Asset pricing for the ASEAN5 equity markets is the main focus of this thesis, although we also develop vector error correction models VECM for GDP, trade and local equity market returns
Introduction
This thesis investigates asset pricing within the ASEAN region, focusing specifically on the five founding member countries: Malaysia, Singapore, Thailand, Indonesia, and the Philippines, collectively referred to as the ASEAN5 The selection of these countries is based on data availability, highlighting their significance in the regional equity markets Given their geographical proximity and shared ASEAN membership, analyzing the determinants of asset pricing in these nations' equity markets is essential for understanding regional market dynamics.
Note: The ASEAN region includes Malaysia, Singapore, Thailand, Indonesia, the Philippines, Brunei
Darussalam, Vietnam, Cambodia, Myanmar and Laos
1 Other ASEAN members include Brunei Darussalam, Vietnam, Laos, Myanmar and Cambodia
Contributions of the thesis
This thesis makes three key contributions, with the primary focus on analyzing the asset pricing of ASEAN5 equity markets using the four-factor model, a study area previously lacking in literature The four-factor model combines Fama and French's (1993) three-factor approach with the momentum factor from Carhart (1997) By examining size-BTME and industry portfolio returns, this research investigates how market, size, value, and momentum factors influence variations in ASEAN5 portfolio returns Additionally, incorporating industry portfolios broadens the scope of asset pricing tests, providing a more comprehensive understanding of return determinants in these emerging markets.
This thesis advances ASEAN5 asset pricing research by incorporating macroeconomic factors into the traditional factor model, creating a macro-factor model Key macro variables include unexpected GDP (UGDP), unexpected total trade (UTT), and unexpected market returns (URI), all derived from vector error correction models (VECMs) or vector autoregressions (VARs), to address cointegration issues Additionally, the model includes the world market excess returns (WRF) to capture global effects This innovative approach extends previous studies, such as Chen, Roll, and Ross (1986) and Chen and Zhang (1997), offering a significant contribution to understanding asset pricing within the ASEAN region.
This thesis's third contribution analyzes cointegration tests among the ASEAN5 equity markets, GDP, and trade, revealing significant linkages within the region Despite previous studies examining ASEAN5 equity markets individually or collectively, this research emphasizes the importance of understanding these interconnected relationships, especially given the comprehensive data set utilized The study carefully considers the impact of the 1997 Asian financial crisis, highlighting variations in ASEAN5 equity market linkages before and after the crisis, and provides a thorough analysis of the entire period to shed light on these dynamic relationships.
Most existing research on GDP focuses on GDP per capita, especially in gravity models and convergence studies (e.g., Cappelen, Castellacci, and Verspagen, 2003; Canova, 2004; Lim and McAleer, 2004) However, the interconnectedness of aggregate GDP among ASEAN economies is rarely explored in the literature, despite its significance for asset pricing Cointegration tests among the ASEAN-5 economies enhance our understanding of their economic linkages and regional integration.
This thesis conducts a comprehensive time series analysis of ASEAN5 GDP, considering both real and nominal US dollar figures By adjusting for the 1997 Asian crisis and seasonal effects, the study provides robust evidence of significant long-term and short-term relationships in the region's economic growth The findings highlight the interconnectedness of ASEAN5 economies, emphasizing the impact of global financial shocks and seasonal patterns on their GDP dynamics.
A comprehensive analysis of ASEAN5 trade linkages provides a number of insights into those links Cointegration tests are performed for bilateral trade, ASEAN5 imports, ASEAN5 exports and ASEAN5 total trade These analyses contribute to the literature by providing a comprehensive time series analysis of trade links that exist between the ASEAN5 countries 2 It is found that there is little evidence of strong trade links between the ASEAN5 countries.
Research motivation
ASEAN was established on August 8, 1967, in Bangkok, and has expanded from five to ten member countries over the past four decades Today, the ASEAN region boasts a population of approximately 560 million, a combined GDP of nearly US$1.1 trillion, and total trade valued at around US$1.4 trillion As a key stabilizing force in Southeast Asia, ASEAN’s economic growth, regional cooperation, and resilience during the 1997 Asian financial crisis underscore its significance This article focuses on the ASEAN5, exploring asset pricing and the interconnections between equity markets, GDP, and trade within the region.
Emerging markets, particularly the ASEAN5 economies, have recently experienced deregulation, leading to distinct financial and economic characteristics compared to developed markets This makes ASEAN5 a valuable sample for testing the applicability of asset pricing models that were originally designed for more mature markets.
