -2. Coeff S.2088 ln M V -0.098287 VSNL 0.6133 -0.0000 Coeff S.0001 0.2576 -0.0269 0.6236 0.0000 0.2506 0.2200 -0.6381 0.0216 -0.1444 6.0000 0.7867 0.7807 -0.0001 -0.1245 0.2719 -0.6353 0.8943 0.5870 3.055311 0.9000 0.0031 0.E t-Stat Prob.0338 ln BM -0.2572 -0.1881 0.1178 0.9084 0.0611 0.0000 0.4417 0.0000 0.7560 0.1178 0.1248 0.0000 0.Lalitha & Rao: Stock Price Model Table 3.7976 0.8514 R-sq = 2.7183 -0.0379 0.0021 0.1262 -2.30 0.8347 0. 16.0000 0.5277 R-sq = 1.0651 -0. Coeff S.0027 0.01 0.1904 7.0601 0.9839 0.1238 0.1226 0.0203 1.4754 0.946069 DW = Prob(F) = 0 t = -1.0710 0.0063 0.1404 0.

1519 ln β Negative -0. WIPRO 0.9385 0.E t-Stat Prob.0000 0.1268 0.E t-Stat Prob.7270 0.0000 0.4017 3.0364 R-sq = 2.64 13.7671 0.7345 -2.1680 0.1444 -2.0580 0.66 and +2.1064 0.9230 -0.0000 0.0439 0.0000 0.2118 83.6019 6.0000 0.5873 0.6783 0.6061 0.944739 DW = 1.4777 0.0985 -0.0111 0.2398 -0.0003 0.0969 -7.0000 0.2493 0.0018 R-sq = 0.0775 -3.5415 12.0665 -4.0566 0.9720 0.1358 7.0361 0.0000 0.9964 0.1265 0.933332 DW = Prob(F) = 0 t = -4.004182 Prob(F) = 0 t = -2.3749 ln M V -0.2948 -7. C Coeff S.2825 0.2149 2.090802 51.2002 0.176874 178.66 .892003 AdjR-sq = 0. ZENSAR 1.1214 0.8431 -0.0000 0.4655 0.5733 -0.7915 -0.E t-Stat Prob.8824 -0.5478 2.3508 -3.2743 0.E t-Stat Prob.9657 5.34 0.0809 -0.0368 0.0738 -2.0370 3. ZEE Coeff S.0615 ln BM -0.0000 5% critical values for t (59) / t (60) are -2 and +2 1% critical values for t (59) / t (60) are -2.9240 1.2614 0.0000 0.22 Table 3.0883 0.7051 -0.0117 0.1306 0.6132 2.0528 0.821649 Prob(F) = 0 t = -3.0113 0.94899 AdjR-sq = 0.8891 0.939486 AdjR-sq = 0.883003 DW = 2.0001 25.1186 0.8487 ln Pt−1 ln Pt−2 0.0389 2.1416 0.1805 0.1304 0.4546 -0.0029 CIISS August 2007 ln β Coeff S.5501 4.0966 0.3004 0.29 0. ZENITH Coeff S.1420 19.863754 DW = Prob(F) = 0 t = -2.1297 0.6038 0.0622 -0.876331 AdjR-sq = 0.9763 0.5124 -0.0000 0.6560 0.0702 -0.5184 -3.0439 R-sq = 0.2996 2.1362 0.1319 3.4461 R-sq = 2.54 2.6103 0. Contd.0016 0.0000 0.6213 2.3991 0.5457 -0.7144 6.4169 0.7655 0.

