Modern theory of investment in financial assets, as opposed to investment in real and derivative assets, has embraced three main building blocks of continued development and refinements. It spans three consecutive decades from 1950s to 1970s namely, the portfolio theory and the single-factor model which are based on the mean-variance efficiency (MVE) for assets allocation pioneered by Markowitz (1952, 1959) and simplified by Sharpe (1963), the capital asset pricing model (CAPM) developed independently by Sharpe (1964), Lintner (1965), and Mossin (1966), and the arbitrage pricing theory (APT) by Ross (1976).

While the MVE is concerned with the diversifiability of idiosyncratic risk associated with the expected rates of return on securities through optimal portfolio selection, the CAPM is more involved with the estimation of the general-equilibrium rates of return on securities in relation to the term structure of the risk-free interest rate and the non-diversifiable market risk premium. In particular, both MVE and CAPM rely on the second moment of the stochastic return-generating function, i.e., variance (s²) and standard deviation (s), as the measures of risk. In general, however, the proxies for risk can be derived from factors other than variance-covariance relationship between the market and the specific security as extended by the APT. All these three interrelated theories and models shall be explored in more detail in the sections to follow. The interim section investigates the empirical tests of CAPM and APT. My own conclusion and critique are supplied in the final section of this paper.

The Mean-Variance Efficient Portfolio Theory Back to Top

The Markowitz's theory of portfolio selection introduces the use of MVE in finding the optimal combination of securities which has the minimum variance and (i.e., the efficient portfolio) and factor analysis in modeling the behavior of security prices and correlation among security returns. With regard to the MVE, the objective is to find the portfolio weights that minimize the level of portfolio risk, proxied by the variance, subject to a given level of expected rate of return, given by the weighted mean returns of n securities. The following presents two alternative notations - statistical and matrix algebra - to derive the required portfolio weights:

min V = S_nS_ns_ijw_iw_j = min 0.5w^TVw

Subject to

E = S_nw_iE(r_i) = w^Te

1 = S_nw_i = w^T1

 

where

V = portfolio variance

E = portfolio expected return

e = n-vector of security expected return

w = n-vector of portfolio weights

s_ij = covariance of securities i and j, i¹j

w_i, w_j = weights of holding securities i and j

Using the Lagrangian optimization technique with two multipliers, l and g, the resultant portfolio weight is given by:

w = l(V^-1e) + g(V^-11)

Further manipulation of matrix algebra yields:

w = D^-1[B(V^-1)-A(V^-1e)] + D^-1[C(V^-1e)-A(V^-11)]E

where

A = (e^TV^-11)

B = (e^TV^-1e)

C = (1^TV^-11)

D = (BC-A2)

and

l = D^-1(CE-A)

g = D^-1(B-AE)

Characteristics of the MVE Portfolio Frontier

1. The Mean-Variance (MV) frontier has a parabolic shape while the Mean-Standard Deviation (MSD) frontier has a hyperbolic shape.
2. The MVE portfolios are located on the frontier above the minimum-variance portfolio.
3. Any frontier portfolio can be generated by a linear combination of two other frontier portfolios.
4. Every MVE portfolio has a counterpart zero-covariance inefficient portfolio whose return is equal to the intercept of the line tangent to that MVE portfolio on the MV parabola.
5. The expected return of any portfolio can be generated by a linear combination of MVE portfolio and its zero-covariance counterpart.
6. The linear combination between the risk-free rate and the MVE portfolio implies a two-fund separation.

