i An update to this article is included at the end

Heliyon 6 (2020) e05185

Contents lists available at ScienceDirect

Heliyon
journal homepage: www.cell.com/heliyon

Research article

An augmented capital asset pricing model using new

macroeconomic determinants
Chinh Duc Pham a, b, Le Tan Phuoc a,b,c, *
a
University of Economics and Law, Ho Chi Minh City, Viet Nam
b
Viet Nam National University, Ho Chi Minh City, Viet Nam
c
Becamex Business School - Eastern International University, Viet Nam

A R T I C L E I N F O A B S T R A C T

Keywords: Using the interview results of 26 experienced scholars, managers, and professional stock traders in conjunction
Asset pricing with findings of recent studies in economics, we proposed an augmented asset pricing model using the macro-
Corporate finance economic determinants representing the macroeconomic state variables to explain the nexus between these risks
Interest rate
and the U.S. stock returns. This non-traded factor model (MAPM) is inspired by and based on the macroeconomic
Government long-term bond rate
Exchange rate
theory and models and consists of the market return, U.S. prime rate, U.S. government long-term bond rate, and
Financial economics exchange rate of USD/EUR as in Eq. (1). Using the Bayesian approach (via two Bayes and t.Bayes estimators) and
Money monthly returns of the S&P 500 stocks from 2007- 2019, our results showed the MAPM consistently yielded a
Pricing statistically significant greater forecasting, explanatory power, and model adequacy compared to the most used
Macroeconomics capital asset pricing model (CAPM) in practice. Interestingly, our study found and confirmed (t-statistic > 3) that
Econometrics the last two macroeconomic determinants have a statistically significant positive effect on the stock returns, which
also supports the MAPM. These findings suggest the MAPM is a more efficient and advantageous model compared
to the CAPM. So, practitioners would be better off employing the MAPM over CAPM in practice and research.

1. Introduction Phuoc, 2018; Zhang, 2017). However, that study does not explore and
capture other risks such as the firm's financial ratios and investment,
The simplest and most common benchmark asset pricing model in stock momentum, and macroeconomic risks.
both practice and research is single-period and one-factor capital asset Other researchers headed in different directions. They attempted to
pricing model (CAPM, Sharpe, 1964) as shown in a report of survey re- verify that the relationship between the real stock returns and the firm's
sults by a professional organization (The Association for Financial Pro- financial traits existed. Some very well-known and influential studies in
fessionals, 2013), a report by a leading financial services company (The this direction (Banz, 1981; Basu, 1983; Bhandari, 1988; Fama and
Credit Suisse, 2013), and recent studies (Barillas and Shanken, 2018; French, 1992, 1993; 1995; Rosenberg et al., 1985) showed that, besides
Barillas et al., 2019; Chib et al., 2020). However, the CAPM possesses the market risk (beta), the stock returns also depended on the firm's
limitations in both research and practices (Barillas and Shanken, 2018; market equity, leverage, book-to-market equity, earnings-to-price, and
Gungor and Luger, 2019; Lee, 2019; Zhang, 2017, 2019). Therefore, size. Using these findings, Fama and French (1992, 1993, 1995, 1996a)
researchers have augmented or propose alternatives to the CAPM. Jensen proposed the three-factor asset pricing model (FF3) consisting of the
(1967) is the very first study that modified the CAPM. That study pro- market and two traded factors: the SMB (size) and HML (value). Also,
posed a two-coefficient model by adding an intercept coefficient alpha Fama & French (1992, 1993, 1995, 1996a) showed that the FF3 yielded
(α) to the original CAPM to represent the stock's expected excess return greater beta and standard deviation compared to the original CAPM.
when the market risk premium is zero (α equals zero in an efficient Hence, the authors claimed that the FF3 could explain more of the
market). Recent studies confirmed the alpha existed for the real stocks relationship between the market and stock returns than the CAPM. After
(Barillas and Shanken, 2017; Fama and French, 1996b, 2015; 2018; Hou the FF3 was published, other studies (Aharoni et al., 2013; Fama and
et al., 2015, 2020b; Hou et al., 2019, 2020a; Pham and Phuoc, 2020; French, 2006, 2008; Novy-Marx, 2013; Titman et al., 2004) showed that

* Corresponding author.
E-mail address: [email protected] (L.T. Phuoc).

https://doi.org/10.1016/j.heliyon.2020.e05185
Received 13 March 2020; Received in revised form 9 June 2020; Accepted 5 October 2020
2405-8440/© 2020 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
C.D. Pham, L.T. Phuoc Heliyon 6 (2020) e05185