070125 VINDHYA 1.019159 23 χ2 Prob χ2 Prob χ2 Prob χ2 Prob χ2 Prob χ2 Prob .7656 0.0695 0.003861 TATA ELXSI 10.001074 MASTEK 0.012861 PENTAMEDIA 3.048645 MOSER 6.018647 ESERVE 4.347959 0.19114 0.188097 0.00027 BLUESTAR 12.69559 0.78141 0.105548 FINOLEX 10.00048 DSQ 4. Wald test of the behavioral inertia model χ2 Prob ACE 2.486939 0.046289 HCL 3.786176 0.93433 0.454485 ORIENT 4.045246 JAIN 10.971055 0.001022 NIIT 8.096308 CREST 3.280081 0.051178 0.000018 COSMO 2.64737 HINDUJA 0.051825 ZENSAR 13.41242 0.616146 MPHASIS 7.044141 ZEE 5.000017 SATYAM 7.013721 VISUALSOFT 3.559443 0.26933 0.99686 0.Lalitha & Rao: Stock Price Model Table 4.341178 0.251322 0.9371 0.61965 0.013754 MTNL 18.010983 AFTE 3.78742 0.03381 TRIGYN 4.004686 PENTASOFT 6.000944 WIPRO 18.011797 CMC 5.009459 0.253203 ZENITH 6.504282 0.04723 CYBERSYS 0.467986 0.534355 0.47962 0.887571 0.032305 INFOSYS 6.005265 VSNL 6.073744 0.582247 0.209232 0.305545 0.

45763 0.30957 0.27766 0.67420 125.37418 0.69860 196.70700 262.08350 318.37947 0.61230 633.88630 883.83590 Theil Ineq.85000 168.44515 0.39057 0.61330 130.18720 288.54730 0.24 Table 5.43930 176.81000 137.95460 176.91980 148.14634 0.07230 370. A dynamic.61953 0.68240 120.38177 0.46158 0.62968 0.02960 114.65740 50602.33182 0. 0.40340 0.78540 198.04350 161.50490 110.25454 0.45227 0.34005 0.16400 851.85130 115.68885 0. in-sample forecast of the Fama and French model Portfolio Categories LOWBEME MEDBEME HIGHBEME LOWBEME WTD MEDBEME WTD HIGHBEME WTD SLBEME SMBEME SHBEME BLBEME BMBEME BHBEME SLBEME WTD SMBEME WTD SHBEME WTD BLBEME WTD BMBEME WTD BHBEMEWTD LOWEPSP MEDEPSP HIGHEPSP LOWEPSP WTD MEDEPSP WTD HIGHEPSP WTD SLEPS SMEPS SHEPS BLEPS BMEPS BHEPS SLEPS WTD SMEPS WTD SHEPS WTD BLEPSWTD BMEPS WTD BHEPS WTD MAPE 138.90670 204.29400 604.46656 0.41984 CIISS August 2007 MAPE : : Mean Absolute Percentage Error .74200 272.46320 186.59581 0.37231 0.27703 0.43200 0.44783 0.57486 0.53040 1193.14610 279.74750 1611.75572 0.32803 0.25380 559.15290 0.18430 126.32650 0.53700 240.36140 11746.24796 0.27699 0.59913 0.36520 201.33056 0.25119 0.65210 123.

733216 58.082944 0.752003 16.023913 0.285069 0.053656 0.284171 2.640259 10.087276 0.243619 5.062769 0.952510 26.025801 0.575910 4.756940 21.949859 4.064224 0.037949 0.501098 8.913490 10.223986 6.076810 10.780920 9. in-sample forecast of the behavioral inertia model SERIES ACE AFTE BLUESTAR CMC COSMO CREST CYBERSYS DSQ ESERVE FINOLEX HCL HINDUJA INFOSYS JAIN MASTEK MOSER MPHASIS MTNL NIIT ORIENT PENTAMEDIA PENTASOFT SATYAM TATA ELXSI TRIGYN VINDHYA VISUALSOFT VSNL WIPRO ZEE ZENITH ZENSAR MAPE 12.118091 0.076916 0.024622 0.273919 17.044740 0.084641 0.258890 16.207320 0.053670 0.041740 0.054045 0.624820 Theil Ineq.075450 0. A dynamic.755490 3.784530 3.070138 0.137682 0.013973 0.073154 25 .548486 8.Lalitha & Rao: Stock Price Model Table 6.072514 12.351700 6.856250 3.029230 0.358880 15.854880 7.034547 0.093213 0.060570 13.058233 0.050925 0.263292 41.279920 6.021038 0. 0.024414 0.355607 4.039356 0.027271 7.381320 15.072841 0.311590 13.