The Factor Model Back to Top

As Sharpe (1984) indicates, a factor model describes the expected-return-generating process which underlies the behavior of security prices. It attempts to capture the major sources of correlation among security returns. The general representation of factor model is given by:

r_j = a_j + b_j1F₁ + b_j2F₂ + ... + b_jkF_k + e_j

E[r_j] = a_j + b_j1F₁ + b_j2F₂ + ... + b_jkF_k + E[e_j]

where

r_j = ex ante rate of return on security j

a_j = non-factor-related rate of return on security j

b_ji = sensitivity of the return on security j to the i^th factor

F_i = value of the i^th factor, i = 1,2,...,k

ej = residual return uncorrelated with any factors in the model

E[e_j] = zero

The Factor Model Assumptions and Implications

The main assumption underlying the factor model is that the residual return of j^th security (i.e., e_j, j = 1,2,...,n) are uncorrelated with one another in addition to its unrelatedness to any factors in the model. Therefore, it is implied that the cross-sectional returns among n securities will be related to each other only through common reactions to one or more of the k factors. Moreover, in the expected-return form, E[r_j], as E[e_j] becomes zero, a_j is assumed to capture all the expected non-factor-related aspects of the return on security j, while each b_ji is set so that each factor-related aspect is captured in the associated relationships.

 

There is no assumption made about the common a and the tastes of the investors, such as time and risk preferences, in the factor model. Thus, there is no guarantee that the return generated by one security whose b_ji differ from those of the other securities will be different, since each of their a_j can be adjusted so that the returns on all securities are identical. Nor can it not be implied that a factor model for one period will be a good one for the next period, and that a good factor model for one individual is applicable to another individual or all other individuals.

 

Many researchers attempt to identify the relevant and pervasive factors to be included in their multi-factor models. Yet, Markowitz (1959) who develops and Sharpe (1963) who simplifies the single-factor model are not satisfied with such implications because it cannot be used to explain or derive the equilibrium rates of return on securities being traded in the capital markets. The search for the equilibrium asset pricing model consummates when Sharpe (1964), Lintner (1965), and Mossin (1966) independently develop the CAPM and Ross (1976) the APT.

 

The Capital Asset Pricing Model (CAPM) Back to Top

The missing assumptions about the common intercept term a and the investor preferences are identified and included in the equilibrium models such as the CAPM and its extensions and variations. The equilibrium model relates the expected rates of return on securities over the next period to several attributes of those securities including the market's agreed-upon a which is the risk-free interest rate and the preferences of the investors. The basic CAPM assumes that investors are risk-averse expected utility maximizers whose utility is a function of wealth generated by expected returns from investment. The expected returns are related to one attribute relative to the expected return on market portfolio. The model is given by:

E[r_j] = r_f + b_j(E[r_m] - r_f)

where

E[r_j] = expected rate of return on security j

E[r_m] = expected rate of return on market portfolio

r_f = rate of return on a risk-free security, i.e., interest rate

b_j = Cov(r_j,r_m)/Var(r_m)

(E[r_m] - r_f) = market risk premium

 

The CAPM Assumptions

1. Perfect Competition: All investors behave as if they have no market power over prices.
2. Frictionless Markets: There are no transaction costs, taxes, or restrictions on security trading. In addition, all assets and securities are infinitely divisible and marketable.
3. Homogeneous Beliefs: All investors have homogeneous prior belief and Bayesian expectations and receive the same relevant information sets that affect market prices.
4. Individual Preferences: All investors care only about the risk-expected return tradeoff.
5. Individual Rationality: All investors are rational-expectations utility maximizers.

The CAPM Implications

 

The b_j is a sensitivity measure of how the expected security return co-moves with the expected return on the market portfolio. Since the market risk premium is positive because of the risk-return tradeoff, the movement of security return perfectly matches with the movement of the market return if b_j is 1.0 . If it is greater (less) than 1.0, security return is expected to move faster (slower) than the market return. However, it should be noted that CAPM does not require a linear relationship between the two returns. The expected return on security j derived from CAPM is related to the b_j through the characteristic line called the security market line (SML) instead of being related to its variance or standard deviation through the capital market line (CML) as characterized by the MVE. One can assume investors to have preferences more than the market risk premium when deriving their expected security return. Relaxation of the basic CAPM's assumptions have been examined and modeled by many financial economists including:

