000080970 THE INFLUENCE OF MACRO FACTORS ON STOCK RETURN IN HO CHI MINH STOCK EXCHANGE ẢNH HƯỞNG CỦA CÁC YẾU TỐ VĨ MÔ ĐẾN LỢI NHUẬN CỔ PHIẾU TẠI SỞ GIAO DỊCH CHỨNG KHOÁN THÀNH PHỐ HỒ CHÍ MINH

IN T R O D U C T IO N

Overview o f the problem and attempt to solve the problem

Vietnam's stock market is an emerging market that was established on July 11, 1998, after the government signed Decree 48/CP to approve its creation The decree paved the way for the Vietnam stock market and led to the formation of its first two stock exchanges, located in Hanoi and Ho Chi Minh City.

The influences o f macro factors on stock return in HOSE

HOSE (the Ho Chi Minh City Stock Exchange) was established by Decision No 127/1998/QD-TTG in 1998, but it did not begin operating until July 2000, when it conducted its first trading session The Hanoi Stock Exchange (HNX) officially started later, in March 2005 HOSE serves as the platform for large-cap listings and major companies, while HNX caters to smaller investors and primarily lists small and medium-sized enterprises.

Vietnam's stock market began in 1998, but the period from 1998 to 2005 was a quiet phase, with the VN-Index hovering around 200–300 points and total market capitalization at the end of 2005 totaling just 0.69% of GDP.

In 2006, Vietnam's stock market posted a broad rally across all three trading platforms—HOSE, HNX, and the OTC market During this upturn, the VN-Index climbed to a peak of 1,170.67 points, while total market capitalization surged to 500,000 billion VND, equal to about 43.7% of GDP (Vneconomy, 2008).

The Vietnam stock market was severely damaged by the global financial crisis, a downturn tied to unsustainable development In the first half of 2008, the VN-Index fell by about 60%, sparking a wave of selling as investors tried to cut further losses By the end of 2008, total trading value had dropped to 27 trillion VND Five years after the downturn, the Vietnam stock market had not yet recovered, and persistent macroeconomic weakness is cited as a key reason for this slow rebound.

This study investigates whether the development of the stock market is hampered by unstable macro factors or by other drivers, and how stock returns in the Vietnam market are affected by different macro factors It analyzes the impact of macroeconomic volatility on market growth and identifies which macro variables most strongly influence Vietnamese stock returns, providing practical insights for investors and policymakers.

To answer this question, my thesis is carried out to find the answer for the questions mentioned above.

Firstly, my thesis will try to find the potential factors affecting return o f the stock listed in HOSE in the period from 2009 to 2014.

Second, in evaluating stock-return models, this thesis compares three models—CAPM, APT, and the Fama–French model—to determine which best explains stock return fluctuations on the Ho Chi Minh Stock Exchange (HOSE).

An overview o f the thesis

This thesis is divided into three main parts to find a clear answer for each o f the above questions.

Section one presents the literature review, introducing the major theories of stock returns with a focus on the CAPM, APT, and Fama–French models It outlines the required assumptions for each pricing framework and explains how these assumptions influence their relative strengths, limitations, and applicability The discussion then proceeds to two empirical results that illustrate how the CAPM and the Fama–French three-factor model perform in real markets, comparing their applications in the NYSE and China's A‑share market.

Part two of this thesis, the core, presents data analysis and findings: it details the sampling process for HOSE-listed stocks and lists the independent variables—risk-free rate, market risk premium, CPI, gold price, exchange rate, crude oil, firm size, and book-to-market ratio—along with the justification for their selection It then explains model construction, using EViews to build and estimate the models, and analyzes the results to determine whether CAPM, APT, and the Fama–French model can be applied to HOSE stocks.

Like any study, this thesis has limitations that cannot be fully resolved, and these limitations are clearly identified and explained to guide future researchers toward practical remedies that enhance the robustness of ensuing work The findings provide a fundamental starting point for broader research on other stock exchanges or on the Vietnam stock exchange, employing longer time horizons and larger sample sizes The final section details the future research opportunities and prospective directions, enabling subsequent studies to build on this work and achieve more comprehensive insights.

L IT E R A T U R E R E V IE W

Overview o f existing literature on the research topic

CAPM model was firstly introduced in 1952 by Harry M arkowitz and known as Harry Markowitz M odem Portfolio Theory (MPT), after that, it was developed 12 years later in

1964 by William Sharpe, John Lintner and Jan Mossin The CAPM model has some specific assumptions:

^ Investors have an identical holding period

S Investments are limited to a universe o f publicly traded financial assets, such as stocks and bonds, and to risk-free borrowing or lending arrangements.

■/ Investors pay no taxes on returns and no transaction costs (commissions and service charges) on trades in securities

S All investors are rational mean-variance optimizers

Under homogeneous expectations, all investors evaluate securities in the same way and share the same view of the economy Given a set of security prices and the risk-free rate, every investor uses the same expected returns and the same covariance matrix of security returns to derive the efficient frontier and identify the unique optimal risky portfolio.

With those above assumptions, the CAPM model demonstrates the relationship between the expected return o f a specific stock and the market return as:

In which: E(r,) is the expected return o f a specific stock r f is the risk free rate o f return [E (rm)- r f] is the market risk premium

E (rm) is the expected market return

The following graph is the visual relationship between stock return and the market return

Graph 4: The Capital Asset Pricing Model

Under the CAPM framework, there is a linear relationship between stock return and market return The model indicates that a one-unit increase in the market risk premium leads to an increase in the stock's expected return by an amount equal to its beta This beta coefficient captures the stock's sensitivity to market movements, tying market performance directly to individual stock performance.

Many analysts have argued that the CAPM's proposed linear relationship rests on an idealized set of assumptions that rarely hold in real markets In practice, this model depicts an ideal economy that hardly ever exists, which raises questions about whether beta alone suffices as the explanatory variable for the return–risk relationship.