the stock returns also depended on the firm's profitability and invest- growth. The authors showed that this three-factor model outperformed
ment. Using this new finding, Fama and French (2015, 2016, Fama et al., both the FF3 and CAPM in terms of explanatory power. Another study,
2017) augmented the FF3 with two more factors: the RMW (profitability) using the q-theory of investment, Hou et al. (2015) argued for an
and CMA (investment). This model considers the five-factor asset pricing empirical four-factor q-model (q4 model) including the market, ME
model (FF5). Fama and French (2015, 2016, Fama et al., 2017) also (size), 1/A (investment), and return-on-equity (profitability) factors. The
demonstrated that the FF5, in general, performed better than the FF3 in authors showed that the q-factor model largely summarizes the
explaining the stock returns since the FF5's GRS statistic (Gibbons et al., cross-section of average stock returns. Also, in many cases, the q-factor
1989) is less than the FF3's. Additionally, Fama & French (2015) also model outperformed the FF3 and Carhart (1997) 4-factor model in
pointed out that the HML factor was redundant in this model. Once again, capturing the significant anomalies. Similarly, Hou et al. (2019) revised
Fama and French (2018) augmented the FF5 with one more factor, the the q4 by adding one more factor, the expected investment growth. Their
UMD (momentum) using the findings of momentum factor-affected stock new model, the q5 (Hou et al., 2020a,b), yielded strong explanatory
returns in the literature (Asness and Frazzini, 2013; Asness et al., 2013; power in the cross-section and outperformed the q4 , FF5, FF6, and the
Barroso and Santa-Clara, 2015; Carhart, 1997; Jegadeesh and Titman, Barillas and Shanken (2018)' six-factor model in terms of maximum
1993; Moskowitz et al., 2012; Stambaugh and Yuan, 2017). This new Sharpe ratio.
model considers the six-factor asset pricing model (FF6). Also, Fama and One of the three parameters of the CAPM is the market risk premium,
French (2018) showed that the FF6 yielded a smaller GRS statistic or but the market is not clearly stated. In reality, there are many different
higher max squared Sharpe ratio (Barillas and Shanken, 2017) compared markets and these markets have different performances, especially the
to the FF5. So, the authors claimed that FF6 performed better than the markets locate in different countries. To make matters worse, many firms
FF5 in explaining the stock returns. – especially multinational firms – are affected by the exchange rate risks.
Overall, the FF3, FF5, and FF6 are more flexible than the CAPM, but Hence, some studies (Adler and Dumas, 1983; Agmon, 1972; Grauer et
they have big drawbacks for individual investors to apply in practice al., 1976; Lessard, 1974, 1976; Solnik, 1974a, 1974b) proposed and
since these models are more complex due to the facts that they are traded worked on the international asset pricing model (IAPM) by adding fac-
factor models; the data are not always available, especially for the tors to the CAPM related to the exchange rate risks, such as the exchange
monthly or daily data. Also, the shortcomings of CAPM's assumptions are rate risk premium. Generally, in the real world, IAPM is more flexible
seen in the FF3, FF5, and FF6 models. In addition, Fama and French compare to the CAPM. Therefore, it is a good theoretical model in asset
(1996a) claimed and showed that the FF3 yielded greater beta and pricing – but not easily applied in practice by individual investors.
standard deviation, hence, more explanation of the relationship between Importantly, this model often yielded poor results in empirical research
stock returns and beta, compared to the CAPM. However, the greater beta as shown in some studies (Solnik, 1977; Wallingford and Bicksler, 1974).
standard deviation means the longer beta confidence interval, which Other studies attempted to examine the relationship between the
leads to the inefficient beta estimation, the main contribution of the stock returns and the macroeconomic state variables. Those studies
CAPM. Besides, other studies showed that the FF3 fails to account for a (Bower et al., 1984; Goldenberg and Robin, 1991; Roll and Ross, 1983,
wide array of asset pricing anomalies (Boons, 2016; Jegadeesh and Tit- 1984; Ross, 1976) developed and worked on the arbitrage pricing theory
man, 1993; Loughran and Ritter, 1995). Importantly, even Fama and (APT), a flexible but complex model, consisting of multiple
French (2004) admitted and other studies (Berk, 1995; Ferson et al., macro-economic risks such as the inflation rate, exchange rate, GNP
1999; Kim et al., 2011; Kothari et al., 1995; Lo and MacKinlay, 1990; growth rate, etc. Those studies showed that the APT outperformed the
MacKinlay, 1995; Wang and Wu, 2011) also pointed out that the traded CAPM in terms of explanatory power. However, there is not a consensus
factors, SMB, HML, and others employed in the FF3 (in FF5 and FF6 as among the researchers and individual investors of how many and what
well) do not have a solid background but brute-force ideas. So, these FF3, risks should be in the model (Dhrymes et al., 1984). Other studies
FF5, and FF6 are just ad-hoc models (Hou et al., 2019). Also, the FF5 is (Latham, 1989; Shanken, 1982) showed that the APT does not imply an
not driven by the valuation theory as claimed (Hou et al., 2019). Also, the exact linear risk-return relation and is very hard to test empirically.
HML (value), RMW (profitability), and CMA (investment) factors were Finally, Latham (1989) showed that the APT is not consistent with either
shown to be non-significant in explaining the stock returns (Fama and the single-period model or the multiperiod model.
French, 2015; Hou et al., 2015, 2020a,b; Kim et al., 2011; Kothari et al., Our study differs from other studies and contributes to the literature
1995; Kubota and Takehara, 2017). as follows. Firstly, we employed the qualitative research techniques using
Barillas and Shanken (2018) tried to examine the FF5 and q-factor interviews (Krueger et al., 2001; Opdenakker, 2006) of, totally, 26 ex-
model (Hou et al., 2015). Using the Bayesian approach and Sharpe ratio, perts of three different groups to explore possible economic determinants
their study confirmed the six-factor model (BS) using the market, in- affecting stock returns. The reason we interviewed three groups is that
vestment, return-on-equity (profitability), size, value, and momentum. we wanted to avoid bias in the results. Also, the number of experts that
Hence, this BS includes both traded and non-traded factors. This BS also employed in our qualitative study is consistent with Fang et al. (2019) –
showed that the size and momentum factors were not redundant factors six experts of two different groups, Kumar et al. (2017) – 15 experts of
as FF5 and q-factor models claimed, respectively. Roy and Shijin (2018) only one group, Raj and Sah (2019) – 10 experts of two different groups,
tried to extend the FF5 with the human capital component based on the and Wu et al. (2019) – 17 experts of two different groups. In fact, we
findings of other studies (Belo et al., 2014; Belo et al., 2017; Kim et al., interviewed 16 scholars with a minimum of seven years of teaching and
2011; Kuehn et al., 2017; Roy and Shijin, 2017). Again, the models research in the fields of economics and finance, 5 industry managers with
proposed by Barillas and Shanken (2018) and Roy and Shijin (2018) are a minimum of 5 years in seniority position, and 5 professional stock
more flexible compare to the CAPM, FF3, and FF5. However, they have traders. Based on the results of our interviews, the findings of recent
similar weaknesses as the FF3, FF5, and FF6. studies in the fields of economics and finance (Ang and Bekaert, 2007;
Due to the criticism of FF3 and other existing asset pricing models in Barinov et al., 2018; Bernanke and Gertler, 1999; Boons, 2016; Campbell
the literature, Kim et al. (2011) proposed a revised version of both FF3 and Yogo, 2006; Dai and Zhou, 2020; Dai and Zhu, 2020; Goyal and
and Jagannathan and Wang (1996)’ models. Their study argued that the Welch, 2008; Gungor and Luger, 2019; Kroencke, 2017; Lee, 2019), and
future labor income growth, a macroeconomic state variable, captures the availability of the related data, our study proposed adding three
the nature of economic risks that the size and value factors in the FF3 macroeconomic determinants/risks (the U.S. prime rate, the U.S. gov-
hope to depict. Also, the authors argued that future labor income growth ernment long-term bond rate, and the exchange rate of USD/EUR) to the
was a better factor than the current income growth (see Jagannathan and original CAPM to explain the nexus between the risks and the U.S. stock
Wang, 1996) to represent the return on human capital. Hence, their returns. These macroeconomic determinants are proxies for the
model consists of market, consumption growth, and future labor income