1. The heterogeneous-expectations version of CAPM by Lintner (1969)
2. The CAPM with taxes implication by Brennan (1970)
3. The zero-beta CAPM by Black (1972) assuming that riskless borrowing is not possible
4. The non-marketable (i.e., human capital) variation of CAPM by Mayers (1972)
5. The intertemporal CAPM by Merton (1973) to include investors' liquidity preference
6. The skewness-preference CAPM by Kraus and Litzenberger (1976)
7. The consumption CAPM by Breeden (1979) based on aggregate real consumption rate
8. The utility-based CAPM by Brown and Gibbons (1985) for testing Breeden's model
9. The contingent-claims CAPM by Lo (1986) based on the contingent claims analysis (CCA)
10. The behavioral CAPM by Statman and Shefrin (1994) incorporating the noise traders' beliefs

The Arbitrage Pricing Theory (APT) Back to Top

 

Ross (1976) revisits the factor model and employs its approach to develop an equilibrium asset pricing theory, which requires fewer assumptions than the CAPM. He assumes that the expected security returns are generated by multiple k factors instead of one pervasive market risk premium factor identified in the CAPM. The model constructed from the APT is given by:

 

r_j = E[r_j] + b_j1l₁ + b_j2l₂ + ... + b_jkl_k + e_j

E[r_j] = b_j0l₀ + b_j1l₁ + b_j2l₂ + ... + b_jkl_k

where

r_j = random rate of return on security j

E[r_j] = expected rate of return on security j

l_i = value of the ith factor, i = 0,1,2,...,k

b_ji = sensitivity of the expected return to the ith factor

b_j0 = unity value since l₀ is set to be a common factor for all securities

e_j = random residual return on security j

E[e_j] = zero

 

The APT Assumptions 

1. The value of l₀ is set to equal the risk-free rate to ensure the no-arbitrage equilibrium.
2. Factors identified are approximately linearly related to the expected security return.
3. The covariances with and among the factors can be positive, negative, or zero.
4. Markets are perfectly competitive, frictionless; and assets are infinitely divisible.
5. Individual investors have homogeneous beliefs and are rational-expectations utility maximizers with unconstrained preferences, i.e., investors are risk-neutral.
6. The random residual returns on n securities are independent of the factors with their expectations equal to zero.

The APT Implications

 

Since the APT has no restriction about investor preferences, it is more difficult to identify the pervasive factors that affect significant numbers of securities and less powerful to provide associated relationships between expected returns and security attributes than the CAPM. However, it combines the benefit of having the common intercept factor, i.e., the risk-free rate, of the CAPM and the flexibility of identifying more than one attribute of the factor model to generate the desirable expected returns. Nevertheless, the relationship between multiple betas and the sensitivities to their factors must be determined. The estimate of the market beta for a security is given by:

 

b_j = b_j1bF₁ + b_j2bF₂ + ... + b_jkbF_k

where

b_j = beta of security j to the change in market portfolio return

bF_i = beta of the ith factor to the change in the market return, i = 1,2,...,k

b_ji = sensitivity of bj to bFi

 

If expected returns are generated by a k-factor model, b_j in the CAPM becomes the terms on the right-hand side of the above equation. Thus, the k-factor CAPM is given by:

 

E[r_j] = r_f + (E[r_m] - r_f)(b_j1bF₁ + b_j2bF₂ + ... + b_jkbF_k)

E[r_j] = r_f + b_j1{(E[r_m]-r_f)bF₁} + b_j2{(E[r_m]-r_f)bF₂} + ... + b_jk{(E[r_m]-r_f)bF_k}

 where

r_f = l₀

(E[r_m]-r_f)bF_i = l_i, i = 1,2,...,k

 

Chamberlain and Rothschild (1983) offer a generalization of the APT with an approximate factor structure with the no-arbitrage assumption being replaced by the assumption that the price functional (p) is continuous. Reisman (1988) provides a simple proof of the generalized APT using the Hahn Banach theorem when the security payoffs have an approximate factor structure. He concludes that the general formulation of the APT which allows an infinite dimensional factor space may be use in the extension of the theory to the dynamic case. Even with Reisman's proof, the use of the mathematical theory of Banach is still far beyond layperson's understanding. Until recently, Shanken (1992) supplies an alternative proof of Reisman's result.