To remove the drawbacks o f the CAPM model, Arbitrage Pricing Model has been developed by Stephen Ross in 1976 APT also has its own assumptions:

S All common variation between returns can be described by a factor model.

S There are no arbitrage opportunities.

S Idiosyncratic risk can be diversified through portfolio formation.

S There is perfect competition in the market.

When those assumptions exist, APT model can be used:

In which: E(rj) is the expected return o f stock i ry is the risk free rate

(3k i is the coefficient beta o f stock i to factor k f k is the factor k o f the model

Using the APT model, investors can choose the most suitable common factors affecting their stock to have a better prediction on stocks expect return.

According to Kent Womack, The APT model has some advantages over the CAPM model such as:

Compared with the CAPM, which uses a single market premium as the only explanatory variable for stock returns, the APT framework allows multiple explanatory variables—beta coefficients corresponding to several common factors—to estimate returns As a result, APT offers flexibility to adapt to different economic environments and time periods by selecting the most impactful common factors This multi-factor approach can capture a wider set of sources of risk and return than CAPM, enabling more accurate modeling of stock performance across varying conditions.

CAPM involves many strict assumptions that rarely match real-market conditions, and it relies on the market portfolio return, E(rm), to determine a stock’s expected return By contrast, the Arbitrage Pricing Theory (APT) makes far fewer assumptions, and its framework better reflects actual stock market behavior, offering an alternative approach to explain asset returns.

However, APT model still has following disadvantages that need to be solved by a more comprehensive model:

Choosing the right common factor to show how macro factors affect a specific stock price is inherently challenging Consequently, differences in investors’ intentions and ability to select the factors produce divergent expected returns at different levels of precision.

Based on the APT model, Eugene Fama and Kenneth French have developed the most popular stock pricing model in 1992, the Fama French model:

E (n )= rf + Pi(rm - 7 y ) + /?sweE(SMB) + /?hm lE(HML)

E(rj) is the expected rate o f portfolio return.

Tf is the risk-free rate o f return

(rm ~ rf ) *s the expected rate o f excess market portfolio return

E (SMB) is the expected value o f the difference between the excess return on a portfolio o f small stocks and the excess return on a portfolio o f big stocks

The application o f CAPM, APT and Farma French model on stock exchange

E (HML) is the expected value o f the difference between the excess return on a portfolio o f high-Book-to-market stocks and the excess return on a portfolio o f low-book-to-market

Fama and French (1993) proposed a three-factor asset pricing model that explains excess portfolio returns using market, size, and value risk factors The model identifies three sources of systematic risk: the market excess return, the SMB factor (the difference between small and big stock returns), and the HML factor (the difference between high and low book-to-market stocks) Portfolios constructed to mimic these factors—capturing market exposure, company size, and book-to-market value effects—show significant explanatory power for stock returns, underscoring the importance of size and value in asset pricing beyond the traditional market factor.

Connor and Sehgal (2001) empirically examined the Fama and French model for India, Faff

(2001) tested the model in Australian stock market by using shelf index, Aksu and Onder

Empirical tests across markets have compared the CAPM with the Fama-French Three-Factor Model A 2003 study on the Istanbul Stock Exchange contrasted CAPM with the Fama-French framework, and Gaunt (2004) examined both models on the Australian Stock Exchange The results from these studies consistently favor the Fama-French model, showing that the added SMB and HML factors beyond the CAPM’s market risk premium significantly influence stock returns.

2.2 The application of CAPM, APT and Fama French model on stock exchange

To test the application of CAPM, APT, and the Fama–French models, a large body of empirical research has been conducted across diverse stock exchanges, with most studies comparing CAPM results to Fama–French results to highlight the model’s limitations; this paper briefly presents three empirical findings drawn from the New York Stock Exchange, representing a highly developed market, the Istanbul Stock Exchange, representing a developing market, and the Mauritius Stock Exchange, representing an emerging market.

The New York Stock Exchange (NYSE) traces its origins to the Buttonwood Agreement of 1792, signed under a buttonwood tree on Wall Street to establish organized securities trading Today, the NYSE lists more than 8,000 securities, making it the world's largest stock exchange by listings.

A study by Malardalen University on the US stock market uses the CAPM model and the Fama-French model as its two primary methods to explain stock return variation.

In this research, the stock price movement on NYSE over a 30-month period from October

2010 to March 2013 was observed 72 stock from 4 different industries listed on the S&P

A sample of 500 stocks was selected over a period of at least two years To prevent selection bias, stocks with negative book-to-market (B/M) equity ratios were excluded and replaced with alternative stocks.

In the CAPM part, a two-pass regression was used and the result of cross-sectional is presented below: rv

As can be seen from the result

The t-value for β0 is -9.12719, which is more extreme than the critical t-value of -2.120, indicating that β0 is significantly different from zero Although β0's estimate is relatively small (-0.07104), this does not prove that β0 does not exist In short, stock price variation does not depend solely on β0; other factors also influence it.

S Although the t-value o f risk premium b (3.45040) is bigger than the critical t-value (2.120), the risk premium result o f 0.015185 is quiet small comparing to the historical evidence o f about 8.5% for the past 80 years.

S The R 2 value is 0.390797, which is at the acceptable range but it does revenge that the /? does not have s strong explaination power on the risk premium o f the stock.

2.2.1.2 The Fama French three factor model on NYSE

Using the Fama-French three-factor model, 72 stocks were initially grouped into four portfolios based on their book-to-market equity ratio; each of these B/M groups was then subdivided into four portfolios by market capitalization, yielding 16 stock portfolios in total The following table presents a portion of the regression results for the Fama-French three-factor test.

Dependent variable: Excess return on 16 stock portfolios formed on size equity and Book-to-market

Size Book- to-Market equity (BE/ME)

Low 2 High Low 2 o High ò is t(s)

Bold values are significant t-value

Table 1 : Fama French three factors model result on NYSE

Table results show that the slope coefficients for the size factor (SMB) are related to size: within each book-to-market quartile, SMB slopes decrease monotonically from the smallest to the largest size quartile, while the slope coefficients for the book-to-market factor (HML) increase monotonically from the lowest to the highest book-to-market quartile, in line with Fama–French theory (Xin Yi, 2004).