2 Most of the literature provides trade analysis using either one of these measures (for example, see
Baharumshah, Lau and Fauntas, 2003; Santos-Paulino and Thirlwall, 2004; Herzer and Nowak-Lehmann,
Objectives of the study
This thesis explores five key objectives across five chapters, beginning with analyzing the interconnections within ASEAN5 equity markets using weekly data from January 1990 to March 2006, including the influence of US, Japanese, and Australian markets It then investigates the relationship between ASEAN5 economies’ GDP and their market linkages to better understand economic growth correlations The third objective examines the connection between trade measures and GDP within ASEAN5 nations to assess if trade linkages reflect in equity market performance The fourth focus tests asset pricing models in individual ASEAN5 markets using a four-factor approach to explain return variations Finally, the thesis assesses the impact of macroeconomic variables—such as unexpected GDP, trade, and market returns—on asset pricing, aiming to determine whether incorporating these factors enhances the model’s explanatory power.
Thesis structure
This thesis is structured to provide a comprehensive analysis of ASEAN5 economies and markets Chapter 2 offers a literature review of previous research related to the study’s subject matter The methodology used for analysis is detailed in Chapter 3, followed by an overview of the time series data sets in Chapter 4 Chapter 5 explores the equity market linkages within the ASEAN5, while Chapter 6 and 7 examine the economic interconnections through GDP and trade relationships The asset pricing of ASEAN5 assets is analyzed in Chapter 8 using the traditional CAPM and the four-factor model Further, Chapter 9 extends this analysis by incorporating four additional macroeconomic factors into the asset pricing models Finally, Chapter 10 summarizes the key findings, discusses research limitations, and proposes directions for future studies.
Introduction
This chapter provides a comprehensive literature review for the studies included in this thesis, covering broad research as well as focusing on key areas related to the research objectives Section 2.2 reviews existing studies on equity market linkages, highlighting the interconnectedness of global financial markets Section 2.3 discusses literature concerning GDP and its impact on economic development, emphasizing key findings in macroeconomic analysis Section 2.4 examines research related to trade links, exploring how international trade influences economic growth and market dynamics Finally, Section 2.5 reviews previous asset pricing studies, offering insights into valuation models and market behavior relevant to this thesis.
International equity market linkages
The correlation and interdependence between equity markets have been extensively studied in academic literature Key factors influencing international stock market correlation include the overall state of the economy, the level of equity market development within a country (Erb, Harvey, and Viskanta, 1998), and prevailing market trends Notably, correlation tends to increase significantly during bear markets, while it remains relatively stable during bull markets (Longin and Solnik, 2001; Yang, Tapon, and Sun, 2006).
Bekeart and Harvey (2000) highlight that following equity market liberalisation, the correlation and beta of emerging markets tend to increase relative to the global market, indicating greater integration Equity market liberalisation enables both foreign and domestic investors to buy and sell securities without restrictions, progressively linking emerging markets to the global capital market In cointegrated markets, assets with similar risk levels are expected to have the same returns regardless of their country of origin (Bekaert and Harvey, 1997) However, market liberalisation alone may not encourage foreign investment due to factors like home bias and other investor concerns.
The size of a national equity market reflects its stage of development, market liquidity, and the associated information and transaction costs Significant differences in market sizes often indicate disparities in liquidity and costs between markets As the size gap between two stock markets widens, their co-movement tends to decrease, whereas closer market sizes are associated with higher synchronization (Pretorius, 2002).
6 such as lack of information on company stocks, may impede international investment (see Bekaert, 1995 and Levine and Zervos, 1996) 5
Mainstream research has traditionally focused on the interdependence of mature stock markets, but there is an increasing interest in understanding the cointegration among emerging equity markets and their relationship with developed markets Emerging markets, such as the ASEAN5, are characterized by higher, more predictable returns that exhibit greater volatility and low correlations with developed markets Their return distributions are non-normal, skewed, and possess fat tails, indicating higher risk Additionally, emerging market returns are often autocorrelated, suggesting potential predictability based on historical data Factors like country-specific market segmentation and regional segmentation significantly influence expected returns in these emerging markets.
Emerging market economies that open their markets to enhance financial integration are more susceptible to external shocks due to their lower resilience compared to developed markets These nations face increased vulnerability, making them more affected by global economic fluctuations According to Michelfelder and Pandya, emerging markets tend to have limited capacity to withstand external shocks, highlighting the risks associated with financial openness in these regions.
Negative shocks tend to have a greater impact on markets than positive shocks, especially in emerging economies that respond more quickly to localized events (Heaney, Hooper, and Jaugetis, 2002; Dungey, Fry, and Martin, 2003; Soydemir, 2000) The transmission of equity market turmoil increases with market integration, and advancements in information technology may further accelerate shock propagation (Fernendez-Izquierdo and Lafuente, 2004) Major financial shocks, such as the 1987 US stock market crash, the 1994 Mexican Peso devaluation, the 1997 Asian crisis, and the 2000 IT crash, illustrate how crises have significantly affected the global financial landscape Notably, crises in emerging markets have challenged traditional financial theories, highlighting unique market behaviors (Buckberg, 1995).