26 APPENDIX 1 Constructed Portfolios CIISS August 2007 In this study portfolios were constructed using various criteria. Stocks were also classified into two groups of small and big. This classification along with BE/ME and EPS/P ranking gave us the size-BE/ME and size-EPS/P sorted portfolios. 11. With BE/ME and EPS/P ranking sub-group of low. Thus the various portfolios areList of Portfolios. Similarly. Portfolio LBEME MBEME HBEME LBEMEWTD MBEMEWTD HBEMEWTD SLBEME SMBEME SHBEME BLBEME BMBEME BHBEME SLBEMEWTD SMBEMEWTD SHBEMEWTD BLBEMEWTD BMBEMEWTD BHBEMEWTD Description Low BE/ME Portfolio Medium BE/ME Portfolio High BE/ME Portfolio Low BE/ME Value Wtd Portfolio Medium BE/ME Value Wtd Portfolio High BE/ME Value Wtd Portfolio Small and Low BE/ME Portfolio Small and Medium BE/ME Portfolio Small and High BE/ME Portfolio Big and Low BE/ME Portfolio Big and Medium BE/ME Portfolio Big and High BE/ME Portfolio Small and Low BE/ME Value Wtd Portfolio Small and Medium BE/ME Value Wtd Portfolio Small and High BE/ME Value Wtd Portfolio Big and Low BE/ME Value Wtd Portfolio Big and Medium BE/ME Value Wtd Portfolio Big and High BE/ME Value Wtd Portfolio . Both equal weighted and value weighted portfolios were constructed with the value weights being equal to the ratio of equity value of the firm to the aggregate total equity of all the sample firms. the composition of portfolio also changed. The next twelve are equal and value weighted portfolios constructed using the size-BE/ME criteria. Stocks were ranked on the basis of size. book-to-market value (BE/ME) and EPS/P and grouped to form portfolios. Over time as the order of rank changed. and 12 stocks were formed. The first six portfolios correspond to equal weighted and value weighted portfolios formed by using BE/ME criteria. the single sort EPS/P criteria generates a set of six portfolios formed using equal weights and value weights and the double sort size-EPS/P criteria produces two sets of equal weighted and value weighted portfolios each consisting of six sorts. medium and high consisting of 12.

Lalitha & Rao: Stock Price Model List of Portfolios : Contd Portfolio LEPSP MEPSP HEPSP LEPSPWTD MEPSPWTD HEPSPWTD SLEPS SMEPS SHEPS BLEPS BMEPS BHEPS SLEPSWTD SMEPSWTD SHEPSWTD BLEPSWTD BMEPSWTD BHEPSWTD Description Low EPS/P Portfolio Medium EPS/P Portfolio High EPS/P Portfolio Low EPS/P Value Wtd Portfolio Medium EPS/P Value Wtd Portfolio High EPS/P Value Wtd Portfolio Small and Low EPS/P Portfolio Small and Medium EPS/P Portfolio Small and High EPS/P Portfolio Big and Low EPS/P Portfolio Big and Medium EPS/P Portfolio Big and High EPS/P Portfolio Small and Low EPS/P Value Wtd Portfolio Small and Medium EPS/P Value Wtd Portfolio Small and High EPS/P Value Wtd Portfolio Big and Low EPS/P Value Wtd Portfolio Big and Medium EPS/P Value Wtd Portfolio Big and High EPS/P Value Wtd Portfolio 27 .