Another key finding from the regression results is that only one t-statistic testing the null hypothesis that the intercept equals zero exceeds the 5% one-tailed critical value of 1.7056 with 26 degrees of freedom This suggests the intercept is not significantly different from zero in the majority of cases Overall, the result aligns with the Fama-French three-factor model, which predicts that the intercept should be zero.

The intluences o f macro factors on stock return in HOSE

Based on NYSE data, Fama-French tests show that the Fama-French model explains stock price variation more effectively than the CAPM This finding highlights the model's stronger explanatory power for equity returns and supports its use in asset pricing analyses.

China's stock markets consist of the Shanghai Stock Exchange and the Shenzhen Stock Exchange, both established in the 1990s In Christopher Gan's study, only Chinese A-shares are used as observations due to the large number of listed shares on the Chinese stock market.

This study uses the average monthly closing price, adjusted for capital asset changes, of A-share stocks listed in the China Stock Market and Accounting Research Database (CSMAR) from 1996 to 2005 as the dependent variables.

2.2.2.1 CAPM model on Chinese A-stock market

In the CAPM framework, the premium return of each of the six SMB and HML portfolios is regressed on the average monthly market return of A-stocks, excluding stocks with negative book value, and the regression results are presented in the table below.

BH BM BL SH SM SL a

* Significant at the 0.05 evel (2-tailec ) ** Significant at the 0.01 level (2-tailec )

Table 2: CAPM result on Chinese A-stock market From the result table:

If the CAPM can explain stock price fluctuations, the alpha value should be significantly different from zero However, in four of the six portfolios, the null hypothesis that alpha equals zero is rejected, indicating that CAPM pricing leaves measurable pricing errors in those portfolios.

M E T H O D O L O G Y

Research questions

The following graph demonstrates the fluctuation o f the market stock return and the risk free rate in the research period

Graph 5: Market return and risk free rate fluctuation

From the graph, market returns show substantial volatility while the risk-free rate remains relatively stable Within the CAPM framework, the fluctuation of our stock portfolio can be explained by changes in the market risk premium scaled by the portfolio’s beta Yet the market risk premium may not be the only factor driving stock return fluctuations, since firm-specific events and other systematic influences can cause deviations from CAPM predictions Practically, this means we should estimate beta, monitor shifts in the market risk premium, and consider alternative explanations or multifactor approaches when explaining and forecasting portfolio returns.

The APT model tries to customize the explanatory by allowing the researcher to use different independent factors base on their choices In this research, the following macro factors are chosen:

Number Factors Base year Source

2 USD/VND Exchange rate 2008 GSO

Table 4: Macro factors o f APT model Among a lot o f macro factors, I choose these four factors because:

The stock prices are selected on the weekly basic, therefore, all the independent factors must be recorded at least on the monthly basic to ensure that the selection bias is minimized Other important macro factors such as: Money supply, Government deficit are recorded only on the quarterly or annual basic, which can increase the test bias.

Foreign currency (USD), Gold and crude oil trading are the most important substitute for stock trading in a flexible market Therefore, their prices will be closely related.

The fluctuation o f the USD price, Gold price, crude oil price and CPI is presented in the following graph:

Between 2009 and 2014, the CPI, exchange rate, gold price, and crude oil price fluctuated significantly, while the VN-Index stock price showed only a small change This divergence implies that macro factors may have limited explanatory power for stock price movements in Vietnam during this period Therefore, empirical analysis is needed to determine whether these macro variables can account for VN-Index dynamics or whether other determinants drive the stock market more strongly.

This study asks whether macroeconomic factors have explanatory power for the returns of a stock portfolio selected from HOSE and, if so, at what confidence level this influence can be statistically detected Using a regression framework that links portfolio return to a set of macro variables—such as GDP growth, inflation, interest rates, exchange rates, and policy shocks—we assess the strength of the macro factor model through standard metrics (R-squared, F-statistics, and p-values) and report the levels at which the null hypothesis of no influence cannot be rejected The findings suggest that macro factors can explain a meaningful portion of HOSE portfolio returns in certain periods or portfolio constructions, with statistically significant relationships emerging at conventional confidence levels (for example, 5%); however, in other cases the evidence is weaker and the null hypothesis cannot be rejected at those levels.

According to the Fama–French three-factor model, stock price fluctuations are linked to three main factors: company size, the book-to-market ratio, and the market risk premium This framework explains how smaller firms and those with higher book-to-market values tend to exhibit distinct return patterns due to their exposure to the market risk premium, capturing variations in stock prices beyond the traditional market factor.

Graph 7: Capitalization and P/B ratio o f listed companies

Like other stock exchanges worldwide, this market shows clear segmentation by market capitalization and the book-to-market ratio While many listed companies have a market capitalization of less than 100 billion VND, many others have a market capitalization above 100 billion VND.

Ten trillion VND highlights the context: the book-to-market ratio varies markedly across firms in this economy As a result of the slow economic recovery, many companies show a book-to-market ratio below one, while others still perform well and have high book-to-market ratios This wide disparity raises the question of whether stock returns for firms of different sizes and with different book-to-market ratios exhibit distinct price fluctuation patterns.

The theoretical framework

This study uses the CAPM, APT, and the Fama–French model as the standard approaches in stock price regression to evaluate how different factors explain stock price fluctuations By performing three regression analyses, we assess each model's explanatory power and compare their advantages and disadvantages in the specific context of HOSE stocks.

CAPM model will be the first model used in this research As in many other researches, the common CAPM model will be used is: £ ( n ) - rf = P [E(rm ) - r f ] + C

In which: E(n) is the expected return o f the stock portfolio rf is the risk free rate in Vietnam market

E (rm) is the market return in Vietnam market

We employ a time-series regression model on the average stock return of our portfolio from January 2009 to March 2014, regressing it against the market risk premium over the same period to estimate the relationship between these two variables.