5 It is suggested that home bias seems to persist for households but this bias is decreasing for financial institutions (Kearney and Lucey, 2004)
The 1997 Asian financial crisis significantly impacted regional stock markets and economies, drawing extensive scholarly attention Studies such as Forbes and Rigobon (2002) reveal high co-movement among Southeast Asian markets immediately following the crisis, initially suggesting contagion but later arguing these changes reflected ongoing market interdependence rather than true contagion, supported by Khalid and Kawai (2003) Although Chiang, Jeon, and Li (2007) find evidence of contagion during the crisis’s early stages, this thesis prioritizes understanding the crisis’s overall impact on the ASEAN5 economies, adjusting analysis accordingly to account for these effects.
This article first reviews existing literature on international equity market linkages to highlight the importance of understanding global market relationships It emphasizes the necessity of examining the interdependence between international markets, particularly focusing on geographically close markets like the ASEAN5 Previous studies, such as Eun and Shim's analysis of nine developed stock markets—Australia, Canada, France, Germany, Hong Kong, Japan, Switzerland, the UK, and the US—provide valuable insights into these interconnected dynamics, laying the groundwork for understanding regional market linkages within the ASEAN5.
(1989) The results show that substantial interdependence exists among these developed equity markets, with the US playing the most dominant role in influencing the other equity markets
Kasa (1992) investigates the presence of common stochastic trends across the equity markets of the US, Japan, England, Germany, and Canada, finding evidence of a single shared trend influencing these major developed markets Recent studies, such as Tsouma (2007), also highlight interdependence between mature and emerging stock markets, emphasizing the interconnectedness of global financial markets.
Chen, Firth, and Rui (2002) document the interdependence of six major Latin American stock markets from 1995 to 2000 through cointegration analysis Their study, which divides the period into three sub-periods, reveals that Latin American equity markets are consistently cointegrated across all sub-periods, highlighting their long-term interconnectedness.
6 They also study the 1994 Mexican crisis and the 1987 US stock market crisis
Contagion is defined as a significant increase in cross-market linkages following a shock to one or more countries, indicating heightened market interconnectedness Alternatively, Bekaert et al (2005) describe contagion as correlations that exceed what is expected based on economic fundamentals, reflecting abnormal market co-movements during turmoil.
8 The stock markets are for Argentina, Brazil, Chile, Colombia, Mexico and Venezuela
Soydemir (2000) explores the relationship between Latin American markets and the US stock market, finding significant linkages that are influenced by trade flow levels Additionally, Al-Khazali, Darrat, and Saad document the existence of equity market connections among Gulf Cooperation Council (GCC) countries from 1994 to 2003, highlighting regional market integration during that period.
Research indicates strong interconnectedness among regional stock markets; Narayan, Smyth, and Nanda (2004) highlight significant linkages between the stock markets of Bangladesh, India, Pakistan, and Sri Lanka, fostering regional financial integration Additionally, Drakos and Kutan (2005) reveal that Turkish and Greek equity markets are interconnected primarily due to shared trading partners and foreign direct investment ties, underscoring the influence of cross-border economic relationships on market linkages Incorporating these insights can enhance understanding of regional market dynamics and improve investment strategies.
European economic integration and capital market liberalisation during the 1980s and 1990s significantly strengthened market linkages, surpassing the impact of monetary integration or the introduction of a single currency (Baele, 2005) Since 1996, European equity markets have become highly integrated, primarily driven by the momentum towards Economic and Monetary Union (EMU), which reduced exchange rate volatility and uncertainty associated with monetary unification (Fratzscher, 2002).
Several studies analyze the linkages between ASEAN5 countries and other equity markets For instance, Choudhry, Lu, and Peng (2007) found that significant long-term relationships exist among Far East equity markets—including ASEAN5, Hong Kong, South Korea, and Taiwan—and the US and Japanese markets across pre-1997 crisis, crisis, and post-crisis periods Their results suggest that the US and Japanese markets may not be essential for the interaction among Far East markets, with the Japanese market exerting a more influential role during and after crises Similarly, Johnson and Soenen (2002) demonstrated that the Japanese equity market has become increasingly integrated with twelve other Asian markets since 1994, emphasizing Japan's growing influence in regional market dynamics.