2
C.D. Pham, L.T. Phuoc Heliyon 6 (2020) e05185

macroeconomic state variables. The number of macroeconomics de- Finally, we present the results and interpretations in section 4. Conclu-
terminants (three) matched with other studies suggested (Bower et al., sions are also provided in section 5. The References are at the end.
1984; Goldenberg and Robin, 1991; Roll and Ross, 1983). This
augmented CAPM (hereinafter, the MAPM), a non-traded factor model as 2. The MAPM
in Eq. (1) includes the excess market return of the original CAPM and the
U.S. prime rate, the U.S. government long-term bond rate, and the ex- The MAPM utilized the simplicity, availability, and ease of accessed
change rate of USD/EUR. Importantly, this MAPM is inspired by and data and the flexibility of the CAPM, APT, and IAPM. This model also
based on the macroeconomic theory and models, a requirement for an indirectly incorporated the stock related characteristics of the FF5, q4 , q5 ;
asset pricing model as suggested by other studies (Hou et al., 2020a,b; and Roy and Shijin (2018)' models. Finally, Jensen's alpha also added to
Fama and French, 2018). Then, we employed the quantitative research the MAPM. Therefore, the MAPM showed the relationship between the
method using the Bayesian approach to confirm our proposed model, the excess return on the stock i, i ¼ 1, 2, 3, …, N at the time t, t ¼ 1, 2, …, n
MAPM. Also, only the S&P 500 stocks, the largest U.S stocks, were pur- and the market, prime rate, government long-term bond risk premiums,
posefully employed in this study to examine the performance of both the and exchange rate as follows:
MAPM and CAPM and the three proposed macroeconomic determinants.
The reason is to minimize the faulty anomalies claim with microcap Rit RFt ¼ αi þ βi ðRMt  RFt Þ þ γ i ðUSt RFt Þ þ ki ðLTBt RFt Þ þ λi EXt þ εit
stocks that discovered in a recent study (Hou et al., 2020a,b). (1)
Secondly, we found only one asset-pricing model study (Roy and
Shijin, 2018) in the literature employing the frequentist approach and where,
both parametric and non-parametric estimators. In contrast, our study
employed the Bayesian approach via both parametric and non-parametric  Rit : the return on the stock i at the time t,
Bayes estimators to obtain consistency in the results of the model  RMt : the return of the market portfolio at the time t,
comparisons.  RFt : the risk-free rate at the time t,
Thirdly, the recent influential studies in the literature employed  αi : Jensen's alpha coefficient (alpha) of the stock i,
either r-squared or adjusted r-squared to measure the model of fit in the  βi : the stock i's sensitivity to the market portfolio (beta),
model comparisons (Barillas and Shanken, 2017, 2018; Fama and  γ i : the interest risk coefficient (gamma) that the stock i is bearing,
French, 2016, 2018; Hou et al., 2015, 2019; 2020a,b; Zhang, 2017) even  USt : the U.S. prime rate at the time t,
though the r-squared or adjusted r-squared would easily increase just by  ki : the government long-term bond yield rate risk coefficient (kappa)
adding more independent factors in the model. This means the r-squared that the stock i is bearing,
and adjusted r-squared alone may not be sufficient to measure the model  LTBt : the government long-term bond rate at the time t,
fit. Hence, our study employed both adjusted r-squared (Bayesian  λi : the exchange rate risk coefficient (lambda) that the stock i is
r-square or R2B) and posterior mean deviance (D_bar), a component of bearing,
deviance information criterion (DIC), to measure the model fit/adequacy  EXt : the exchange rate of USD/EUR at the time t,
to obtain consistency in the results of our model comparisons.  εit : the random error term that has mean zero and variance σ 2
Fourthly, it is worth noting that the recent influential studies (Fama (Sigma2).
and French, 2016, 2018; Hou et al., 2019, 2020a,b; Zhang, 2017) did not
employ the model error in their model comparisons. Hence, our study To evaluate the MAPM, we examined and compared its performance
employed both the mean square error (MSE) and model variance against the most used asset pricing model in practice, the CAPM, using
(Sigma2) to measure the forecasting power and efficiency of the model both parametric and non-parametric Bayes estimators for consistency in
and their confidence intervals in our model comparisons. the results and both advanced and common statistical measures (as
Finally, unlike the recent asset-pricing model studies (Barillas and described in Methodologies). We also examined how the U.S prime rate,
Shanken, 2017, 2018; Fama and French, 2015, 2016, 2018; Hou et al., the U.S. government long-term bond rate, and the exchange rate of USD/
2019, 2020a,b), we employed both the confidence interval approach EUR affected the stock returns.
instead of p-value and test hurdle of the absolute t-statistic of both 2.78
and 3.0 in our asset pricing model comparisons (Dyckman, 2016; Dyck- 3. Data and Methodologies
man and Zeff, 2019; Halsey, 2019; Harvey et al., 2015; Hou et al., 2020a,
b; Lewellen et al., 2010; Wasserstein and Lazar, 2016). 3.1. Data
Using the S&P 500 stocks from 2007- 2019, the empirical results
showed the MAPM consistently yielded a statistically significant lesser Only the S&P 500 stocks, the largest U.S stocks, were purposefully
model error and greater model fit/adequacy compared to the CAPM. In selected due to their efficiency to examine the performance of both the
other words, in asset pricing work, the MAPM yielded greater fore- MAPM and CAPM and how the three macroeconomic determinants affect
casting, explanatory power, and model adequacy compared to the CAPM. the stock returns. Besides, we wanted to avoid bias in the data due to the
Our empirical results also found and confirmed (t-statistic > 3) that the most recent financial crisis that had a strong negative effect on both the
last two macroeconomic determinants, the U.S. government long-term macroeconomics and stock market, especially the financial stocks as
bond rate, and the exchange rate of USD/EUR), have a statistically sig- shown in two studies (Bullard et al., 2009; Smaga, 2014). So, the
nificant positive effect on the stock returns. These findings suggest the medium-horizon monthly returns of the 450 S&P 500 stocks and three
MAPM is a more efficient and advantageous asset pricing model macroeconomic determinants from 2007- 2019 were collected from the
compared to the CAPM. So, our findings may help both the policy-makers Federal Reserve Economic Data (FRED). The risk-free rate was the
and investors to draft their decisions in monetary policy and investment, three-month U.S. Treasury secondary market rate. The S&P 500 index
respectively. considered the market because it is widely considered the best gauge of
The remainder of this paper is organized as follows. Section 2 de- large-cap U.S. stocks.
scribes the MAPM. Section 3 is the Data and Methodologies. In this
section, we provide the details of the data. We also present and reason for 3.2. Methodologies
the Bayesian approach and both parametric and non-parametric Bayes
estimators used in this study. Then, we set up the benchmarks and 3.2.1. The approach and estimators
rationalize both the proposed benchmarks and confidence interval Our study employed a Gibbs sampler, a Markov chain Monte Carlo
approach in the model comparisons between the CAPM and MAPM. (MCMC), on the real data and the Bayesian approach (via two Bayes

3
C.D. Pham, L.T. Phuoc Heliyon 6 (2020) e05185

estimators and weakly informative normal priors) as in our previous stocks) shows the support for use of the CAPM in both practice and
study (Pham and Phuoc, 2020; Phuoc and Pham, 2020). research. This is consistent with other studies (The Association for
Financial Professionals, 2013; Barillas and Shanken, 2018; Barillas et al.,
3.2.2. Benchmarks 2019; Chib et al., 2020; Da et al., 2012; Fama and French, 1996b).
We evaluated and compared the performance of the MAPM against In contrast with the CAPM, we assumed that the MAPM does not hold
the CAPM using the following benchmarks: i) the model error (MSE and for a stock if all coefficients beta, gamma, kappa, and lambda from Eq. (1)
Sigma2) since these statistics showed the model forecasting power and were all zero. The MAPM using the Bayes estimator yielded the 95%
precision, respectively. The model with a lower MSE and Sigma2 would confidence interval (not shown here) of the beta, gamma, kappa, and
be a preferred model in practice. ii) The second benchmark was the lambda of six stocks including zero (those six stocks were the same stocks
model fit/adequacy (R2B and posterior mean deviance (D_bar)) since as in the case of the CAPM using Bayes estimator). However, the MAPM
they provided information about the explanatory power of the model and using the t.Bayes estimator yielded the 95% confidence interval (not
model adequacy. The model with a greater R2B and/or lower D_bar shown here) of the beta, gamma, kappa, and lambda of only one stock
would be a preferred model (Pham and Phuoc, 2020; Phuoc and Pham, including zero (this stock is one of the six stocks in the case of MAPM
2020; Spiegelhalter et al., 2002, 2014; Van der Linde, 2005). using the Bayes estimator). These findings mean that the MAPM might
In model comparisons, we employed the mean and 95% confidence not hold for only six stocks (1.3% of big U.S. stocks). Therefore, we could
interval of the mean difference of benchmarks as suggested and claim that in asset pricing work, the MAPM worked with more U.S stocks
employed by other studies (Dyckman, 2016; Dyckman and Zeff, 2019; compared to the CAPM.
Halsey, 2019; Lewellen et al., 2010; Pham and Phuoc, 2020; Phuoc and
Pham, 2020). 4.1. The model errors

3.2.3. The effect of the U.S. prime rate, the government long-term bond rate, 4.1.1. The MSE
and the exchange rate of USD/EUR on the S&P 500 stock returns Panels (a) and (b) of Figure 2 showed that the differences between
Furthermore, we evaluated the MAPM by looking at how the U.S two MSE's of the CAPM and MAPM were distributed on both sides of
prime rate, the U.S. government long-term bond rate, and the exchange zero. However, Table 1 illustrates the mean difference and 95% confi-
rate of USD/EUR affected the stock returns. If these three factors were dence interval of the mean difference between CAPM and MAPM in terms
found significantly affecting the stock returns, then it would provide of MSE using the Bayes estimator of 0.749 and (0.018, 1.480), respec-
more evidence supporting the MAPM over CAPM – since the main tively. Similarly, Table 1 illustrates the mean difference and 95% confi-
contribution of the CAPM is that the stock returns depend only on market dence interval of the mean difference between CAPM and MAPM in terms
risk. Again, we used the mean, 95% confidence interval of the mean, and of MSE using the t.Bayes estimator of 0.161 and (-0.088, 0.411),
test hurdle of the absolute t-statistic of both 2.78 and 3 to exam these respectively. These findings mean that in asset pricing work, the CAPM
macroeconomic determinants as suggested by Harvey et al. (2015) and yields a statistically significant greater MSE (greater model error or lesser
Hou et al. (2020a,b). forecasting power) compare to the MAPM.