To test the application of CAPM model, we test the following two hypothesises

S Test on the constant term

Under the CAPM framework, if the model fully explains the relationship between stock portfolio returns and the market risk premium, the intercept term C should be zero, and the null hypothesis that C equals zero would not be rejected by the data.

Under the CAPM, if the model explains the relationship between stock portfolio returns and the market risk premium, the portfolio beta must be different from zero, and the null hypothesis that beta equals zero should be rejected A statistically significant, nonzero beta indicates that portfolio returns move with the market risk premium as CAPM predicts Rejection of the null hypothesis H0: beta = 0 provides empirical evidence that systematic market risk is priced in portfolio returns, supporting CAPM's risk–return relationship and its explanatory power in equity markets.

Under the Capital Asset Pricing Model (CAPM), expected returns are linked to a single general factor—the market risk premium While this relationship explains how market-wide movements influence asset returns, CAPM cannot specify which particular factors affect a given stock portfolio, nor quantify their individual impact or the confidence with which those effects can be measured.

Applying the Arbitrage Pricing Theory (APT), we identify and select the independent variables most likely to influence our stock portfolio’s returns The four factors chosen are the Consumer Price Index (CPI), the VND/U.S dollar exchange rate, the gold price, and the crude oil price, with the rationale for their selection to be presented in the next section Once these independent variables are established, we construct an APT-based time-series regression to quantify their impact on portfolio returns.

The stock portfolio return is modeled in the Vietnam market by the risk-free rate, with coefficients that measure the impact of four key factors: the Consumer Price Index (CPI), the VND/USD exchange rate, the gold price, and the crude oil price Each factor has its own coefficient, and the model also incorporates the time changes (percent changes) of CPI, the VND/USD rate, the gold price, and the crude oil price to reflect how dynamic movements affect portfolio performance.

After the using o f CAPM model, the APT model is used because it has some advantages that can fit well in the stock market o f Vietnam compared to CAPM model.

Firstly, many CAPM assumptions do not hold in the Vietnam stock market Investors in the Vietnam stock market do not share identical holding periods, reflecting varied investment horizons Like other markets, Vietnam’s stock market offers two main investing forms: long-term investing and shorter-term strategies, highlighting the diversity of investors and approaches within the market.

Macro factors influence stock returns for HOSE institutional investors as well as short-term investing (stock surfing) by individual investors The public information assumption cannot be fully applied in the Vietnam stock market, which remains an emerging market where company disclosures are not updated in time and investors struggle to keep up with operations According to Vietnam Stock, only 19 of 704 companies listed on both HOSE and HNX meet Government Circular 52/2012 on the public information of listed companies This finding suggests that an ideal CAPM-based stock market does not exist in Vietnam.

Although the CAPM framework assumes investors pay no taxes and incur no transaction costs, in reality stock investors—both individuals and institutions—face different tax brackets and trading costs that can dampen their willingness to buy or sell, thereby influencing supply and demand and ultimately stock prices Investors also do not share homogeneous expectations: perspectives diverge on both global and Vietnam domestic economic conditions, especially during the volatile 2009–2014 period that followed the 2008 financial crisis, and they differ in risk preference, with some investors being risk-averse and seeking to protect returns while others are risk-seeking and pursue higher returns, leading each investor to be attracted to different stocks with varying levels of risk.

To test the application o f APT model on the stock price, we made four hypothesis testing

S Testing on the CPI factor

An arbitrage pricing theory (APT) model that explains the relationship between stock portfolio returns and CPI fluctuations shows that the CPI factor's coefficient is nonzero, and the null hypothesis that the CPI factor's effect is zero should be rejected This finding signals a statistically significant CPI-driven component in portfolio performance, confirming the APT’s explanatory power for inflation-related movements in asset returns.

The three other tests will be repeated for Gold price factors, VND/USD Exchange rate factor, and Crude oil factors with the same testing method.

Generally, when CPI, the VND/USD exchange rate, gold prices, and crude oil prices each influence stock portfolio returns, all four betas are significantly different from zero, and the null hypotheses that these betas equal zero are rejected.

The Fama-French model is widely regarded as an efficient framework for assessing the factors that drive stock portfolio returns It posits that returns are influenced primarily by three factors: market risk as captured by the market premium, company size captured by SMB (small minus big), and the book-to-market value captured by HML (high minus low) The standard specification expresses the expected return of a portfolio i as r_i ≈ r_f + β_i (r_m − r_f) + s_i SMB + h_i HML, where r_f is the risk-free rate, r_m is the market return, SMB measures size, and HML measures value This decomposition helps investors understand how market, size, and value effects contribute to portfolio performance.

Pi is the coefficient o f the market risk premuim factor rm is the market risk premium

PsMB is the coefficient on the company size factor (SMB) is the company size factor

P hml is the coefficient on the company Book-to-market ratio (HML) is the company Book-to-market ratio factor

Size factor is the average difference in returns between small-cap and large-cap companies, capturing the premium often associated with smaller firms The book-to-market ratio factor is the average difference in returns between companies with high book-to-market ratios and those with low book-to-market ratios The exact formulas for these factors will be provided later in the data analysis section.

The Fama-French model posits that stock portfolio returns can be explained by three factors: the market risk premium, the size factor, and the book-to-market factor Numerous studies have demonstrated the applicability of the Fama-French framework across stock exchanges worldwide, finding that both firm size and the book-to-market ratio have explanatory power for stock return fluctuations.

DATA A N A LY SIS

Stock portfolio form ation

Data were collected from stocks continuously listed on the Ho Chi Minh Stock Exchange (HSX) from 2009 to 2014 From cophieu68.vn, 367 stocks satisfied the criteria and their company information—especially market capitalization and the book-to-market ratio—was downloaded These companies were then grouped into 16 industry-based portfolios The number of stocks in each industry group was proportional to the representation of that industry in the overall market, and within each group stocks were selected to reflect the capitalization distribution within that industry Finally, the stocks from all groups were combined to form the research stock portfolio.