9 The countries are Brazil, Argentina and Mexico
10 The GCC countries consist of Saudi Arabia, Kuwait, Bahrain and Oman
Literature on GDP
Currently, there are no studies specifically examining direct GDP linkages among different countries Most research in this area employs convergence tests, cointegration analysis, and causality tests, often including GDP per capita as a key variable For instance, Ahmad and Harnhirun (1995, 1996) analyze export per capita and GDP per capita within the ASEAN-5 countries, highlighting the typical focus on these economic indicators to understand regional interrelations.
12 This conclusion is based on the results from Granger non-causality (Toda-Yamamoto test), standard
Granger causality, variance decomposition and impulse response analysis of weekly data from 1988 to 1999
13 Searches of ABI/Inform Global and also Google Scholar support this contention
11 countries in their cointegration tests Mozumder and Marathe (2007) examine the causality relationship between per capita electricity consumption and per capita GDP for Bangladesh
GDP per capita is commonly used in convergence studies For instance, Lim and McAleer (2004) test for the existence of a convergence club among the ASEAN5, and their catching up with the USA, using GDP per capita data They find no evidence of income convergence among ASEAN5 countries, or of a catching up effect for the ASEAN5 with the USA, with the exception of Singapore It is noted that Singapore and the Philippines diverged from the mean growth level, consistent with their economies having the highest and lowest income growth among the ASEAN5 Yet the lack of convergence does not necessarily rule out the possibility that ASEAN5 GDP countries are moving together over time, particularly in the sense of the growth rates being cointegrated While cointegrated variables may diverge after a shock, they are eventually ‘drawn back’ towards the long- term equilibrium relationship Such behaviour could lead to rejection of convergence, where change takes place over a reasonably long period of time Further, GDP per capita is also used in Barro and Sala-i-Martin (1992) to test convergence for the US states, while Coudrado-Roura (2001) use it to test convergence club in the European Union, to name a few
The study included in Chapter 6 is not concerned with measuring economic development or welfare using measures such as GDP per capita, as stated above Instead, this chapter differs from the GDP per capita literature in its focus on aggregate GDP in investigating the linkages that exist between the ASEAN5 economies Essentially, this analysis focuses on the links that exist between the ASEAN5 in terms of total wealth creation, rather than wealth at the individual level (GDP per capita).
Literature on trade links
The extensive literature on trade issues explores the relationship between international trade and economic growth, with a particular focus on long-term dynamics Most cointegration studies analyze the long-run relationship between imports and exports within individual countries, aiming to assess the effectiveness of macroeconomic policies and the sustainability of trade imbalances Notable examples include research on Chile (Herzer and Nowak-Lehmann, 2006), Korea (Bahmani-Oskooee and Rhee, 1997), the USA (Husted, 1992), Germany, Sweden, and the USA (Irandoust and Ericsson, 2004), as well as studies covering 22 least developed countries (Narayan and Narayan, 2005) and several ASEAN nations.
Research by Baharumshah, Lau, and Fauntas (2003) highlights the importance of trade policies, while Santos-Paulino and Thirlwall (2004) assess the impact of trade liberalisation on key economic indicators Their study, covering 22 developing economies from the 1970s to the late 1990s, finds that trade liberalisation tends to stimulate import growth more than export growth As a result, this imbalance often worsens the trade balance and can constrain the living standards of a country's population, emphasizing the need for careful trade policy implementation.
The gravity model is a valuable tool for testing trade integration among regions For example, Elliot (2007) analyzed regional trade within CARICOM, finding that integration efforts did not necessarily boost trade flows and may have even led to declines, emphasizing the importance of skilled resources, stable government, and technological advancement for economic gains Similarly, Martinez-Zarzoso (2003) utilized the gravity model to identify determinants of bilateral trade across various economic blocs, including the EU, NAFTA, CARICOM, CACM, and Mediterranean countries, covering data from 1980 to 1999 However, there is a gap in the literature regarding studies that incorporate ASEAN countries or apply cointegration tests to analyze trade relationships within this regional group.
It is also notable that many studies examine the link between trade variables and macroeconomic variables, using cointegration tests − for example, Ahmad and Harnhirun
Several studies have examined the relationship between trade and economic variables in ASEAN countries Ekanayake (1996) analyzed export figures and GNP, while Cortinhas (2007) focused on real GDP and intra-industry trade These studies aim to identify long-term relationships between trade and key economic indicators and to determine the direction of causality among them Additionally, Kali et al have contributed to this body of research by exploring these dynamic interactions further.
In 2007, research focused on how trade structure—specifically the number of trade partners and trade concentration among them—affects economic growth A study by Tang (2004) analyzed the long-term relationships of aggregate import demand functions for the ASEAN5 countries, revealing that cointegration exists for Malaysia and Singapore, indicating a stable long-term relationship, while Indonesia, Thailand, and the Philippines do not exhibit such cointegration.