 

The conclusion of Reisman's proof is that the infinite space of expected returns is approximately equal to some linear function of the securities' factor betas. Shanken attempts to show that if the betas on factors (F) are approximately proportional to the betas on a proxy (P), then expected returns are approximately equal to a linear function of the betas on P. The factor representation for the first N security returns is given in vector form as:

 

r_N = a*_N + b_NF + e_N

 

where

E[e_N] = Cov(e_N,F) = 0

 

Let C_N be the N x N covariance matrix for e_N and let u be an upper bound on the eigenvalues of C_N. By the approximate factor model assumption, u can be taken to be independent of N.

 

Theorem: If returns conform to an approximate factor structure and there is no arbitrage, then expected returns are approximately linear in the betas on any proxy that is correlated with the factor.

 

Lemma 1: Let e_p be the residual from a regression of P on F and a constant, and let d_N be the N-vector of covariances between e_p and the components of e_N. Then d__Nd_N £ Var(e_p)u.

 

Lemma 2: If returns conform to an approximate factor structure, then the betas on P are approximately proportional to the betas on F. If P and F are correlated, then the betas on F are approximately proportional to the betas on P.

 

Empirical Studies in Portfolio Theory, Factor Models, CAPM, and APT Back to Top

Many of the empirical studies and testing conducted in the areas of portfolio diversification, factor models, CAPM, and APT have been subject to mixed sentiments and criticisms. On the applications of portfolio theory, there have been strong advocacy for diversification across assets with a wide range of risk-classes (i.e., mixed assets) as well as across national borders (i.e., multinational and global assets). As noted by Friedman (1971), diversifying portfolios in both financial and real assets (e.g., real estate) can further improve portfolio efficiency. Curcio (1983), and Webb, Curcio, and Rubens (1988) investigate the risk-reduction gains from mixed-asset diversification and find that the correlation between the returns on financial assets and real estate is less than the correlation among the returns on the assets within either class alone. Grubel (1968) observes that diversification of portfolios across countries results in further reduction in portfolio risk. Indirect cross-border diversification through multinational firms' stocks can also reduce systematic risk, as suggested by Errunza and Senbet (1981) and Adler and Dumas (1983). Both dimensions of portfolio diversification can be combined to form the international/global mixed-asset portfolios in order to drive down systematic risk even further as suggested by Curcio and Ziobrowski (1991). By using the data on financial assets including T-bills and bonds, corporate bonds common stocks and three types of real estate (commercial, residential, and farmland) for the U.S., the U.K. and Japan from 1973 to 1987 with returns adjusted to reflect exchange rate variations but omitting taxes, transaction costs, hedging gains, and short-selling restriction, they find that, over long horizon, the global mixed-asset portfolios perform the best trailing by domestic (U.S.) mixed and foreign financial assets, domestic and foreign financial assets, domestic mixed assets, and domestic financial assets, respectively.

 