Sector Group Number of company

2 Rubber price 5 10 Bank, Finance and 8

Table 5: Number o f stock in each sector o f stock sample

Starting with 376 HOSE stocks, we construct a representative sample by assembling a portfolio of 145 stocks to serve as the stock parameter This sampling approach minimizes selection bias and ensures the sample accurately reflects the size distribution and book-to-market levels of companies in the market.

Stock analysis

After the stock portfolio formation, we start analyzing the stock information to get the series for the time regression test.

By using the website: www.Vndirect.vn, the daily stock price from 1 January, 2009 to 31 March, 2014 for each stock is downloaded to make a comprehensive price table for the stock portfolio.

To obtain the weekly stock price, the average of each block of five consecutive trading days is calculated, and this five-day average is used to represent the weekly stock return From the daily stock prices, 152 weekly observations are then produced, forming the dependent variable series for the thesis data analysis Each weekly observation is labeled with the format ww-mm-yyyy, for example 01-09-2009 denotes the first week of September 2009.

After getting the weekly price for each stock, the stock then, are grouped into different categories based on their level o f capitalization and their book to market ratio.

All 152 company capitalization levels are listed with their stock symbols Using Excel, the data are sorted by market capitalization to create an ascending list from the smallest to the largest cap, with values ranging from 5 billion VND to 119,188 billion VND The 51st position in this ranking identifies the corresponding capitalization level.

Starting from 101 as the baseline, we classify a list of 152 companies into three market-cap tiers—small-cap, mid-cap, and large-cap Companies with market capitalization exceeding 794 billion VND are grouped as large-cap, while those below that threshold are categorized as small-cap or mid-cap The 101st ranked company helps define the boundary for the high-cap segment, anchoring the thresholds used to separate the three capitalization levels.

Market capitalization is used to classify companies into big-cap, mid-cap, and small-cap segments: companies with 794 billion VND or more are listed as big-cap, those with capitalization below 203 billion VND (the 51st-ranked company has 203 billion VND) are listed as small-cap, and all remaining firms with capitalization between 203 billion VND and 794 billion VND are listed as medium-capitalization (M).

A consistent procedure is applied to the company's book-to-market ratio to form three groups: Low, for ratios below 0.76; Medium, for ratios between 0.76 and 1.24 (inclusive); and High, for ratios above 1.24 This classification supports comparative analysis across firms with different book-to-market profiles.

The final result is that we have 9 different stock portfolios with different number o f stocks based on the company capitalization and the company Book-to-market ratio as following:

Table 6: Number o f companies in each stock portfolio After finishing with the stock portfolios formation we have to calculate the stock return to get the observation for our time regression.

Weekly stock return can be calculated from the weekly stock price under the assumption that the price already reflects all relevant information, including dividends, other income, and taxes on capital gains and dividends This approach, aligned with the idea that markets efficiently absorb new information, implies that the return arises from the combination of price changes and cash distributions The stock return formula is Return = (P_t - P_{t-1} + D_t) / P_{t-1}, where P_t is the closing price at the end of the week, P_{t-1} is the price at the start of the week, and D_t denotes dividends paid during the period; if other income or taxes need to be accounted for, you can adjust the numerator accordingly.

After forming the 9 stock return portfolios, we will form the observations for the CAPM model The CAPM model requires having the average stock return o f the whole sample,

The influences o f macro factors on stock return in HOSE which can be calculated by taking the average function to get the stock return o f 145 stocks

In this thesis, the market stock return choose is the return on VN-Index and the risk free rate is the refinancing interest rate o f the State Bank o f Vietnam.

The market risk premium can be calculated easily by taking the minus of VN-Index return to the risk free rate.

To generate the input data for the APT model, we collect four key indicators: Vietnam CPI, gold price, the VND/USD exchange rate, and crude oil price Because Vietnam's CPI is not published on a weekly basis, we use the monthly CPI figure for each week In addition, the CPI base year changes from year to year, so all CPI values must be converted to a common base year through a normalization step before they are used in the model.

2008 The CPI will be calculated as following:

To generate the data required for the Fama-French model, we compute the SMB (small minus big) and HML (high minus low) factors In the Fama-French framework, SMB and HML are constructed by sorting stocks into portfolios based on size (market capitalization) and value (book-to-market ratio), forming long–short portfolios, and averaging their excess returns to produce the factor time series These SMB and HML factors capture size and value effects in asset returns and serve as essential inputs for the three-factor model used in asset pricing and empirical finance research.

SMB is the company size factor

HML is the company Book-to-market ratio factor

SM is the stock portfolio o f small capitalization and medium Book-to-market ratio stock

SL is the stock portfolio o f with small capitalization and low Book-to-market ratio stock

SH is the stock portfolio o f small capitalization and high Book-to-market ratio stock

ML is the stock portfolio o f medium capitalization and low Book-to-market ratio stock

MH is the stock portfolio o f medium capitalization and high Book-to-market ratio stock

BL is the stock portfolio of big capitalization and low Book-to-market stock

BM is the stock portfolio o f big capitalization and medium Book-to-market ratio stock

BH is the stock portfolio o f big capitalization and high Book-to-market stock

Regression re su lt

Using the Eview to run the time regression on the stock return o f our portfolios, we get the following result:

After running the CAPM model on the stock portfolio return again the market risk premium, the following table o f result is presented:

Variable Coefficient Std Error t-Statistic Prob.

S.E of regression 0.037154 Akaike info criterion -3.739586

Sum squared resid 0.345105 Schwarz criterion -3.711575

Log likelihood 473.1879 Hannan-Quinn criter -3.728315

Table 7: Regression result on CAPM model

From this table, we get the linear relationship between the stock portfolio and the market risk premium as: r i - r f = -0 0 0 2 4 2 5 + 0 211709 ( r VN_,ndex - r f )

The t-values o f the test are:

S T-value for the constant term is -1.0349 which is larger than the t-critical value o f - 1.9719 at 5% significance and 252 -2= 250 degree o f freedom

We can not reject the hypothesis that the constant term is equal to zero

S T-value for the beta coefficient is 3.41, which is larger than the t-critical value of 1.9719 at 5% significance and 252-2 = 250 degree o f freedom.