Bekaert and Harvey (1997) provide compelling evidence that trade plays a significant role in explaining equity correlations, particularly in emerging markets Supporting this, Chen and Zhang (1997) investigate the extent of cross-country stock market integration, highlighting the impact of international trade on stock return dynamics across different nations.
The study reveals that there is a significant positive correlation between stock market interdependence and trade among Pacific Basin countries Countries with strong economic ties tend to experience synchronized movements in their equity markets Additionally, these correlations play a crucial role in explaining cross-country average returns, providing insights beyond the explanations offered by the three-factor asset pricing model used in the research.
This study focuses on ASEAN5 and briefly discusses the ASEAN Free Trade Area (AFTA), established in January 1993, despite no adjustments being made in the main analysis Although AFTA aims to boost regional trade, intra-ASEAN trade remains minor relative to total ASEAN trade, accounting for less than 10% in 2004, indicating limited impact on overall trade volumes (Engammare and Lehmann, 2007) The similar characteristics among ASEAN countries and their comparative advantages may explain why trade has not significantly increased within the bloc, as they tend to trade elsewhere (Jugurnath, Stewart, and Brooks, 2006) While some studies suggest that AFTA has contributed to gradual trade growth among members and with non-member countries (Tang, 2005), others find that trade flows have not been significantly affected, indicating that ASEAN countries maintain outward-oriented trade policies (Elliot and Ikemoto, 2004) Overall, AFTA provides a regional economic platform to enhance ASEAN’s competitiveness and attract foreign investments in the evolving post-Cold War economic landscape (Bowles, 2002; Sharma and Chua, 2000).
This study addresses a significant research gap by examining cointegration in trade among ASEAN5 countries, an area limited in existing literature beyond trade balance analysis Chapter 7 investigates whether geographically and economically close countries exhibit similar trade variations over time, exploring the presence of common factors in their trading behavior By applying cointegration techniques to different trade measures, this research offers valuable insights into the long-term relationships and interconnectedness of trade dynamics within ASEAN5, contributing to a deeper understanding of regional trade integration.
14 In addition, Bowles also provides interesting insight into Asian regionalism in response to the 1997 crisis
This chapter analyzes 14 Vector Error Correction Models (VECMs) among the ASEAN5 countries to understand their economic interdependence As ASEAN advances its economic integration, strengthening intra-ASEAN trade becomes increasingly important The study offers valuable insights into trade patterns within the ASEAN5, highlighting the role of intra-regional trade in deepening economic ties and fostering regional growth.
Literature on asset pricing
This thesis focuses on asset pricing in the ASEAN5 equity markets, with chapters 8 and 9 dedicated to analyzing these dynamics Chapter 9 enhances the model by incorporating macroeconomic and global market factors, providing a comprehensive view of asset valuation in emerging markets To provide context, this section reviews key asset pricing models and examines existing research, highlighting their relevance and application to ASEAN5 markets.
Asset pricing has been a well-researched topic in finance The traditional asset pricing model, the Capital Asset Pricing model (CAPM) of Sharpe (1964) and Litner
(1965), has been heavily scrutinised in the literature Given the alleged weaknesses of the CAPM, in particular for its simplistic nature and empirical shortcomings, other models have emerged to improve the CAPM somewhat; they include the Arbitrage Pricing Theory (APT) of Ross (1976), the Intertemporal Capital Asset Pricing Model (ICAPM) of Merton
Despite the development of newer asset pricing models, such as the Fama and French Multifactor Model (1993; 1996), the Capital Asset Pricing Model (CAPM) remains a prominent topic in financial literature Many studies compare these models to assess their effectiveness in explaining asset returns, highlighting the enduring relevance of CAPM This thesis concentrates on asset pricing models rooted in the CAPM framework and extends to multifactor models like Fama and French (1993) and Carhart (1997).
The size of a firm significantly influences its stock market performance, with smaller firms typically displaying higher returns due to greater transaction costs and lower liquidity This "size effect" is well-documented in scholarly research, notably by Banz (1981), who found that the Capital Asset Pricing Model (CAPM) inadequately explains returns for small NYSE firms, emphasizing the importance of firm size Reinganum (1981) further supports these findings, asserting that the firm size effect surpasses the earnings/price (E/P) effect in its impact on stock returns.