4. Results and discussion 4.1.2. The Sigma2

Panels (a) and (b) of Figure 3 showed that the differences between the
We assumed that the original CAPM holds for a stock if the beta is not two Sigma2s of the CAPM and MAPM were positive for the majority of
zero. So, we eliminated the stocks if their betas were zero from our an- stocks, especially in the case of the t.Bayes estimator. Importantly,
alyses. Panel (a) of Figure 1 showed that the CAPM using the Bayes Table 1 illustrates the mean difference and 95% confidence interval of
estimator yielded the 95% confidence interval of the beta of seven stocks the mean difference between CAPM and MAPM in terms of Sigma2 using
including zero. Similarly, Panel (b) of Figure 1 showed that CAPM using the Bayes estimator of 0.747 and (0.02, 1.48), respectively. Also, Table 1
the t.Bayes estimator yielded the 95% confidence interval of the beta of illustrates the mean difference and 95% confidence interval of the mean
three stocks including zero (two of those three stocks were the same difference between CAPM and MAPM in terms of Sigma2 using the
stocks as in the case of CAPM using the Bayes estimator). These findings t.Bayes estimator of 0.257 and (0.102, 0.411), respectively. Again, these
mean that the CAPM might not hold for eight stocks (seven and one findings mean that for an asset pricing work, the CAPM yields a statis-
stocks from Panel (a) and (b), respectively). Therefore, we decided to tically significant greater Sigma2 (lesser model precision and efficiency)
eliminate these 8 stocks (442 stocks remained) in our next analyses. The compare to the MAPM.
finding that the CAPM might not hold for only 8 stocks (1.7% of big U.S

(a) The 95% confidence interval of (b) The 95% confidence interval of
the CAPM's beta Using the Bayes the CAPM's beta using the t.Bayes
estimator estimator
4
4
3

2 2

1
0
0
30
59
88
117
146
175
204
233
262
291
320
349
378
407
436
1

30
59
88
117
146
175
204
233
262
291
320
349
378
407
436
1

-2 -1
Stock Stock

Figure 1. The 95% confidence intervals of the CAPM's beta. Notes: This figure reported the 95% confidence intervals of beta using the CAPM and Bayes, (Panel (a))
and t.Bayes (Panel (b)) estimators of 450 S&P 500 stocks. Panel (a) also showed seven stocks (91, 185, 207, 230, 231, 296, and 411) with zeroed betas. Panel (b) also
showed three stocks (91, 161, and 411) with zeroed betas.

4
C.D. Pham, L.T. Phuoc Heliyon 6 (2020) e05185

Figure 2. The differences between CAPM and

(a) The difference between two (b) The difference between two MAPM in terms of MSE. Notes: This figure re-
MSEs using the Bayes estimator MSEs using the t.Bayes estimator ported the differences between CAPM and MAPM
in terms of MSE using Bayes (Panel (a)) and
MSE (CAPM - MAPM)_Bayes MSE (CAPM - MAPM)_t.Bayes t.Bayes (Panel (b)) estimators of 442 S&P 500
stocks. For each stock, we calculated the MSE
10 10
from both CAPM and MAPM using Bayes esti-
8 8 mators. Then, we took the difference between
these two MSEs. We also calculated the sample
6 6
mean of all these differences of MSEs (mean dif-
4 4 ference). We repeated the same process for CAPM
and MAPM using t.Bayes estimator. Besides,
2 2
Panels (a) and (b) showed the means difference of
0 0 these MSEs of 0.749 and 0.161, respectively.

-2 -2

31
61
91
121
151
181
211
241
271
301
331
361
391
421
1
31
61
91
121
151
181
211
241
271
301
331
361
391
421
1

Stock Stock

Table 1. The differences between CAPM and MAPM.

Evaluation Criterion Estimator Used Mean from CAPM Mean from MAPM The Difference Between the CAPM and MAPM

Mean Difference 95% Confidence Interval of Mean Difference

Model error MSE Bayes 133.83 133.08 0.749 (0.018, 1.480)
t.Bayes 139.47 139.31 0.161 (-0.088, 0.411)
Sigma2 Bayes 135.75 135.01 0.747 (0.02, 1.48)
t.Bayes 26.72 26.47 0.257 (0.102, 0.411)
Model fit/adequacy R2B Bayes 0.30 0.30 -0.002 (-0.003, -0.001)
t.Bayes 0.65 0.66 -0.0032 (-0.0042, -0.0022)
D_bar Bayes 955.58 955.17 0.411 (0.129, 0.694)
t.Bayes 939.16 938.34 0.827 (0.527, 1.127)

Note: This table reported the mean difference and 95% confidence interval of the mean difference between CAPM and MAPM in terms of model error (MSE and Sigma2)
and model fit/adequacy (R2B and D_bar) using both Bayes and t.Bayes estimators of the S&P 500 stocks.

Figure 3. The differences between CAPM and MAPM

(a) The difference between two (b) The difference between two in terms of Sigma2. Notes: This figure reported the
Sigma2s using the Bayes estimator Sigma2s using the t.Bayes differences between CAPM and MAPM in terms of
estimator Sigma2 using Bayes (Panel (a)) and t.Bayes (Panel (b))
Sigma2 (CAPM - MAPM)_Bayes estimators of 442 S&P 500 stocks. For each stock, we
Sigma2 (CAPM - MAPM)_t.Bayes calculated the Sigma2 from both CAPM and MAPM
5
4 5 using Bayes estimators. Then, we took the difference
3 between these two Sigma2s. We also calculated the
2 3 sample mean of all these differences of Sigma2s (mean
difference). We repeated the same process for CAPM
1
1 and MAPM using t.Bayes estimator. Besides, Panels (a)
0
and (b) showed the means difference of these two
-1 -1 Sigma2s of 0.747 and 0.257, respectively.
-2
-3 -3
29
57
85
113
141
169
197
225
253
281
309
337
365
393
421
1

29
57
85
113
141
169
197
225
253
281
309
337
365
393
421
1

Stock Stock

4.2. The model fit/adequacy interval of the mean difference between CAPM and MAPM in terms of
R2B using the t.Bayes estimator of -0.0032 and (-0.0042, -0.0022),
4.2.1. The R2B respectively. These findings mean that in asset pricing work, the CAPM
Panels (a) and (b) of Figure 4 showed that the differences between the yields a statistically significant lower R2B (lesser model fit and explan-
two R2Bs of the CAPM and MAPM were distributed on both sides of zero. atory power) compare to the MAPM.
However, Table 1 illustrates the mean difference and 95% confidence
interval of the mean difference between CAPM and MAPM in terms of 4.2.2. The D_bar
R2B using the Bayes estimator of -0.002 and (-0.003, -0.001), respec- Panels (a) and (b) of Figure 5 showed that the differences between the
tively. Also, Table 1 illustrates the mean difference and 95% confidence two D_bars of the CAPM and MAPM were distributed on both sides of

5
C.D. Pham, L.T. Phuoc Heliyon 6 (2020) e05185

Figure 4. The differences between CAPM and MAPM

(a) The difference between two (b) The difference between two in terms of R2B. Notes: This figure reported the dif-
R2Bs using the Bayes estimator R2Bs using the t.Bayes estimator ferences between CAPM and MAPM in terms of R2B
using Bayes (Panel (a)) and t.Bayes (Panel (b)) esti-
R2B (CAPM - MAPM)_Bayes R2B (CAPM - MAPM)_t.Bayes mators of 442 S&P 500 stocks. For each stock, we
0.03 0.03 calculated the R2B from both CAPM and MAPM using
0.02 0.02 Bayes estimators. Then, we took the difference be-
0.01 0.01 tween these two R2Bs. We also calculated the sample
0.00 0.00 mean of all these differences of R2Bs (mean differ-
-0.01 -0.01 ence). We repeated the same process for CAPM and
-0.02 -0.02 MAPM using t.Bayes estimator. Besides, Panels (a) and
-0.03 -0.03 (b) showed the means difference of these R2Bs of
-0.04 -0.04
-0.002 and -0.0032, respectively.
-0.05 -0.05
31
61
91
121
151
181
211
241
271
301
331
361
391
421