■=> We can reject the hypothesis that the beta is equal to zero at 95% level of significant

To test the validity o f our regression value, we use two tests on autocorrelation ansd heteroscedasticity

■S The Dubin-Watson test: We have: at 0.05 level o f significance and 252 observation, d L= 1.758, d u = 1.758.Because d= 1.62< d L, there is evidence that the autocorrelation does exist.

S The White test: We have the following Eview result for the White test

Obs*R-squared 6.089397 Prob Chi-Square(2) 0.0476

Scaled explained SS 8.320064 Prob Chi-Square(2) 0.0156

Because n*R2 = 6.089 > 5.99 (chi square value at 5% significance and 2 degree o f freedom), we can not conclude that there is no heteroscedasticity.

To improve the ordinary least square regression in the CAPM, we use the Newey-West estimator to run the regression again:

Newey-West HAC Standard Errors & Covariance (lag truncation=4)

Using the Newey-West estimator, the relationship between stock returns and the market risk premium remains unchanged relative to the ordinary CAPM result This finding indicates that the link between returns and the market premium is robust to the estimator used.

Although the t-value is statistically significant, the model’s R-squared is only 0.044, meaning that just 4% of the variation in stock returns is explained by changes in the market risk premium This low explanatory power suggests that factors beyond the market risk premium drive stock returns in this analysis.

With this result, we can come to a conclusion that:

At 95 level of confidence, when the market premium increases by 1%, the stock portfolio return increases by 0.21%.

After running the regression on the APT model o f the stock portfolio return against the four macro factors, we get the following result:

Table 8: Regression result on the APT model From the result table, we get the following relationship equation: r f - ry = 0 0 1 5 8 + 0 0 00 723 CPI - 0 0 0 0 1 2 9 CRU - 0 0 0 0 2 5 1 6 0 0

The absolute critical t-value at 95% confidence level and 247 degree o f fredoom is 1.972, when comparing with the t-values from Eview, we have following result

S T-value for CPI factor is 3.3> 1.972

^ We can reject the hypothesis that the beta for the CPI factor is equal to zero at 95% level o f confidence

S Absolute t-value for Crude oil price factor is 0.39 < 1.972

■=> We can not reject the hypothesis that the beta for the Crude oil factor is equal to zero at 95% level o f confidence.

■S Absolute t-value for the VND/USD exchange rate factor is 1.0788 < 1.972

■=> We can not reject the null hypothesis that the beta for the VND/USD factor is equal to zero at 95% level o f confidence.

•/ Absolute t-value for the gold price exchange rate factor is 1.95 < 1.972 However, this t-value is higher than 1.65 (t-critical value at 90% level o f confidence and 247 degree o f freedom).

We can not reject the null hypothesis that the beta for gold price is equal to zero at 95% level o f confidence.

The regression's R-squared value is 4.7%, indicating that only 4.7% of the variation in stock returns is explained by the CPI, gold price, VND/USD exchange rate, and crude oil price.

With this result, we can only come to a conclusion that:

W hen other factors are unchanged, at 95% level of confidence, when CPI increases b y l %, the stock return will increase by 0.0073%.

To understand why three of the four macro factors appear to have insignificant effects on stock returns, we generate a residual plot to check for violations of t-test assumptions This approach helps ensure the inferences drawn about macro-factor effects are reliable by signaling whether the residuals meet the normality, independence, and homoscedasticity conditions required for the t-test.

Graph 9: The residual o f the APT Model

Analysis of the residual graph suggests the error term may increase over time, potentially limiting the explanatory power of the APT model To test for a time pattern in the error term Uj, we conducted a White heteroscedasticity test The EViews output for this test summarizes whether the variance of the error term changes with time, indicating heteroscedasticity if present The heteroskedasticity results from EViews are reported alongside the test statistics and p-values, informing interpretation of the APT model's significance in the presence of potential time-varying error variance.

Obs*R-squared 35.68971 Prob Chi-Square(14) 0.0012

Table 9: Regression result o f White test From the result table, we have: n.R' = 35.689, while chi square value for 15 degree o f freedom at 95% level o f confidence is 24

=> We can reject the null hypothesis that there is no heteroscedasticity

At the 95% confidence level, we conclude that heteroscedasticity is present in the error term Uj, violating the ordinary least squares (OLS) assumption of constant variance that underpins the i-test This finding implies that the reliability of OLS estimates and the associated inferences in the i-test may be compromised when heteroscedasticity is present.

To remove the heteroscedasticity, we try to use the weighted least squares method to correct our formula After several trying, the final variable is chosen for transforming is the CPI factor Effecting this transformation, we obtain the following result:

Dependent Variable: RI_RF Method: Least Squares

Log likelihood 486.8939 Hannan-Quinn enter -3.796377

S.E of regression 0.037343 Sum squared resid 0.344441

Table 10: Regression result o f WLS method After applying the weight least square method, we have the following equation:

(rf - r f )CPI = 0 0 1 8 4 CPI + 0 0 00 695 CPI2 - (8 2 6 E - 0 5 )CRU CPI

Although the transformed model still has very low R2, it helps to remove the heteroscedascity problem.

Obs'R-squared 23.83965 Prob Chi-Square(15) 0.0679

After testing the explanation power o f gold price, CPI, crude oil price and VND/USD exchange rate, we can conclude that at 95% level o f confidence, only CPI factor has influence on the stock return fluctuation.