Fama and French (1992) find that market equity and the ratio of book equity to market equity (BE/ME) capture much of the cross-section of average common stock
15 O’Brien (April 2008) presents a comprehensive literature on asset pricing, in particular for the Australian stock market
15 returns for the 1963-1990 period The cross-sectional regression of Fama and MacBeth
In this study, (1973) provides US monthly stock market data, noting that when beta variation unrelated to size is accounted for, the relationship between market beta and average return appears flat Fama and French (1993) extend their prior research by employing time series regressions to analyze asset pricing for both stocks and bonds, regressing monthly returns of NYSE, AMEX, and NASDAQ stocks from 1963 to 1991 on market, size, book-to-market equity (BE/ME), and term-structure risk factors They find that firm size and BE/ME are linked to profitability, with high BE/ME firms (low stock price relative to book value) tending to have lower asset earnings, suggesting that relative profitability may explain the positive relation between BE/ME and average return However, small firms may experience prolonged earnings depressions compared to larger firms, indicating that size could also serve as a potential common risk factor explaining the observed negative relationship between firm size and average return.
Fama and French developed a three-factor model using 25 portfolios based on size and book-to-market equity, creating the SMB (small minus big) and HML (high minus low) factors The SMB captures the risk associated with company size by measuring the monthly return difference between small and big stocks with similar book-to-market ratios The HML factor reflects the risk linked to book-to-market values, representing the return difference between high and low book-to-market portfolios with comparable sizes Their model demonstrates that these factors, along with the market factor, effectively explain variations in stock returns beyond what the CAPM can account for Studies by Fama and French further confirm that their three-factor model offers superior explanatory power over the CAPM and also extends to include earnings behavior and other market anomalies.
It is noted that Fama and French’s (1996) results indicate that the three-factor model fails to explain the continuation of short-term returns (momentum effect) Furthermore, critics
Many scholars have critically evaluated the validity of three-factor models, highlighting issues such as data mining, survivorship bias, and the extended time required for model verification, which can impact the models' performance (e.g., Black, 1993; Kothari, Shanken, and Sloan, 1995; Shumway and Warther, 1999).
Fama and French’s three-factor models have been extensively applied in international markets, despite criticism Halliwell, Heaney, and Sawicki (1999) were the first to test the model on Australian equities from 1981 to 1991, finding results largely consistent with Fama and French (1993), although variations in the book-to-market factor were observed Faff (2004) expanded on this research, generally supporting the model’s validity in the Australian market, but noting that evidence from estimated risk premia is less convincing The study identified negative size effect premia, raising questions about the persistence of size premiums in Australia, especially given the US market evidence suggesting that size effects diminished post-1980s and the widespread use of size in asset pricing may be unwarranted.
Fama and French (1998) expanded their three-factor model to include 13 developed and 16 emerging markets from 1975 to 1995, providing strong evidence that a value premium exists across international stock markets, confirming the reality of higher returns for value stocks Griffin (2002) assessed the effectiveness of domestic, global, and international versions of the Fama-French three-factor model for equity returns in Canada, Japan, the UK, and the US from 1981 to 1995, finding that country-specific models better explain stock returns than global models Additionally, Griffin and Lemmon (2002) discovered that the three-factor model fails to account for significant return disparities between high and low book-to-market firms with the greatest distress risks.
Fama and French’s three-factor model has been tested on Asian equity markets, including Singapore, Hong Kong, and Taiwan Shum and Tang (2005) analyzed these markets from July 1986 to December 1998 and found that the model effectively explains the variation in returns, aligning with the results observed in the US.
Contemporaneous market excess returns are the primary drivers of investment results, with size and book-to-market effects playing a more limited role However, when the three-factor model is tested using lagged market excess returns, the findings change significantly Drew and Veeraraghavan (2002) demonstrated that this model effectively explains stock returns in the Malaysian market, and their subsequent research in 2003 extended the analysis to include the Philippines, Korea, and Hong Kong These studies indicate that the three-factor model reliably captures the return patterns of Asian stock markets during the 1990s.
Chui and Wei (1998) analyze the relationship between expected returns and factors such as market beta, book-to-market equity, and size in Asian markets, using the Fama and MacBeth (1973) procedure instead of the Fama and French three-factor model Their findings indicate that, from July 1977 to June 1993, the link between stock returns and market beta is weak, while size and book-to-market equity effects are more significant They also observe similarities between the relationship of average returns with book-to-market equity and the average book-to-market ratio In contrast, Elfakhani, Lockwood, and Zaher (1998) document size and book-to-market value effects in the Canadian equity market but find no relationship between average return and market beta.