31
61
91
121
151
181
211
241
271
301
331
361
391
421
1

1
Stock Stock

Figure 5. The differences between CAPM and MAPM

(a) The difference between two (b) The difference between two in terms of D_bar. Notes: This figure reported the
D_bars using the Bayes estimator D_bars using the t.Bayes differences between CAPM and MAPM in terms of
estimator D_bar using Bayes (Panel (a)) and t.Bayes (Panel (b))
D_bar(CAPM - MAPM)_Bayes estimators of 442 S&P 500 stocks. For each stock, we
D_bar(CAPM - MAPM)_t.Bayes calculated the D_bar from both CAPM and MAPM
12
using Bayes estimators. Then, we took the difference
12
8 between these two D_bars. We also calculated the
8 sample mean of all these differences of D_bars (mean
4 difference). We repeated the same process for CAPM
4
0 and MAPM using t.Bayes estimator. Besides, Panels (a)
0
and (b) showed the means difference of these D_bars
-4 -4 of 0.411 and 0.827, respectively.
31
61
91

31
61
91
121
151
181
211
241
271
301
331
361
391
421

121
151
181
211
241
271
301
331
361
391
421
1

Stock Stock

(a) The gamma using the Bayes (b) The gamma using the t.Bayes
estimator estimator
Gamma_MAPM_12 Yrs_Bayes Gamma_MAPM_12 Yrs_t.Bayes
75 75
50 50
25 25
0 0
-25 -25
-50 -50
-75 -75
31
61
91

31
61
91
121
151
181
211
241
271
301
331
361
391
421

121
151
181
211
241
271
301
331
361
391
421
1

Stock Stock

Figure 6. The gamma and distribution. Notes: This figure reported the gamma and distribution using both Bayes (Panel (a)) and t.Bayes (Panel (b)) estimators of 450
S&P 500 stocks. For each stock, we derived the gamma. Then, we calculated the sample mean, confidence interval, and t-statistic.

Table 2. The mean and 95% confidence interval of the MAPM's coefficients.

Coefficient Estimator Used Mean 95% Confidence Interval t-statistic

Gamma Bayes -1.17 (-4.18, 1.84) -0.76
t.Bayes -2.78 (-5.75, 0.19) -1.84
Kappa Bayes 2.36 (1.38, 3.33) 4.76
t.Bayes 1.61 (0.79, 2.42) 3.86
Lambda Bayes 2.01 (1.41, 2.62) 6.52
t.Bayes 1.27 (0.71, 1.83) 4.44

Note: This table reported the mean and 95% confidence interval of the gamma, kappa, and lambda using both Bayes and t.Bayes estimators of 450 S&P 500 stocks.

6
C.D. Pham, L.T. Phuoc Heliyon 6 (2020) e05185

(a) The kappa using the Bayes (b) The kappa using the t.Bayes
estimator estimator
Kappa_MAPM_12 Yrs_Bayes Kappa_MAPM_12 Yrs_t.Bayes

25 25
20 20
15 15
10 10
5 5
0 0
-5 -5
-10 -10
-15 -15
-20 -20
-25 -25
31
61
91

31
61
91
121
151
181
211
241
271
301
331
361
391
421

121
151
181
211
241
271
301
331
361
391
421
1

1
Stock Stock

Figure 7. The kappa and distribution. Notes: This figure reported the kappa and distribution using Bayes (Panel (a)) and t.Bayes (Panel (b)) estimators of 450 S&P 500
stocks. For each stock, we derived the kappa. Then, we calculated the sample mean, confidence interval, and t-statistic.

(a) The lambda using the Bayes (b) The lambda using the Bayes
estimator estimator
Lambda_MAPM_12 Yrs_Bayes Lambda_MAPM_12 Yrs_t.Bayes

25 25
15 15
5 5
-5 -5
-15 -15
-25 -25
31
61
91

31
61
91
121
151
181
211
241
271
301
331
361
391
421

121
151
181
211
241
271
301
331
361
391
421
1

Stock Stock

Figure 8. The lambda and distribution. Notes: This figure reported the lambda and distribution using Bayes (Panel (a)) and t.Bayes (Panel (b)) estimators of 450 S&P
500 stocks. For each stock, we derived the lambda. Then, we calculated the sample mean, confidence interval, and t-statistic.

zero. However, Table 1 illustrates the mean difference and 95% confi- interval, and test hurdle of absolute t-statistic of both 2.78 and 3 that
dence interval of the mean difference between CAPM and MAPM in terms suggested by two recent influential studies (Harvey et al., 2015; Hou
of D_bar using the Bayes estimators of 0.411 and (0.129, 0.694), et al., 2020a,b). Panels (a) and (b) of Figure 6 showed that the gamma
respectively. Similarly, Table 1 illustrates the mean difference and 95% using the Bayes and t.Bayes estimator, respectively, of almost all of the
confidence interval of the mean difference between CAPM and MAPM in stocks, were not zero; were distributed on both sides of zero. Importantly,
terms of D_bar using the t.Bayes estimator of 0.827 and (0.527, 1.127), Table 2 illustrates the mean, 95% confidence interval, and t-statistic of
respectively. Once more, these findings mean that the CAPM yields a the gamma using the MAPM and Bayes estimator of -1.17, (-4.18, 1.84),
statistically significant higher D_bar (lesser model adequacy) compare to and -0.76, respectively. Similarly, Table 2 illustrates the mean, 95%
the MAPM. confidence interval t-statistic of the gamma using the MAPM and t.Bayes
To summarize, this study shows that the MAPM yields greater pre- estimator of -2.78, (-5.75, 0.19), and -1.84, respectively. These findings
cision, forecasting, and explanatory power than the CAPM in asset pric- mean that the U.S. prime rate has a negative effect, on average, on the
ing via the S&P 500 stocks. However, Table 1 does not show how the stock returns. However, it does not clear the test hurdle of the absolute
three macroeconomic determinants affecting the S&P 500 stock returns. t-statistics of either 2.78 or 3. This finding contradicts some previous
So, we would examine this issue in the next section. studies (Dai and Zhou, 2020; Dai and Zhu, 2020; Wong et al., 2005) but is
consistent with others (Bernanke and Gertler, 1999; Kim, 2003).
Next, we examined the kappa. Panels (a) and (b) of Figure 7 showed
4.3. The effect of the macroeconomic determinants on the S&P 500 stock
that kappa using the Bayes and t.Bayes estimators, respectively, of almost
returns
all of the stocks, were not zero; were distributed on both negative and
positive sides. Importantly, Table 2 illustrates the mean, 95% confidence
The S&P 500 stocks are the large U.S stocks and known to be efficient.
interval, and t-statistic of the kappa using the MAPM and Bayes estimator
The main contribution of the CAPM is that the returns on stocks depend
of 2.36, (1.38, 3.33), and 4.76, respectively. Similarly, Table 2 illustrates
only on the market. However, the empirical results in the previous sec-
the mean, 95% confidence interval, and t-statistic of the kappa using the
tion showed that the MAPM consistently yielded a statistically significant
MAPM and t.Bayes estimator of 1.61, (0.79, 2.42), and 3.86, respectively.
greater precision, forecasting, and explanatory power in asset pricing
These findings mean that the U.S. government long-term bond rate has a
compared to the CAPM. So, it is well worth it to examine and confirm the
statistically significant positive effect on the stock returns and clears the
effect of the U.S. prime rate (gamma), U.S. government long-term bond
test hurdle of the absolute t-statistic of both 2.78 and 3. This finding is
risk premiums (kappa), and the exchange rate of USD/EUR (lambda) on
consistent with some studies (Dai and Zhou, 2020; Dai and Zhu, 2020)
the S&P 500 stock returns via the MAPM using the mean, confidence