To get the result for the Fama French model, we have to run nine Eview regression models for nine stock portfolios The result is consolidated as following:

4.3.3.1 Eview result of the market risk premium

The return o f each stock portfolios are regressed against the market risk premium based on the CAPM model for each o f 9 stock portfolios

Low Medium High Low Medium High ò t(ò)

Bold values are significant at 5% level o f significance, 250 d f (two-tail)

Table 11: Regression result o f the CAPM model based on Fama French three factors

From the result table, we can see that the beta coefficient o f the stock portfolio does follow the rule based on the company capitalization and Book-to-market ratio.

Across all market capitalizations, the beta coefficient of small-cap companies is lower than that of large-cap companies, regardless of the book-to-market (B/M) ratio Likewise, the beta coefficient of stocks with high B/M ratios is lower than that of stocks with low B/M ratios, regardless of capitalization.

Another key finding is that all estimated intercepts (constant terms) are essentially zero, and none is statistically different from zero This result strongly supports the CAPM, which implies that the intercept term does not exist in the model.

Although none of the R2 values exceed 10%, the t-values of seven out of nine stock portfolios are highly significant (t-critical = 1.96 at the 95% confidence level with 250 degrees of freedom) Consequently, we can reject the null hypothesis and conclude that the CAPM model can explain the relationship between stock returns and the market risk premium, even when stocks are grouped by size and book-to-market (B/M) ratios.

4.3.3.2 Eview result regarding the Fama French three factor model

The following table presents the result o f Fama French three factors model regression (the detailed result is presented on the appendix)

Low Medium High Low Medium High ò s i t-value

Big 0.6714 0.5029 0.4115 0.002 0.001 0.001 t-critical value: 1.96 and -1.96 at 95% level o f confidence and 248 degree o f freedom

Bold value is significant t-value at 95% confident level, two-tail

Table 12: Regression result o f the Fama French three factors model

From the Eview result, we can see that:

For HOSE-listed stocks, the size factor affecting portfolio return aligns with the Fama-French model, showing that small-cap stocks have higher beta than large-cap stocks This indicates greater sensitivity to market movements for small-cap shares and suggests they can contribute higher returns to a portfolio, albeit with increased risk compared with large-cap stocks.

The influences o f macro factors on stock return in HOSE companies with the smaller market capitalization will have higher return than the stock of companies with bigger market capitalization

Regarding the Book-to-market ratio factors, again, the Fama French model theory is followed

Results show that stocks of companies with higher book-to-market (B/M) ratios exhibit higher beta coefficients than stocks of companies with lower B/M ratios This suggests that higher B/M stocks tend to deliver higher returns than lower B/M stocks.

Analysis of the 18 t-values with 248 degrees of freedom shows that only three are not significant at the 95% confidence level; however, two of these three non-significant t-values become significant at the 90% level (t = 1.78 and 1.76), both exceeding the 1.645 critical value This finding indicates that company size and the book-to-market ratio have a meaningful impact on stock return variation.

Under the Fama-French three-factor framework, the market risk premium is not proven by HOSE-listed stocks If the model is applied strictly, the beta coefficient for the market risk factor should equal 1 and exhibit a significant t-value However, HOSE stock data show the market beta is nearly zero, and only two of the nine t-values are significant at the 95% confidence level with 248 degrees of freedom This result suggests that the market risk premium has limited explanatory power for stock return variation when stocks are grouped by company size and book-to-market ratio.

Finding sum m ary

After compiling comprehensive stock data and constructing a diversified stock portfolio, we conducted regression analyses using the three most common models to assess how macroeconomic factors affect stock prices for stocks listed on the HOSE The results for HOSE-listed stocks are reported in this section.

Results for HOSE-listed stocks support the CAPM model, indicating that stock returns can be explained by the market risk premium However, the returns do not fluctuate as much as the CAPM would predict, implying potential deviations from the model or the influence of other factors on price movements.

Macro factors influence stock returns on the Ho Chi Minh City Stock Exchange (HOSE), with the market risk premium as a key driver: a 1% increase in the market risk premium is associated with about a 0.2% rise in HOSE stock returns.

Among CPI, the VND/USD exchange rate, gold price, and crude oil price, only the CPI factor shows explanatory power for stock return fluctuations; the other variables cannot explain returns due to heteroscedasticity in the error term.

Among the three factors of the Fama-French model, firm size and the book-to-market ratio exhibit strong explanatory power for stock return fluctuations The results of this thesis robustly support the Fama-French theory: smaller market capitalization is associated with higher stock returns, and lower book-to-market ratios correspond to higher returns However, a persistent heteroscedasticity problem arises, which prevents the market risk premium—the third factor—from explaining HOSE stock return fluctuations.

Research lim itation

In this thesis, there are some limitations that can not be solved to have a more precise and comprehensive result.

Firstly, the weekly stock returns are not recorded exactly but they are calculated from the daily stock value

Secondly, two macro factors chosen are also not recorded weekly The weekly data are generated from the monthly data, which can reduce the exact o f the test.

Thirdly, while the Fama-French three-factor model has been supported by numerous studies, there are additional factors that can influence stock returns One notable factor is momentum, introduced by Fama and French in 2012; however, momentum is not used as an independent factor in this thesis.

C O N C L U SIO N

Summ ary

Stock investment remains one of the world’s most popular ways to build wealth, offering potentially high returns but also substantial risk that varies across countries and sectors To help investors manage this risk, this article introduces three widely used stock return evaluation models: CAPM (Capital Asset Pricing Model), APT (Arbitrage Pricing Theory), and the Fama–French three-factor model It explains how these models assess expected stock performance and risk, and it presents real-world examples from developed markets such as the New York Stock Exchange and the Chinese A-share market to illustrate their practical applications in evaluating stock returns.

Using the results from the NYSE and the Chinese A-share market, this study conducts data analysis to evaluate whether the CAPM, the APT, and the Fama–French three-factor model can predict returns on stocks listed on the Ho Chi Minh Stock Exchange (HOSE) If these models fail to predict HOSE stock returns, the paper explains the reasons for their limited applicability in the Vietnam stock market and outlines directions for future research to address these limitations and develop a more comprehensive model for forecasting HOSE stock returns.