02755 1% Critical Value * 5% Critical Value 10% Critical Value .24905 -5.59867 -5.90226 -4. Portfolio categories LBEME MBEME HBEME LBEME WTD MBEME WTD HBEME WTD SLBEME SMBEME SHBEME BLBEME BMBEME BHBEME SLBEME WTD SMBEME WTD SHBEME WTD BLBEME WTD BMBEME WTD BHBEME WTD ADF Test Statistic -4.50817 -3.17855 -4.72802 -4.56958 -3.4801 -3.28 CIISS August 2007 APPENDIX 2 Unit Root Test on Returns on Various Portfolios.97865 -4.26832 -3.4449 -3.66814 -3.1675 ADF Test Statistic -4.30645 -4.58284 -3.11845 -4.1587 -4.52882 Portfolio categories LEPSP MEPSP HEPSP LEPSP WTD MEPSP WTD HEPSP WTD SLEPS SMEPS SHEPS BLEPS BMEPS BHEPS SLEPS WTD SMEPS WTD SHEPS WTD BLEPS WTD BMEPS WTD BHEPS WTD SMB HML EX RET BSE -4.02462 -4.95087 -3.89775 -4.01194 -4.7389 -3.45268 -5.90823 -4.84653 -4.02282 -4.39183 -3.64291 -3.15495 -4.1059 -3.35248 -4.18031 -3.68762 -4.39274 -4.55762 -3.62454 -4.66651 -4.7908 -4.69531 -3.8988 -3.

65933 4.86096 3.65260 -3.24104 ACE AFTE BLUESTAR CMC COSMO CREST CYBERSYS DSQ ESERVE FINOLEX HCL HINDUJA INFOSYS JAIN MASTEK 3.47584 4.17294 0.26175 4.03148 0.24613 0.36177 -2.94376 5.60139 3.07082 0.04889 4.07396 -1.80383 6.05328 -0.59224 3.41327 -2.03034 0.99447 -1.38188 -3.01103 11.46176 4.10800 .22190 0.51209 -0.29002 -2.55981 -1.81554 -2.70040 -3.65562 4.42558 10.45127 0.53597 -1.51687 -3.49399 -1.94296 4.29531 6.01011 0.11550 4. Trend and intercept Intercept First Difference ADF Test Statistic 4.50963 11.01687 LR statistic p value ADF Test Statistic -1.97545 5.60502 -1.18307 -1.04929 0.22589 0.Lalitha & Rao: Stock Price Model APPENDIX 3 : Unit Root Test on Behavioral Model Variables A: Unit root test on lnPt series.86471 -3.28890 -2.49361 29 Company Name ADF Test Statistic -1.21061 0.99037 2.38762 5.66241 -2.02134 0.06762 0.62226 3.63110 3.16278 -0.30391 -1.02015 15.83399 4.31309 -0.09915 0.05534 -2.79965 2.64174 -1.87510 4.61098 -1.61218 3.45132 3.22767 -1.70677 -2.02434 0.

19205 0.81744 2.69573 -1.58313 -2.26034 4. A: Unit root test on lnPt series.00724 3.71294 MOSER MPHASIS MTNL NIIT ORIENT 14.66568 4.75306 4.70897 CIISS August 2007 Company Name ADF Test Statistic -3.29305 5.14019 -2.71885 LR statistic p value ADF Test Statistic -3.39461 5.31814 -1.26908 4. Trend and intercept Intercept First Difference ADF Test Statistic 4.69345 -3.20752 11.27396 -1.28438 0.23041 3.24155 3.51994 -1.64867 4.30433 -1.60495 -1.15441 0.87475 -2.09223 -1.14635 0.02929 0.73631 6.95324 -1.11882 0.02007 0.05630 -2.00977 -2.61819 3.30000 12.30 Contd.15712 -1.78012 4.01304 5.02116 -1.51537 -1.84356 4.89891 -0.49261 4.88117 -0.79180 7.61317 -0.12200 0.09109 0.59670 0.02217 0.55242 PENTASOFT SATYAM TATAELXSI TRIGYN VINDHYA VISUALSOFT VSNL WIPRO ZEE -2.73524 -1.53922 4.03724 0.02415 4.17511 -1.97569 6.19582 -1.10042 PENTAMEDIA -1.46158 -2.10579 0.87954 .32108 4.10937 -2.01774 0.88054 -3.03802 0.17209 4.10155 5.51490 7.18448 4.