In the area of factor models, two types of factor analysis are differentiated: covariance factors and expected returns factors. The first involves the time-series autocorrelational studies between specific stock returns and their event factors, e.g., macrofactors, microfactors, and industry factors. The second deals with the cross-sectional returns variability due to firm's specific factors such as risk, liquidity, price level, growth potential, and technical factors. There are attempts to empirically identify relevant factors and estimate the associated values as follows. King (1966) uses market and industry factors while Feeney and Hester (1967) employ principal components analysis to identify cross-sectional factors. Elton and Gruber (1973) estimate the dependence structure of stock prices. Fama and MacBeth (1973) attempt to test expected returns factor model for equilibrium setting. Farrell (1974) analyzes covariance of stock returns to determine the groupings of homogeneous stocks. Rosenberg and Marathe (1975) use multifactor model to estimate and predict systematic and residual risks. Arnott (1980) relies on cluster analysis to identify covariance factors that determine stock price movement. Sharpe (1982) identifies expected returns factors in the NYSE using multifactor model. Estep, Hanson, and Johnson (1983) attempt to find the multiple factors which determine both stock prices and risk. Rosenberg, Reid, and Lanstein (1985) identify book-to-price ratios as another pervasive factor beside market risk premium which influence stock returns. Fama and French (1992, 1995, 1996) add size factor and book-to-price ratios to the CAPM and propose it as the three factor model to explain most of the market anomalies. Haugen and Baker (1996) employ all firm's specific factors in their cross-sectional, cross-national factor model and conclude that they are common to all stocks in all major equity markets around the world.

 

As Roll (1977) argues on the CAPM's front, the definitive tests of CAPM can never be performed simply because it is intended to be an ex ante, not ex post, model. Moreover, the definition and measurement of the market portfolio is still doubtful in the sense that whether or not other low-liquidity and non-domestic assets such as real estate, foreign assets, commodity, and collectible items should be or are actually included. His argument is supported by the fact that, even if capital markets are efficient and the CAPM is valid, the cross-sectional SML cannot be used as a means to measure the ex post performance of portfolio selection models as a result of the following realities:

1. Because of various actual trading restrictions such as unlimited riskless borrowing and short-selling, investors may not be able to hold MVE frontier portfolios.
2. Due to the fact that security returns are not normally distributed, investors may have skewness preferences and end up holding inefficient portfolios.
3. Since transaction costs and taxes affect security returns, investors who face different costs may take inefficient portfolio positions gross of cost.
4. Investors may suboptimize their portfolio positions when holding indivisible assets such as their own human capital (i.e., the present value of their future earnings).

To perform the ex post tests, the expectations models must be transformed into the realizations models both in the time-serial and cross-sectional forms:

 

Time-Serial Form:

r_jt - r_ft = (r_mt - r_ft)b_j + e_jt

where

e = actual residual returns

 

Cross-Sectional Form:

r_pt = l₀ + l₁b_p + e_pt

where

p = portfolio

 

The tests of CAPM are chronologized as follows. The direct test of CAPM is first conducted by Douglas (1969) who finds that the time-serial average returns on securities over seven periods of five years from 1926 to 1960 are positively related to their variances. Lintner (1969) uses the cross-sectional data from 1954 to 1963 to test the variance of residual returns and observes that the risk-free rate is higher and the SML is flatter than expected by the CAPM. Another cross-sectional study is conducted by Blume and Friend (1970, 1973) on portfolios of stocks listed on the NYSE from 1960 to 1968. Using the Jensen, Treynor, and Sharpe ratios, they find that low-beta stocks tend to be undervalued while high-beta stocks overvalued. Lintner's study is retested by Miller and Scholes (1972) using different data set with similar conclusions. Black, Jensen, and Scholes (1972) examine the average returns on ten NYSE portfolios from 1931 to 1965 and find a linear relationship between risk and return. Fama and MacBeth (1973) increase the number of portfolios to twenty and the time span from 1926 to 1968 using five-year sub-periods. They obtain similar results to those of Black, Jensen, and Scholes with additional findings that the betas are non-linear.

 

Other CAPM tests are conducted in conjunction with the tests of markets efficiency under their joint hypothesis. Basu (1977) tests the low price-earnings-ratio stocks based on monthly returns on portfolios from 1957 to 1971 and find them to outperform the CAPM's forecast. Banz (1981) examines returns on stock portfolios together with the size effect and find that firm size had negative relationship with returns. Reinganum (1981) studies stock market returns to earnings announcements between 1975 and 1977 and concludes that portfolios based on size and earnings-price ratios exhibit abnormal monthly returns that persist for at least two years. Levy (1978) and Mayshar (1979, 1981, 1983) propose a non-CAPM explanation for size effect. Rosenberg, Reid, and Lanstein (1985) look at monthly stock returns from 1973 to 1980 and show that book-to-price ratios and return reversals help explain stock returns performance. Lakonishok and Shapiro (1986) test the twenty-year monthly stock returns from 1962 to 1981 using the same technique as Fama and MacBeth and find that neither beta nor variance is significant in predicting returns.