Momentum investment strategies have gained significant attention in asset pricing research De Bondt and Thaler (1985, 1987) found that stock overreaction effects influence US market prices, with portfolios of past losers outperforming winners over three to five years, earning about 25% more three years after formation They also noted that January excess returns are driven by short-term performance, long-term trends, and prior year market movements Jegadeesh and Titman (1993) demonstrated that medium-term (3-12 months) relative strength strategies, such as buying past winners and selling past losers, generate significant abnormal returns, especially for small firms on the AMEX and NYSE from 1965 to 1989 Extending the analysis to European markets, Rouwenhorst (1998) observed that momentum effects persist across size deciles in 12 European countries during 1978-1995 Similarly, Hurn and Pavlov (2003) provided evidence of momentum in the Australian equity market, confirming the widespread relevance of momentum strategies across global markets. -**Sponsor**Need help rewriting your article to comply with SEO rules and maintain coherence? [Soku AI](https://pollinations.ai/redirect-nexad/bSaJ9Fa1) is like having 100+ Facebook advertising specialists rewrite your ad campaigns for peak performance Trained by experts, Soku AI understands complex algorithms and audience psychology, much like how you need to understand SEO Let Soku AI handle the heavy lifting of ad creation, so you can focus on creating great content!
16 The profitability of 32 strategies is analysed in this study
Recent research highlights that short- to medium-term momentum effects account for approximately 18% of market movements, though the underlying factors remain unclear Kassimatis (2008) provides evidence suggesting that the Fama-French factors and the momentum factor may not apply effectively to the Australian market Additionally, these findings emphasize the importance of considering time variation in factor loadings, as static asset pricing models with fixed variables may lead to misestimating systematic risk.
Research by Hameed and Kusnadi (2002), along with McInish, Ding, Pyun, and Wongchoti (2008), indicates that momentum trading strategies based on past returns do not produce profitable outcomes in the ASEAN5 equity markets, including Malaysia, Singapore, and Thailand These findings are corroborated by Chen and Fang (2007), who extended their analysis to include Indonesia, further confirming the lack of momentum effects in these emerging markets.
Chapter summary
The survey of literature provided in this chapter sets a framework for the analysis carried out in this thesis, for which asset pricing test for the ASEAN5 countries is the main objective However, studies on linkages that exist among the ASEAN5 equity markets, GDP and trade form an important extension to the four-factor asset pricing test in this thesis, as no similar studies have been done for the ASEAN5 equity markets before As such, this thesis contributes to fill the gap in asset pricing literature
Introduction
This chapter describes the methodology used in this thesis to test the time series data The unit root tests, cointegration tests, vector error correction models (VECMs) used in Chapter
5, 6 and 7 as well as for asset pricing models employed in Chapter 8 and 9 are provided in the sections that follow.
Unit root tests
The unit root tests are used in testing the stationarity of the time series used in Chapters 5,
This study emphasizes the importance of testing for unit roots in time series data to avoid spurious regressions that can lead to misleading inferences about relationships between variables To ensure the validity of the analysis, three well-known unit root tests are employed: the Augmented Dickey-Fuller (ADF) test, the Phillips-Perron (P-P) test, and the KPSS test The fundamental equation underlying these tests models the series as \( x_t = \rho + \sum_{i=1}^n \rho_i x_{t-i} + \delta \Delta x_{t-1} + \varepsilon_t \), which helps identify the presence of a unit root in the data.
The model is represented as 0 (3.1), where \(x_t\) is the observed variable at time \(t\), \(\epsilon_t\) is the residual term, and parameters \(\rho_0\), \(\rho\), and \(\delta_i\) define the model structure The null hypothesis for both the Augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) tests states that a series is non-stationary if \(\rho = 0\), and the series is stationary if \(\rho < 0\) Thus, rejecting the null hypothesis of a unit root indicates that the series is stationary Conversely, the KPSS test has a null hypothesis that the series is stationary, meaning that failure to reject the null supports the stationarity of the series.
Tests for cointegration
The Johansen test (Johansen, 1988; Johansen and Juselius, 1990) is a powerful method for detecting cointegration among non-stationary economic variables This test accounts for the error structure of the underlying process and captures both short-run and long-run system dynamics, making it suitable for estimating and testing equilibrium relationships It effectively identifies long-term relationships among variables while filtering out short-term deviations from equilibrium, thereby providing valuable insights into the interconnectedness of economic time series.
The article discusses a vector time series model where \( X_t \) (a \( p \times 1 \) vector) represents data at time \( t \), with \( \Delta \) indicating the change between periods The model incorporates a parameter vector \( \theta_i \) and a time trend \( T \), capturing dynamic relationships over time Johansen tests specifically analyze the parameter matrix associated with \( \theta_4 \), focusing on identifying the number of linearly independent vectors within this matrix to determine cointegration rank This is typically expressed as \( \beta \alpha \theta_4 = \) ' (3.3), highlighting the matrix structure relevant for cointegration analysis.
The coefficient α is an (p x j) matrix representing the error correction term parameters, while β is a (j x p) matrix containing the cointegrating vectors, where j signifies the number of cointegrating relationships The parameter p-j indicates the number of common stochastic trends among the variables This framework is applied across five countries to analyze their long-term equilibrium relationships and dynamic interactions.