65304 -1.55714 ZENITH ZENSAR 3.18838 0.09658 4.14573 -1. A: Unit root test on lnPt series.85018 LR statistic p value ADF Test Statistic -1.28964 31 Company Name ADF Test Statistic -1.Lalitha & Rao: Stock Price Model Contd.29169 4.12317 .21261 0. Trend and intercept Intercept First Difference ADF Test Statistic 5.

63905 6.43957 -0.45216 4.51057 4.57227 -1.09700 -0.24075 7.48791 4.06931 6.49499 4.35138 -2.62910 6.70587 -1.12341 0.46875 -1.07589 4.52998 4.99117 -1.04809 0.25576 -1.18705 0.40410 -1.01763 0.32 B: Unit root test on lnBM series Trend and intercept Intercept First Difference ADF Test Statistic 4.27267 -1.27988 -1.04414 0.02205 0.44662 4.90305 -1.46972 4.34830 -2.58385 LR statistic p value ADF Test Statistic -1.98408 -2.61597 -2.53345 4.69387 -1.08523 -1.45441 -2.95558 4.52801 -1.74690 -1.89377 -2.35276 6.56365 4.53970 4.03178 0.42389 3.84192 -1.77535 -1.96977 0.60665 4.45668 CIISS August 2007 Company Name ADF Test Statistic -1.18449 3.03937 0.31826 -2.79174 -0.56252 4.04028 0.70001 4.50575 4.46987 5.73121 -2.05963 0.43327 ACE AFTE BLUESTAR CMC COSMO CREST CYBERSYS DSQ ESERVE FINOLEX HCL HINDUJA INFOSYS JAIN MASTEK MOSER 3.34250 6.69359 -2.03725 0.21934 0.11404 0.41144 -2.22653 .92264 -1.17848 0.48074 4.58005 8.10700 0.48453 4.56533 -1.89768 2.03424 6.

07367 0.40941 7.03022 0.39026 -2.89420 4.05325 -2.04332 0.99881 2.44255 4.08048 0.32218 -2.69646 -1.03037 0.46106 LR statistic p value ADF Test Statistic -1.78966 -1.48835 4.47237 4.60752 -1.77333 -0.Lalitha & Rao: Stock Price Model Contd.66177 0.23525 0.80392 4.36106 -1.41817 5.78718 4.44561 4.30145 6.11472 -1.49689 4.92930 -1.01880 0.11640 0.22251 7.43463 -2.04771 0.45388 PENTASOFT SATYAM TATAELXSI TRIGYN VINDHYA VISUALSOFT VSNL WIPRO ZEE ZENITH ZENSAR -2.21641 5.02169 PENTAMEDIA -1.61062 4.65237 -0.42894 -1.08525 7. B: Unit root test on lnBM series Trend and intercept Intercept First Difference ADF Test Statistic 4.02605 0.44764 4.67975 4.85833 -1.03950 4.43571 4.62594 -1.60236 4.24286 0.48221 -1.65490 -2.77276 -2.73857 .73398 33 Company Name ADF Test Statistic -1.47746 4.86790 MPHASIS MTNL NIIT ORIENT 4.12109 0.83058 6.43765 4.18183 0.22202 3.29549 6.53894 -2.10980 0.64671 -2.60426 -0.94784 3.19969 0.03260 -0.94517 -2.98864 6.16089 -1.27813 2.02192 -1.23658 -2.43791 -1.51874 4.25882 -0.