Implication of the th e sis

Based on the results of this thesis, stock returns are clearly influenced by multiple factors, with company size and the book-to-market ratio emerging as the most significant For issuers, these findings provide a practical framework to evaluate how changes in market capitalization and the book-to-market ratio may affect their stock price The results can be extended by incorporating additional factors believed to influence stock prices, enabling more accurate predictions of future stock performance.

From the point o f view o f the individual investors, the result can be used as a guidance to evaluate the significance of the impact from different factors on the stock price Knowing

Macro factors drive stock returns on HOSE, and understanding how variables like GDP growth, inflation, interest rates, exchange rates, and policy changes influence market performance helps investors choose stocks that fit their time horizon and risk tolerance By analyzing these macro drivers, investors can better predict sector trends, assess risk-adjusted returns, and make disciplined stock selections on the Ho Chi Minh Stock Exchange This approach supports investment decisions aligned with individual preferences for time frame and risk, enabling more effective asset allocation and portfolio construction on HOSE.

These findings indicate that institutional investors managing diversified stock portfolios can preserve their return levels when there is a favorable balance between market capitalization and the book-to-market ratio Expanding the research to identify additional influential factors can reveal how stock returns relate to these variables, enabling more informed investment decisions and the potential to achieve higher returns than peers.

Future research opportunities

This study acknowledges limitations that remain unresolved at present, yet these constraints create opportunities for future research to advance the work, address the outstanding issues, and refine methods to overcome these limitations.

The research can be developed more by adding more macro factors and also the momentum factors as suggested by Frama French model in 2012

The research can be developed more and may have a more reliable result if the data is selected for a longer time period with a more stable economics situation

This study analyzes stocks primarily from southern Vietnamese companies, which do not fully reflect the Vietnamese stock market as a whole To broaden coverage and improve representativeness, future research could include stocks listed on both the Hanoi Stock Exchange (HNX) and the Ho Chi Minh City Stock Exchange (HOSE).

Another important research opportunity is the global applicability of this methodology, enabling evaluation of the CAPM model, the APT model, and the Fama-French model across stock markets worldwide at different levels of development This cross-market analysis can reveal how these asset pricing models perform under varying market conditions and development stages, informing researchers and investors about model robustness and relevance in diverse financial environments.

Key finding

After building a stock portfolio and testing the stock return using CAPM model, APT model and Fama French three factors model, some key findings are found as following:

Across the NYSE and the Chinese A-share market, the CAPM and the Fama–French three-factor model show strong applicability; however, in the HOSE market, neither model is fully supported.

Using an Arbitrage Pricing Theory (APT) framework to predict stock returns with four covariates—the consumer price index (CPI), gold price, the VND/USD exchange rate, and crude oil price—we find that only the CPI factor has a statistically significant impact on stock returns The other three factors show no robust predictive power within the model, suggesting that CPI-driven inflation dynamics largely drive cross-sectional risk premiums in the data This result positions CPI as the dominant macro factor in explaining stock return variation under the APT approach, while gold, exchange rate, and oil prices contribute little to expected returns in the studied period Investors can interpret this as CPI-related signals being more informative for risk pricing and portfolio allocation when using APT, though mindful of model specification and data window limitations.

Macro factors have strong explanatory power for fluctuations in stock returns on HOSE, whereas the three other factors show little to no explanatory power over stock return volatility.

Stock return fluctuations on the HOSE largely support the Fama-French three-factor model, driven by size and book-to-market effects The regression results indicate that smaller-capitalization stocks generate higher returns than larger-cap stocks, while stocks with lower book-to-market ratios outperform those with higher ratios However, the market risk premium does not substantially explain stock returns, a limitation possibly due to heteroscedasticity arising from market instability observed from 2009 to 2014.

Concluding rem ark

This thesis aims to lay the foundational step for predicting stock returns in Vietnam’s stock market, an emerging market known for its high volatility While acknowledging several limitations that remain unresolved, the study follows rigorous, standard procedures to minimize bias in stock selection and in the construction of the regression model Although it cannot explain all fluctuations in stock returns, the work yields a reasonable model that can be developed further by future researchers to achieve more comprehensive results, helping investors forecast stock returns, reduce investment risk, and protect their funds.

Table 15: Regression o f Fama French three factors model on small capitalization and low Book-to-market ratio companies

Table 16: Regression o f Fama French three factors model on small capitalization and medium Book-to-market ratio

Table 17: Regression o f Fama French three factors model on small capitalization and high Book-to-market ratio

Table 18: Regression o f Fama French three factors model on medium capitalization and low Book-to-market ratio

Table 19: Regression o f Fama French three factors model on medium capitalization and medium Book-to-market ratio

Newey-West HAC Standard Errors & Covariance (lag tru n ca tio n ^)

Table 20: Regression o f Fama French three factors model on medium capitalization and high Book-to-market ratio

Table 21 : Regression o f Fama French three factors model on big capitalization and low Book- to-market ratio

Table 22: Regression o f Fama French three factors model on big capitalization and medium Book-to-market ratio

Newey-West HAC Standard Errors & Covariance (lag tru n ca tio n ^)

Table 23: Regression o f Fama French three factors model on big capitalization and high Book-to-market ratio

0.411562 Mean dependent var 0.0005590.404443 S.D dependent var 0.0334160.025788 Akaike info criterion -4.4620640.164926 Schwarz criterion -4.406041566.2200 Hannan-Quinn criter -4.43952157.81815 Durbin-Watson stat 1.9513510.000000

Tiêu đề	The Influence of Macro Factors on Stock Return in Ho Chi Minh Stock Exchange
Tác giả	Nguyễn Thị Hồng Nhung
Người hướng dẫn	Dr. Dao Thanh Binh
Trường học	Hanoi University
Chuyên ngành	Finance and Banking
Thể loại	Bachelor thesis
Năm xuất bản	2014
Thành phố	Hanoi

Định dạng
Số trang	53
Dung lượng	5,1 MB