 

Most notably, Fama and French (1992) examine stock returns from two sample periods from 1963 to 1990 and 1940 to 1990 using 100 portfolios from NYSE, AMEX, and NASDAQ and find that size and book-to-price ratios are related to monthly returns but beta is not related when controlled for size factor. Their conclusion is that size and book-to-price ratios can be the proxies for non-market risk factor and that returns in excess of market risk premium can be the results of investors' irrational overreaction. Lakonishok, Shleifer, and Vishney (1994) test the contrarian strategies and conclude that value stocks outperform growth and glamour stocks. Kothari, Shanken, and Sloan (1995) examine a cross-section of annualized expected returns from 1927 to 1990 and find a significant relationship between beta and returns, and a weak relationship between book-to-price ratios and returns. Fama and French (1995) attempt to explain the size and book-to-price ratios anomalies and find that both of them are persistently related to earnings, while size is more highly related to annual returns than book-to-price ratios. Jaganathan and Wang (1993) argue that betas vary over the business cycle and develop a three-beta model. Fama and French (1996) propose a three factor model which can explain most market anomalies found by CAPM and provide its explanation that multiple common sources of variation in returns are necessary, that investors' irrational behavior may not cause anomalies, and that spurious betas result from survivor bias, data snooping, and bad market proxies. Haugen and Baker (1996) suggest two alternative approaches to tackle market anomalies by using the multiple factor model and the behavioral model. For the first approach, forty-one factors are grouped into six firm characteristics namely, 10 risk factors, 4 liquidity factors, 10 price factors, 9 growth potential factor, 7 technical factors, and 1 sectoral factor. The results are that 1) stocks with higher expected and realized rate of return are lower in risk than stocks with lower returns, 2) important determinants of expected stock returns are common to the major equity markets around the world, and 3) the efficient markets hypothesis should be rejected.

 

Tests of APT, though occurred about ten years later than those of CAPM, are less controversial because the APT itself requires no assumptions about returns distribution, investor preferences, and market portfolio, while being able to predict relative pricing of any subset of securities. Roll and Ross (1980) embark on empirical investigation of APT by looking at daily returns on NYSE and AMEX stocks between 1962 and 1972. After correcting for dependence between mean and standard deviation that lognormality causes, they find that total variance of returns does not add explanatory power to the model. They also find the same result for the intercept term which proves that APT is robust and should not be rejected. Reinganum (1981) forms portfolios of like factors based on previous years' returns and finds excess returns. His conclusion is to reject the APT. Shanken (1982) questions the testability of the APT that 1) since it requires a large number of securities to approximate returns, it may not work on smaller samples, and 2) since the basic construct of APT is the linear factor model, it means that different grouping of stocks may result in different slope coefficients which would be inconsistent. Chen (1983) groups stocks into high-market value and low-market value groups and finds that grouping based on market value is not significant. He also compares the results with those from CAPM and finds that APT explains significant part of CAPM's residual but not vice versa. Chen, Roll, and Ross (1983) use four macroeconomic factors in the APT, i.e., index of industrial production, ratio of inflation, yield spread between T-bonds and BB corporate bonds, and yield spread on short-term and long-term bonds. Connor (1983) tests the Nash or competitive equilibrium APT and finds it can eliminate the need to assume an infinite number of securities for the market portfolio. Roll and Ross (1984) critically reexamine the empirical evidence on the APT and argue that it is testable and can be use in strategic portfolio planning. Dybvic and Ross (1985) address Shanken's critique by noting that returns explained by factor structure in actual securities is much different from returns explained by factor structure in arbitrary portfolios. Wei (1988) uses Nash equilibrium APT by adding market portfolio as another factor which effectively combines CAPM and APT. Robin and Shukla (1991) examine the monthly returns from 1976 to 1985 using the Chicago's Research on Security Prices (CRSP) data and find that pricing errors and variance are high and statistically significant. Chatterjee and Pari (1990) use a non-parametric method to estimate factors for APT and find that the number of factors increase linearly with the number of securities. Shukla and Trzcinka (1990) employ principal components method to identify the factors and find that 40% of variation in returns can be explained in a five-dimensional APT model. Mei (1993) use the data from 1963 to 1990 and find that APT can explain size and dividend yield but fails to explain book-to-price ratios and price-earnings ratios.