56535 3.36729 5.44177 4.78896 -1.90081 -1.44668 -1.36305 -2.58942 4.69785 -2.09524 0.03391 9.51793 -2.57577 2.45680 4.50805 5.44491 -1.04895 0.66261 6.12859 -2.03546 0.75451 -0.39192 -1.22814 0.04069 0.01066 0.43807 4.11263 0.66497 CIISS August 2007 Company Name ADF Test Statistic -2.70790 4.38008 ACE AFTE BLUESTAR CMC COSMO CREST CYBERSYS DSQ ESERVE FINOLEX HCL HINDUJA INFOSYS JAIN MASTEK MOSER 5.82017 -1.89845 5.03495 0. Trend and intercept Intercept First Difference ADF Test Statistic 4.66209 4.05200 -2.93344 -2.49163 4.91776 -2.13013 0.31318 -2.16731 0.29033 LR statistic p value ADF Test Statistic -1.95557 6.11083 -2.83477 -1.48169 4.95382 -1.15568 -1.37317 -1.06336 .78711 6.10495 -1.70272 6.60052 4.05538 0.34031 -1.07852 6.43471 4.45677 4.40378 3.97142 -1.58803 4.67853 6.59142 -1.38163 -1.42819 -0.56581 4.51792 0.03753 0.49494 -2.43803 4.05894 0.03177 0.16025 0.52703 -2.40748 -2.60634 -2.34 C: Unit root test on lnMV series.02213 4.08200 4.56803 4.

23713 0.43578 4.44698 -1.44177 4.04732 0.01998 -1.03124 PENTAMEDIA -1.47290 4.37427 -2.39523 6.87831 4.46998 4.47385 4.82651 6.03264 0.47348 4.04630 -2.43776 4.03959 4.42122 -1.48565 -2.63038 -1.68868 -0.04978 -2.25233 -1.81016 MPHASIS MTNL NIIT ORIENT 3.69817 -2.13890 -2.03659 0.00601 6.15735 3.33388 -1.43741 4.69795 -2.08842 0.56124 -1.Lalitha & Rao: Stock Price Model Contd.93208 0.63077 35 Company Name ADF Test Statistic -1.34824 -0.78138 7.58112 4.61590 2.72643 -0.76691 -2.43866 4.09157 0.48183 4.50661 LR statistic p value ADF Test Statistic -1. C: Unit root test on lnMV series Trend and intercept Intercept First Difference ADF Test Statistic 4.21729 6.85137 4.77324 4.63825 -2.67812 -1.92836 -1.11632 -1.04881 0.10152 6.11107 0.20410 0.34276 -2.50294 4.44896 PENTASOFT SATYAM TATAELXSI TRIGYN VINDHYA VISUALSOFT VSNL WIPRO ZEE ZENITH ZENSAR -2.95845 4.04978 0.34249 -1.44734 4.57356 -1.00026 3.74205 -1.13493 0.84418 6.03083 0.04509 -1.20016 0.01998 0.47360 4.17826 6.20625 0.23992 -1.56653 .

27136 -1.43536 4.47579 4.62973 3.55389 4.83675 -1.42086 4.08732 -1.93612 -1.45580 4.99547 -1.07642 -1.09400 -1.25853 -1.01351 -1.52630 4.46930 4.13770 0.05603 0.67184 4.24993 5.12559 0.06941 0.93059 -1.14948 4.09536 0.67929 -1.92160 -1.96533 3.02711 -1.47489 -1.76894 -2.92159 -1.89370 5.28300 -1.83557 ACE AFTE BLUESTAR CMC COSMO CREST CYBERSYS DSQ ESERVE FINOLEX HCL HINDUJA INFOSYS JAIN MASTEK MOSER 4.98368 -0.77483 4.70030 6.00997 5.01602 -1.81387 LR statistic p value ADF Test Statistic -1.57807 4.55766 4.36 D: Unit root test on lnBeta series.00279 -2.33535 5.04284 0.52444 4.10965 0.07863 0.98758 -2.05560 .83253 -2.87919 -2. Trend and intercept Intercept First Difference ADF Test Statistic 4.44790 CIISS August 2007 Company Name ADF Test Statistic -2.11284 -2.38496 -1.72998 -0.47875 4.08657 0.13466 0.10228 0.30071 4.16286 0.60670 -1.15467 0.76365 3.73295 4.77724 -2.11944 0.15146 0.67338 -2.44256 4.77913 0.52708 -2.44652 4.08610 3.43495 4.47052 4.56002 4.