7
C.D. Pham, L.T. Phuoc Heliyon 6 (2020) e05185

but contradicts others (Anderson et al., 2008; Campbell et al., 2019; Competing interest statement
Jare~no and Negrut, 2016; Wong et al., 2005).
Finally, we examined the lambda. Panel (a) and (b) of Figure 8 showed The authors declare no conflict of interest.
that lambda using the Bayes and t.Bayes estimators, respectively, of
almost all of the stocks, were not zero; were distributed on both positive Additional information
and negative sides. Importantly, Table 2 illustrates the mean, 95% con-
fidence interval, and t-statistic of the lambda using the MAPM and Bayes No additional information is available for this paper.
estimator of 2.01, (1.41, 2.62), and 6.52, respectively. Similarly, Table 2
illustrates the mean, 95% confidence interval, and t-statistic of the lambda
Acknowledgments
using the MAPM and t.Bayes estimator of 1.27, (0.71, 1.83), and 4.44,
respectively. These findings mean that the exchange rate of USD/EUR has
We also thank our colleague, Mr. Kirk Jordan – Eastern International
a statistically significant positive effect on the stock returns and clears the
University, a professional English editor and lecturer for valuable com-
test hurdle of the absolute t-statistic of both 2.78 and 3. This finding is
ments on the writing that greatly improved the manuscript.
consistent with some studies (Ajayi and Mougoue, _ 1996; Ajayi et al.,
1998) but contradicts others (Kim, 2003; Nieh and Lee, 2001).
References
5. Conclusions Adler, M., Dumas, B., 1983. International portfolio choice and corporation finance: a
synthesis. J. Finance 38, 925–984.
The CAPM is the most commonly used asset pricing model in practice, Agmon, T., 1972. The relations among equity markets: a study of share price co-
movements in the United States, United Kingdom, Germany and Japan. J. Finance 27
even with its deficiencies. Using both the qualitative and quantitative (4), 839–855.
research methods, our research proposed adding three macroeconomic Aharoni, G., Grundy, B., Zeng, Q., 2013. Stock returns and the Miller Modigliani valuation
determinants/risks (the U.S. prime rate, the U.S. government long-term formula: revisiting the Fama French analysis. J. Financ. Econ. 110 (2), 347–357.
Ajayi, R., Mougoue, _ M., 1996. On the dynamic relation between stock prices and
bond rate, and the exchange rate of USD/EUR) to the original CAPM to
exchange rates. J. Financ. Res. 19 (2), 193–207.
explain the nexus between the risks and the U.S. stock returns. The Ajayi, R., Friedman, J., Mehdian, S., 1998. On the relationship between stock returns and
MAPM, a non-traded factor model, is more flexible than the CAPM and exchange rates: tests of Granger causality. Global Finance J. 9 (2), 241–251.
Anderson, M., Krylova, E., Vahamaa, S., 2008. Why does the correlation between stock
very easy to apply since the data are always available. Unlike the other
and bond returns vary over time? Appl. Financ. Econ. 18, 139–151.
traded factor models, this MAPM is inspired by and based on the mac- Ang, A., Bekaert, B., 2007. Stock return predictability: is it there? Rev. Financ. Stud. 20,
roeconomic theory and models. Using a Gibbs sampler, the Bayesian 651–707.
approach (via both parametric and non-parametric Bayes estimators), Asness, C., Frazzini, A., 2013. The devil in HML’s details. J. Portfolio Manag. 39 (4),
49–68.
confidence interval approach, model error (MSE and Sigma2), and model Asness, C., Moskowitz, T., Pedersen, L., 2013. Value and momentum everywhere.
fit/adequacy (R2B and D_bar), we examined and compared the perfor- J. Finance 68 (3), 929–985.
mance of both CAPM and MAPM on the S&P 500 stocks from 2007-2019. Banz, R., 1981. The relationship between return and market value of common stocks.
J. Financ. Econ. 9, 3–18.
This study found: 1) the MAPM worked with more U.S stocks than the Barillas, F., Shanken, J., 2017. Which alpha? Rev. Financ. Stud. 30, 1316–1338.
CAPM. 2) the MAPM consistently yielded a statistically significant Barillas, F., Shanken, J., 2018. Comparing asset pricing models. J. Finance 73, 715–754.
greater forecasting, explanatory power, and model adequacy compared Barillas, F., Kan, R., Robotti, C., Shanken, J., 2019. Model comparison with Sharpe ratio.
J. Financ. Quant. Anal. 1–35.
to the CAPM. 3) Both the U.S. government long-term bond rate and ex- Barinov, A., Xu, J., Pottier, S., 2018. Estimating the cost of equity capital for insurance
change rate of USD/EUR had a statistically significant positive effect on firms with multiperiod asset pricing models. J. Risk Insur. 87 (1), 213–245.
the S&P 500 stock returns and cleared the test hurdle of absolute t-sta- Barroso, P., Santa-Clara, P., 2015. Momentum has its moments. J. Financ. Econ. 116 (1),
111–120.
tistic of both 2.78 and 3. These findings question the main contribution of Basu, S., 1983. The relationship between earnings yield, market value, and return for
the CAPM of that the stock returns depend only on the market; therefore, NYSE common stocks: further evidence. J. Financ. Econ. 12, 129–156.
this is another piece of evidence supporting the MAPM over the CAPM in Belo, F., Li, J., Lin, X., Zhao, X., 2017. Labor-force heterogeneity and asset prices: the
importance of skilled labor. Rev. Financ. Stud. 30 (10), 3669–3709.
asset pricing. These conclusions suggest that the MAPM is a more effi-
Belo, F., Lin, X., Bazdresch, S., 2014. Labor hiring, investment, and stock return
cient and advantageous asset pricing model compared to the CAPM. predictability in the cross section. J. Polit. Econ. 122 (1), 129–177.
Since the MAPM's data is always available, the investors and firm man- Berk, J., 1995. A critique of size-related anomalies. Rev. Financ. Stud. 8, 275–286.
agers would be better off employing the MAPM over the CAPM to predict Bernanke, B., Gertler, M., 1999. Monetary policy and asset price volatility. Econ. Rev. 84,
17–51.
the stock returns and a firm's cost of equity in practice. Also, these Bhandari, L., 1988. Debt/Equity ratio and expected common stock returns: empirical
findings may help both the policy-makers and investors to draft their evidence. J. Finance 43 (2), 507–528.
decisions in monetary policy and investment, respectively. Boons, M., 2016. State variables, macroeconomic activity, and the cross section of
individual stocks. J. Financ. Econ. 119 (3), 489–511.
For the smaller U.S. stocks and less efficient markets, we expect that Bower, D., Bower, R., Logue, D., 1984. Arbitrage pricing theory and utility stock returns.
the MAPM will yield even better performance compared to the CAPM. J. Finance 39 (4), 1041–1054.
Practitioners such as the investors and a firm's managers are advised to Bullard, J., Neely, C., Wheelock, D., 2009. Systemic risk and the financial crisis: a primer.
Fed. Reserv. Bank St. Louis Rev. 91, 403–417.
consider the MAPM over CAPM in their asset pricing work and cost of Campbell, J., Yogo, M., 2006. Efficient tests of stock return predictability. J. Financ. Econ.
equity estimation. 81, 27–60.
Campbell, J., Pueger, C., Viceira, L., 2019. Macroeconomic Drivers of Bond and Equity
Risks (Harvard Business School Working Paper No. 14-031). https://papers.ssrn.c
Declarations om/sol3/papers.cfm?abstract_id¼2332106.
Carhart, M., 1997. On persistence in mutual fund performance. J. Finance 52, 57–82.
Author contribution statement Chib, S., Zeng, X., Zhao, L., 2020. On comparing asset pricing models. J. Finance 75 (1),
551–577.
Da, Z., Gou, R., Jagannathan, R., 2012. CAPM for estimating the cost of equity capital:
L.T. Phuoc and C.D. Pham: Conceived and designed the experiments; interpreting the empirical evidence. J. Financ. Econ. 103, 204–220.
Performed the experiments; Analyzed and interpreted the data; Dai, Z., Zhou, H., 2020. Prediction of stock returns: sum-of-the-parts method and
Contributed reagents, materials, analysis tools or data; Wrote the paper. economic constraint method. Sustainability 12 (2), 541.
Dai, Z., Zhu, H., 2020. Stock return predictability from a mixed model perspective. Pac.
Basin Finance J. 60, 101267.
Funding statement Dhrymes, P., Friend, I., Gultekin, N., 1984. A critical reexamination of the empirical
evidence on the arbitrage pricing theory. J. Finance 39 (2), 323–346.
Dyckman, T., 2016. Significance testing: we can do better. Abacus 52 (2), 319–342.
This research is funded by the University of Economics and Law, Dyckman, T., Zeff, S., 2019. Important issues in statistical testing and recommended
Vietnam National University – Ho Chi Minh City/VNU-HCM. improvements in accounting research. Econometrics 7 (18), 1–11.