 

Conclusion Back to Top

It is overwhelm to give a precise conclusion about modern investment theory amid its structural successes and empirical flaws. Yet, without theoretical grounds to be contrasted against, we would by no means be realized that there are no simple ways to explain the actual behavior of financial assets' prices and returns. Modern investment theory should be perceived as the indispensable building block for every academician and practitioner to start off, but not to be engrossed, with. Our judgement should be based upon benefit/cost analysis it contributes/incurs to our understanding about and stakes in the real-world investment phenomena. Academically, as Friedman (1953) suggests, theories should be judged not on the basis of their assumptions but rather on the validity of their predictions. Practically, the techniques of modern investment theory have enabled individuals to make their investment decisions wisely with clear purposes and criteria. Thus, it depends on which of the two relative stances we take in order to justify the trade-offs between adopting and rejecting the modern investment theory.

Today, we have quite enough empirical evidence to comfortably reject most of the works done during the last forty years. But, such rejection should not mislead us to favor other alternative paradigms altogether for they are also subject to the same destiny as the current ones. As Sharpe (1984) concludes, "While the relative importance of various factors changes over time, as do the preferences of investors, we need not completely abandon a valuable framework within which we can approach investment decision methodically. We have developed a useful set of tools and should certainly continue to develop them. Meanwhile, we can use the tools we have, as long as we use them intelligently, cautiously, and humbly."

---

References

Adler, M. and B. Dumas (1983) International Portfolio Choice and Corporation Finance: A Synthesis, Journal of Finance.

Banz, R.W. (1981) The Relationship between Return and Market Value of Common Stocks, Journal of Financial Economics.

Basu, S. (1977) The Investment Performance of Common Stocks in Relation to Their Price-Earnings Ratios, Journal of Finance.

Breeden, D.T. (1979) An Intertemporal Asset Pricing Model with Stochastic Consumption and Investment Opportunities, Journal of Financial Economics.

Brennan, M. (1970) Taxes, Market Valuation and Corporate Finance Policy, National Tax Journal.

Black, F. (1972) Capital Market Equilibrium with Restricted Borrowing, Journal of Business.

Black, F., M.C. Jensen, and M. Scholes (1972) The Capital Asset Pricing Model: Some Empirical Tests, Jensen, ed., Studies in Theory of Capital Markets Praeger, New York.

Chamberlain, G. and M. Rothschild (1983) Arbitrage, Factor Structure, and Mean-Variance Analysis on Large Asset Markets, Econometrica.

Chen, N.F. (1983) Some Empirical Tests of the Theory of Arbitrage Pricing, Journal of Finance.

Chen, N.F., R. Roll, and S.A. Ross (1983) Economic Forces and the Stock Markets, Unpublished Manuscript. Yale University, New Heaven, Connecticut.

Curcio, R.J. (1983) Diversification Gains from Including Real Estate in Mixed-Asset Portfolios, Paper presented at the Financial Management Association Annual Meeting.

Elton, E. and M.J. Gruber (1987) Modern Portflio Theory and Investment Analysis. John Wiley & Sons, New York, NY.

Errunza, V.R. and L.W. Senbet (1981) The Effects of International Operations on the Market Value of the Firm: Theory and Evidence, Journal of Finance.