14038 0.82140 -1.04950 0.49386 6.Lalitha & Rao: Stock Price Model Contd.15542 0.58659 -1.13234 0.12984 -1.55780 4.47025 -1.19380 PENTAMEDIA -2.25219 -1.05076 0.45237 4.11015 0.19842 3.65162 -2.15884 -1.56539 5.10201 0.12255 0.95760 -0.94429 -1.89884 -1.72324 4.90340 -1.33023 LR statistic p value ADF Test Statistic -1.82715 4.48242 4.56911 4.51317 37 Company Name ADF Test Statistic -1.46276 4.74860 -2.55546 4.92688 3.52773 4.41190 3.84458 -2.85829 -2.77565 MPHASIS MTNL NIIT ORIENT 4.60406 4.95294 -1.16882 0.65334 -1.22010 4.27673 6.33018 .16833 3.43507 4.13750 0.63400 4.96829 3.30598 -0.04470 6.34514 -1.96121 5.15255 -0.42502 -2.04460 0. Trend and intercept Intercept First Difference ADF Test Statistic 4.88091 -1.58258 4.12441 0.04686 -2.78388 -1.13622 4.50024 4.43946 4.69949 -1.04651 0.43471 4.31915 -1.06413 0.72157 -2.28186 0.01159 3.17161 PENTASOFT SATYAM TATAELXSI TRIGYN VINDHYA VISUALSOFT VSNL WIPRO ZEE ZENITH ZENSAR -2.09463 -1.19430 0.43766 4.94196 -1. D: Unit root test on lnBeta series.

41190 3.36461 .229586 0.82460 .76599 3.36461 -0.05506 0.43144 5.66881 4.58659 -1.43471 4.15213 0.43471 4.76599 4.79885 4.82460 4.10908 0.83529 -2.43471 4.29146 -1.34507 3.14250 -1.15213 0.44202 4.90693 -0.98384 -1.15213 0.62444 -2.76599 5.00838 -1.11015 0.60524 -1.58258 CIISS August 2007 Company Name ADF Test Statistic LR statistic p value ADF Test Statistic BLUESTAR 1.4406 4.58258 4.43471 4.76599 3.76113 -1.43471 4.05435 0.90693 -1.90693 -1.58258 4.74990 -1.08674 0.43471 4.05953 0.56294 -2. Trend and intercept Intercept First Difference ADF Test Statistic -4.83529 -2.942952 5.43479 4.64263 3.83529 -1.90693 -1.98384 2.576960 COSMO -2.88978 5.93064 -1.90340 -1.83529 -1.15213 0.05435 -1.31972 ESERVE FINOLEX HCL JAIN MOSER MPHASIS MTNL NIIT TRIGYN WIPRO ZEE -2.38 E: Unit root test on lnBeta negative series.18777 0.90693 -1.

5903 First Difference -3.1669 Intercept -3.9069 -2.5907 .Lalitha & Rao: Stock Price Model 39 ADF Critical values for the regression with trend and intercept. intercept (level).5328 -2.4790 -3.1035 -3. and intercept (first difference) Critical Value* 1% 5% 10% Trend and intercept -4.9062 -2.5345 -2.

6762 COSMO 0.7686 INFOSYS 0.18023 0.5602 BLUESTAR 3.83592 VISUALSOFT 2.31789 FINOLEX 2.22343 0.193374 0.340922 0.83439 PENTAMEDIA 0.71662 PENTASOFT 2.65839 0.9238 CREST 3.99735 VSNL 2.40 APPENDIX 4 Breusch-Godfrey Serial Correlation Test.26538 MASTEK 2.22588 0.19743 ESERVE 3.24474 0.28626 CMC 0.713122 0.630086 0.69886 0.36929 0.19466 0.278009 0.363496 0.02398 AFTE 2.254206 0.273 DSQ 1.28925 0.8151 VINDHYA 0.7392 MTNL 0.890049 0.97329 Prob Company LM Statistic MOSER 3.80806 CIISS August 2007 Prob.42704 MPHASIS 2.87743 0.108377 0.26152 ZENSAR 1.23296 NIIT 4. Company LM Statistic ACE 2.72375 CYBERSYS 3.155381 0.791218 0.99236 ZEE 3.44429 ORIENT 0.2119 ZENITH 0.680927 0.22613 0.40494 .65889 0.46836 JAIN 0.28327 0. 0.875737 0.9755 HCL 2.2007 0.52272 TATAELXSI 2.48095 WIPRO 1.549519 0.1522 HINDUJA 0.19034 0.