8
C.D. Pham, L.T. Phuoc Heliyon 6 (2020) e05185

Fama, E., French, K., 1992. The cross-section of expected stock returns. J. Finance 47, Lessard, D., 1976. World, country and industry relationships in equity returns:
427–465. implications for risk reduction through international diversification. Financ. Anal. J.
Fama, E., French, K., 1993. Common risk factors in the returns on stocks and bonds. 32 (1), 32–38.
J. Financ. Econ. 33, 3–56. Lewellen, J., Nagel, S., Shanken, J., 2010. A skeptical appraisal of asset pricing tests.
Fama, E., French, K., 1995. Size and book-to-market factors in earnings and returns. J. Financ. Econ. 96, 175–194.
J. Finance 50 (1), 131–155. Lo, A., MacKinlay, A., 1990. Data-snooping biases in tests of financial asset pricing
Fama, E., French, K., 1996a. Multifactor explanations of asset pricing anomalies. models. Rev. Financ. Stud. 3, 431–467.
J. Finance 51, 55–84. Loughran, T., Ritter, J., 1995. The new issues puzzle. J. Finance 50, 23–51.
Fama, E., French, K., 1996b. The CAPM is wanted, dead or alive. J. Finance 51 (5), MacKinlay, A., 1995. Multifactor models do not explain deviations from the CAPM.
1947–1958. J. Financ. Econ. 38, 3–28.
Fama, E., French, K., 2004. The capital asset pricing model: theory and evidence. J. Econ. Moskowitz, T., Ooi, Y., Pedersen, L., 2012. Time series momentum. J. Financ. Econ. 104
Perspect. 18 (3), 25–46. (2), 228–250.
Fama, E., French, K., 2006. Profitability, investment and average returns. J. Financ. Econ. Nieh, C.-C., Lee, C.-F., 2001. Dynamic relationship between stock prices and exchange
82 (3), 491–518. rates for G-7 countries. Q. Rev. Econ. Finance 41, 477–490.
Fama, E., French, K., 2008. Dissecting anomalies. J. Finance 63 (4), 1653–1678. Novy-Marx, R., 2013. The other side of value: the gross profitability premium. J. Financ.
Fama, E., French, K., 2015. A five-factor asset pricing model. J. Financ. Econ. 116 (1), Econ. 108 (1), 1–28.
1–22. Opdenakker, R., 2006. Advantages and disadvantages of four interview techniques in
Fama, E., French, K., 2016. Dissecting anomalies with a five-factor model. Rev. Financ. qualitative research. Forum Qual. Sozialforschung 7 (4), 11. https://www.qualitative
Stud. 29 (1), 69–103. -research.net/index.php/fqs/article/view/175.
Fama, E., French, K., 2017. International tests of a five-factor asset pricing model. J. Pham, C., Phuoc, L., 2020. Is estimating the capital asset pricing model using monthly and
Finance Econ. 123 (3), 441–463. short-horizon data a good choice? Heliyon 6 (7), e04339.
Fama, E., French, K., 2018. Choosing factors. J. Financ. Econ. 128 (2), 234–252. Phuoc, L., 2018. Jensen’s alpha estimation models in capital asset pricing model. J. Asian
Fang, H., Wang, B., Song, W., 2019. Analyzing the interrelationships among barriers to Finance Econ. Bus. 5, 19–29.
green procurement in photovoltaic industry: an integrated method. J. Clean. Prod. Phuoc, L., Pham, C., 2020. The systematic risk estimation models: a different perspective.
249, 119408. Heliyon 6 (2), e03371.
Ferson, W., Sarkissian, S., Simin, T., 1999. The alpha factor asset pricing model: a parable. Raj, A., Sah, B., 2019. Analyzing critical success factors for implementation of drones in
J. Financ. Mark. 2, 49–68. the logistics sector using grey-dematel based approach. Comput. Ind. Eng. 138,
Gibbons, M., Ross, S., Shanken, J., 1989. A test of the efficiency of a given portfolio. 106118.
Econometrica 57, 1121–1152. Roll, R., Ross, S., 1983. Regulation, the capital asset pricing model and the arbitrage
Goldenberg, D., Robin, A., 1991. The arbitrage pricing theory and cost of capital pricing. Public Util. Fortn. 111, 22–28.
estimation: the case of electric utilities. J. Financ. Res. 14 (3), 181–196. Roll, R., Ross, S., 1984. A critical reexamination of the empirical evidence on the
Goyal, A., Welch, I., 2008. A comprehensive look at the empirical performance of equity arbitrage pricing theory: a reply. J. Finance 39 (2), 347–350.
premium prediction. Rev. Financ. Stud. 21, 1455–1508. Rosenberg, B., Reid, K., Lanstein, R., 1985. Persuasive evidence of market inefficiency.
Grauer, F., Litzenberger, R., Stehle, R., 1976. Sharing rules and equilibrium in an J. Portfolio Manag. 11, 9–17.
international capital market under uncertainty. J. Financ. Econ. 3 (3), 233–256. Ross, S., 1976. The arbitrage theory of capital asset pricing. J. Econ. Theor. 13, 341–360.
Gungor, S., Luger, R., 2019. Exact inference in long-horizon predictive quantile Roy, R., Shijin, S., 2017. Dissecting anomalies and dynamic human capital: the global
regressions with an application to stock returns. J. Financ. Econom. 1–43. evidence. Borsa Istanbul Rev. 18 (1), 1–32.
Halsey, L., 2019. The reign of the p-value is over: what alternative analyses could we Roy, R., Shijin, S., 2018. A six-factor asset pricing model. Borsa Istanbul Rev. 18,
employ to fill the power vacuum? Biol. Lett. 15, 1–8. 205–217.
Harvey, C., Liu, Y., Zhu, H., 2015. … and the cross-section of expected returns. Rev. Shanken, J., 1982. The arbitrage pricing theory: is it testable? J. Finance 37 (5),
Financ. Stud. 29 (1), 5–68. 1129–1140.
Hou, K., Mo, H., Xue, C., Zhang, L., 2019. Which factors? Rev. Finance 23, 1–35. Sharpe, W.F., 1964. Capital asset prices: a theory of market equilibrium under conditions
Hou, K., Mo, H., Xue, C., Zhang, L., 2020a. An augmented q-factor model with expected of risk. J. Finance 19, 425–442.
growth. Rev. Finance 1–40. Smaga, P., 2014. The concept of systematic risk. Systematic Risk Center 1–26.
Hou, K., Xue, C., Zhang, L., 2015. Digesting anomalies: an investment approach. Rev. Solnik, B., 1974a. An equilibrium model of the international capital market. J. Econ.
Financ. Stud. 28, 650–705. Theor. 8, 500–524.
Hou, K., Xue, C., Zhang, L., 2020b. Replicating anomalies. Rev. Financ. Stud. 33 (5), Solnik, B., 1974b. The international pricing of risk: an empirical investigation of the
2019–2133. world capital market structure. J. Finance 29 (2), 48–54.
Jagannathan, R., Wang, Z., 1996. The conditional CAPM and the cross-section of expected Solnik, B., 1977. Testing international asset pricing: some pessimistic views. J. Finance 32
returns. J. Finance 51, 815–849. (30), 503–512.
Jare~
no, F., Negrut, L., 2016. U.S stock market and macroeconomic factors. J. Appl. Bus. Spiegelhalter, D., Best, N., Carlin, B., Van der Linde, A., 2002. Bayesian measures of
Res. 32 (1), 325–340. model complexity and fit. J. Roy. Stat. Soc. 64, 583–639.
Jegadeesh, N., Titman, S., 1993. Returns to buying winners and selling losers: Spiegelhalter, D., Best, N., Carlin, B., Van der Linde, A., 2014. The deviance information
implications for stock market efficiency. J. Finance 48, 65–91. criterion: 12 years on. J. Roy. Stat. Soc. B 1–9.
Jensen, M., 1967. The performance of mutual funds in the period 1945-1964. J. Finance. Stambaugh, R., Yuan, Y., 2017. Mispricing factors. Rev. Financ. Stud. 30 (4), 1270–1315.
Kim, D., Kim, T., Min, B.-K., 2011. Future labor income growth and the cross-section of The Association for Financial Professionals, 2013. Estimating and Applying Cost of
equity returns. J. Bank. Finance 35 (1), 67–81. Capital: Report of Survey Results. https://www.afponline.org/docs/default-source/r
Kim, K., 2003. Dollar exchange rate and stock price: evidence from multivariate egistered/mbr/2013-afp-cost-of-capital-complete?_n¼2351.
cointegration and error correction model. Rev. Financ. Econ. 12 (3), 301–313. The Credit Suisse, 2013. Estimating the Cost of Capital: A Practical Guide to Assessing
Kothari, S., Shanken, J., Sloan, R., 1995. Another look at the cross-section of expected Opportunity Cost. Credit Suisse Group AG. https://plus.credit-suisse.com/rpc4/ra
stock returns. J. Finance 50, 185–224. vDocView?docid¼P8cINv.
Kroencke, T., 2017. Asset pricing without garbage. J. Finance 72, 47–98. Titman, S., Wei, K., Xie, F., 2004. Capital investments and stock returns. J. Financ. Quant.
Krueger, R., Casey, M., Donner, J., Kirsch, S., Maack, J., 2001. Social Analysis: Selected Anal. 39 (4), 677–700.
Tools and Techniques. Social Development Papers, no. 36. World Bank Group. https Van der Linde, A., 2005. DIC in variable selection. Stat. Neerl. 59 (1), 45–56.
://documents.worldbank.org/en/publication/documents-reports/documentdetail/ Wallingford, B., Bicksler, J., 1974. Discussion. J. Finance 29, 392–398.
568611468763498929/social-analysis-selected-tools-and-techniques. Wang, J., Wu, Y., 2011. Risk adjustment and momentum sources. J. Bank. Finance 35,
Kubota, K., Takehara, H., 2017. Does the Fama and French five-factor model work well in 1427–1435.
Japan? Int. Rev. Finance 18 (1), 137–146. Wasserstein, R., Lazar, N., 2016. The ASA’s statement on p-values: context, process, and
Kuehn, L.-A., Simutin, M., Wang, J., 2017. A labor capital asset pricing model. J. Finance purpose. Am. Statistician 70 (2), 129–133.
72 (5), 2131–2178. Wong, W., Khan, H., Du, J., 2005. Money, Interest Rate, and Stock Prices: New Evidence
Kumar, A., Kaviani, M., Hafezalkotob, A., Zavadskas, E., 2017. Evaluating innovation from Singapore and the United States (Working Paper No. 0610). https://papers.ss
capabilities of real estate firms: a combined fuzzy delphi and Dematel approach. Int. rn.com/sol3/papers.cfm?abstract_id¼1607605.
J. Strat. Property Manag. 21 (4), 401–416. Wu, K.-J., Zhu, Y., Chen, Q., Tseng, M.-L., 2019. Building sustainable tourism hierarchical
Latham, M., 1989. The arbitrage pricing theory and supershares. J. Finance 44 (2), framework: coordinated triple bottom line approach in linguistic preferences.
263–282. J. Clean. Prod. 229, 157–168.
Lee, S., 2019. Money, asset prices, and the liquidity premium. J. Money Credit Bank. 52 Zhang, L., 2017. The investment CAPM. Eur. Financ. Manag. 23, 545–603.
(6), 1435–1466. Zhang, M., 2019. Labor-technology substitution: implications for asset pricing. J. Finance
Lessard, D., 1974. World, national and industry factors in equity returns. J. Finance 29 74, 1793–1839.
(2), 379–391.