Fama, E. and K. French (1992) The Cross-Section of Expected Stock Returns, Journal of Finance.

__________. (1995) Size and Book-to-Market Factors in Earnings and Returns, Journal of Finance.

__________. (1996) Multifactor Explanations of Asset Pricing Anomalies, Journal of Finance.

Fama, E. and J. MacBeth (1974) Tests of Multiperiod Two Parameter Model, Journal of Political Economy.

__________. (1973) Risk, Return and Equilibrium: Empirical Tests, Journal of Political Economy.

Friedman, M. (1953) The Methodology of Positive Economics, Essays in Positive Economics. University of Chicago Press, Chicago, Illinois.

Friedman, H.C. (1971) Real Estate Investment and Portfolio Theory, Journal of Financial and Quantitative Analysis.

Grubel, H.R. (1968) International Diversified Portfolios: Welfare Gains and Capital Flows, American Economic Review.

King, B.F. (1966) Market and Industry Factors in Stock Price Behavior, Journal of Business.

Kraus, A. and R. Litzenberger (1976) Skewness Preference and the Valuation of Risky and Assets, Journal of Finance.

Lakonishok, J, A. Shleifer, and R. Vishny (1994) Contrarian Investment, Extrapolation, and Risk, Journal of Finance.

Levy, H. (1983) The Capital Asset Pricing Model: Theory and Empiricism, Economic Journal.

Lintner, J. (1965) The Valuation of Risk Assets and the Selection of Risky Investment in Stock Portfolio and Capital Budgets, Review of Economics and Statistics.

__________. (1969) The Aggregation of Investor's Diverse Judgements and Preferences in Purely Competitive Markets, Journal of Financial and Quantitative Analysis.

Lo, A.W. (1986) Statistical Tests of Contingent-Claims Asset-Pricing Models, Journal of Financial Economics.

Markowitz, H.M. (1952) Portfolio Selection, Journal of Finance.

__________. (1959) Portfolio Selection: Efficient Diversification of Investment, John Wiley & Sons, New York, NY.

Merton, R.C. (1973) An Intertemporal Capital Asset Pricing Model, Econometrica.

Mossin, J. (1966) Equilibrium in a Capital Market, Econometrica.

Reinganum, M.R. (1981) Misspecification of Capital Asset Pricing: Empirical Anomalies Based on Earnings Yields and Market Values, Journal of Financial Economics.

Reisman, H. (1988) A General Approach to the Arbitrage Pricing Theory, Econometrica.

Roll, R. and S.A. Ross (1980) An Empirical Investigation of the Arbitrage Pricing Theory, Journal of Finance.

Ross, S.A. (1976) The Arbitrage Theory of Capital Asset Pricing, Journal of Economic Theory.

Shanken, J. (1982) The Arbitrage Pricing Theory: Is It Testable?, Journal of Finance.

__________. (1992) The Current State of the Arbitrage Pricing Theory, Journal of Finance.

Sharpe, W.F. (1963) A Simplified Model of Portfolio Analysis, Management Science.

__________. (1964) Capital Asset Prices: A Theory of Market Equilibrium, Journal of Finance.

__________. (1984) Factor Models, CAPM, and The APT, Journal of Portfolio Management.

Statman, M. and H. Shefrin (1994) Behavioral Capital Asset Pricing Theory, Journal of Financial and Quantitative Analysis.

Webb, J.R., R.J. Curcio, and J.H. Rubens (1988) Diversification Gains from Including Real Estate in Mixed-Asset Portfolios, Decision Sciences.

Back to Top

* Worapot Ongkrutaraksa is a lecturer in Finance and Strategic Management at Maejo University's Faculty of Agricultural Business, Chiang Mai, Thailand. He used to conduct his post-graduate research in financial economics at Kent State University and international political economy at Harvard University through the Fulbright sponsorship between 1995 and 1998.

E-mail: [email protected]

Home Page