” University of Chicago. Journal of Financial Economics 49 283-306. F. M. 8. (2002) Presidential address: Rational Asset Prices. (1988) “Habit Formation: A resolution of the equity premium puzzle. M. C. October.” Graduate school of Business. (2000) Trading is hazardous to your wealth: The common stock investment performance of individual investors. (1981) The relationship between return and market value of common stock. 9. 6. (2004) Why do individual investors hold underdiversed portfolios? Working paper.B. September. Brav A. Review of Financial Studies 14 659-680. Goetzmann W. Journal of Finance 51 55-84. F. . and Kumar A. Columbia University. 3. and French K R. M. (1992) The cross-section of expected stock returns. Fama E. Daniel K. Heaton J. and Odean T.Lalitha & Rao: Stock Price Model REFERENCES 41 1. 10. 11. Yale International center of finance and University of Notre Dame. 4. (1997) Evidence on the characteristics of cross-sectional variation in common stock returns. and Heaton J. Journal of Financial Economics 9 3-18. Fama E. and Titman S. (1989)“General equilibrium with time complementarity: consumption. Constandinides G. Fama E. 5. 12. Barber B. Journal of Finance 55 773-806. D. (1988) Permanent and temporary components of stock prices. Journal of Political Economy 96 246-273. (1995) An empirical investigation of asset pricing with temporally dependent preference specifications. (1998) Market efficiency. 7. F. 14. and the equity premium. Huberman G. (1996) Multifactor explanations of asset pricing anomalies. and French K R. (2002) Competing theories of financial anomalies. Review of Financial Studies 15 575-606. Econometrica 63: 681-717. 2. Constandinides G. interest rates. Banz R. Journal of Finance 57 1567-1591. (2001) Familiarity breeds investment. Journal of Finance 52 1-33. November. 13. long-term returns and behavioral finance. and French K R. Fama E. Journal of Finance 47 427-465. Detemple J. F. N.

16. (1993) Returns to buying winners and selling losers: Implications for stock market efficiency. D. Stattman D. interest and money. University of Minnesota. Quarterly Journal of Economics 69 99-118. (2003)Household portfolio diversification. (1966) Equilibrium in a capital asset market. University of Notre dame. Stanley T. Journal of Finance 50 131-155. and Lanstein R. (2000) Challenging time series-limits to knowledge inertia and caprice. Working paper Carlson School of management. 19. Journal of Portfolio Management 11 9-17. Kumar A. F. (1936) The general theory of employment. (1989) Intertemporally dependent preferences and the volatility of constitution and wealth. Cambridge University Press. . 21. Keynes J. 25. 17. Edward Elgar Publishing Ltd. UK. Econometricia 34 768-783. February 2003. 22. Polkovnichenko V. 24. M. August. Jegadeesh N. M. 23. A. 26. (1955) A behavioral model of rational choice. (1985) Persuasive evidence of market inefficiency. Review of Economics and Statistics 47 13-37. Sunderasan S. Mossin J.42 CIISS August 2007 15. 20. Rosenberg B. Sharpe W. Working paper. Journal of Finance 48 65-91.. (1965) The valuation of risk assets and the selection of risky investments in stock portfolios and capital budgets. Simon H. (1964)Capital asset prices: a theory of market equilibrium under conditions of risk. (1980) Book values and stock returns. and Titman S. Review of Financial Studies 2: 73-89. Reid K. The Chicago MBA: A Journal of selected Papers 4 5-45. Mendoza College of Business. (2004) One trade at a time: Narrow framing and stock investment decisions of individual investors. and Lim S S. 18. Lintner J.