9
 Update 1 of 2
Heliyon
Volume 6, Issue 12, December 2020, Page

DOI: https://doi.org/10.1016/j.heliyon.2020.e05627
 Heliyon 6 (2020) e05627

Contents lists available at ScienceDirect

Heliyon
journal homepage: www.cell.com/heliyon

Corrigendum

Corrigendum to “An augmented capital asset pricing model using new

macroeconomic determinants” [Heliyon 6 (10) October 2020 e05185]
Chinh Duc Pham a, b, Le Tan Phuoc c, *
a
University of Economics and Law, Ho Chi Minh City, Viet Nam
b
Viet Nam National University, Ho Chi Minh City, Viet Nam
c
Becamex Business School - Eastern International University, Viet Nam

In the original published version of this article, author Chinh Duc “Viet Nam National University Ho Chi Minh City, Viet Nam”. The authors
Pham's affiliation was listed as “University of Economics and Law, VNU- apologise for this mistake. Both the HTML and PDF versions of the article
HCM, Viet Nam”. The affiliations for this author should be listed as both have been updated to correct the error.
“University of Economics and Law, Ho Chi Minh City, Viet Nam” and

DOI of original article: https://doi.org/10.1016/j.heliyon.2020.e05185.

* Corresponding author.
E-mail address: [email protected] (L.T. Phuoc).

https://doi.org/10.1016/j.heliyon.2020.e05627
Received 26 November 2020; Accepted 26 November 2020
2405-8440/© 2020 The Author(s). Published by Elsevier Ltd. All rights reserved.
 Update 2 of 2
Heliyon
Volume 9, Issue 11, November 2023, Page

DOI: https://doi.org/10.1016/j.heliyon.2023.e20847
 Heliyon 9 (2023) e20847

Contents lists available at ScienceDirect

Heliyon
journal homepage: www.cell.com/heliyon

Corrigendum

Corrigendum to “An augmented capital asset pricing model using

new macroeconomic determinants” [Volume 6, Issue 10 (October
2020) e05185]
Chinh Duc Pham a, b, Le Tan Phuoc a, b, c, *
a
University of Economics and Law, Ho Chi Minh City, Viet Nam
b
Viet Nam National University, Ho Chi Minh City, Viet Nam
c
Becamex Business School - Eastern International University, Viet Nam

In the original published version of this article, the author has included the wrong affiliation as shown below.

The correct affiliation should be as follows:

Becamex Business School - Eastern International University, Viet Nam and both University of Economics and Law, VNU-HCM, Viet
Nam and Viet Nam National University Ho Chi Minh City, Viet Nam.

DOI of original article: https://doi.org/10.1016/j.heliyon.2020.e05185.

* Corresponding author. University of Economics and Law, Ho Chi Minh City, Viet Nam.
E-mail address: [email protected] (L.T. Phuoc).

https://doi.org/10.1016/j.heliyon.2023.e20847
Received 9 October 2023; Accepted 9 October 2023
Available online 14 October 2023
2405-8440/© 2020 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).