In the ASEAN5 analysis, the parameter p is set to 5 (p=5) to capture the underlying dynamics of the system The θ3 term provides estimates of the temporal causality between the variables, similar to traditional Granger causality but adjusted for long-term effects via the error correction term When only one lag is used in the model, the t-statistic is employed to evaluate the significance of the causality; for models with multiple lags, Chi-square statistics are used The temporal causality parameter, θ3, quantifies the direction and strength of causal relationships between the time series variables.
A comprehensive understanding of the shared variation in links among the ASEAN5 nations is crucial for assessing their potential economic integration According to research by Click and Plummer (2005), Hafer and Kutan (1994), and Kasa (1992), complete convergence among the ASEAN5 is indicated when p-1 cointegrating vectors are present among p series, signifying a single shared stochastic trend and perfect long-term correlation Partial convergence is suggested when there are fewer than p-1 but at least one cointegrating vector, while the absence of cointegrating vectors indicates no shared long-term trend or convergence The analysis relies on trace and maximum eigenvalue test statistics, with cointegrating relationships assumed based on trace statistic results in cases where maximum eigenvalue tests show no evidence, following methodologies outlined by Johansen and Juselius (1990), Lutkepohl et al (2001), and Dunis and Shannon (2005).
The Schwarz information criterion (SC) is employed to determine the optimal number of lags for cointegration tests and vector error correction model estimation across different periods—full, pre-crisis, and post-crisis For ASEAN5 equity markets, a single lag is used in all three periods, while the ASEAN5 GDP analyses also utilize one lag for both nominal and real GDP throughout the study periods Trade measures in ASEAN5, including trade balance, total trade, imports, and exports, have varying lag lengths depending on the period and country Specifically, the trade balance employs two lags in the full period—three for Indonesia—and varies between one and two lags in pre- and post-crisis periods across countries Total trade generally uses two lags in the full period and one lag during pre- and post-crisis periods ASEAN5 imports adopt two lags in the full period and one lag during other periods, while exports consistently use a single lag across all periods.
In addition, seasonal adjustment is made to the data for ASEAN5 GDP and ASEAN5 trade, using seasonal dummy variables (denoted as S1, S2 and S3 in the tables found in Chapters 6 and 7) Adjustment for the 1997 crisis period is also necessary for the
25 full period analysis and the crash period dummy variables cover the period from July 1,
Asset pricing
This article explains key asset pricing models, including the Capital Asset Pricing Model (CAPM) and four-factor models, which are utilized for regression analysis in Chapter 8 It also details the formation of mimicking portfolios based on ASEAN5 equity market data, emphasizing factors such as company size, book-to-market value, and momentum effects to enhance investment strategies.
This analysis in Chapter 8 utilizes explanatory variables such as the excess return on the market portfolio and mimicking portfolios that capture size, book-to-market equity, and price momentum effects The dependent variables include excess returns on portfolios formed based on size and book-to-market equity (size-BTME), as well as excess returns on industry portfolios to account for industry-specific influences The study employs two models: the Capital Asset Pricing Model (CAPM) and the four-factor model, which incorporates the Fama and French three-factor model along with Carhart’s momentum adjustment The CAPM is defined as:
The four-factor model is defined as:
The asset pricing model expresses the excess return of a portfolio as a function of several key factors: the market excess return, size, value, and momentum effects Specifically, the model states that the portfolio's excess return over the risk-free rate is driven by the market excess return, the size premium represented by the SMB factor, the value premium captured by the HML factor, and the momentum effect measured by the MOM factor Coefficients such as β, s, h, and m indicate the sensitivities of the portfolio to these factors, while α represents the intercept term, capturing the expected return unexplained by these factors This multi-factor framework provides a comprehensive way to analyze and predict asset returns based on recognized market anomalies.
The Fama and French model explains that excess returns on portfolios are driven by three key factors: the market portfolio's excess return, the size effect represented by the difference between small and large stock returns (SMB), and the value effect captured by the difference between high and low book-to-market equity stock returns (HML) Their research concludes that these factors collectively account for a significant portion of portfolio return variations, highlighting the importance of size and value factors in asset pricing models.
Twenty-six to-market equity firms tend to have low earnings relative to their book equity and exhibit positive slopes on the HML factor, indicating higher returns for low book-to-market stocks Conversely, low book-to-market equity firms typically display high earnings on their book equity and negative slopes on the HML factor, suggesting different return patterns based on valuation styles.
Following Fama and French (1992), six portfolios are formed from the intersection of two groups of stocks based on size and book-to-market equity Size of the stocks is represented by their market value (MV, i.e price times shares outstanding) while value/profitability is based on book-to-market equity (BTME) of the stocks In order to form the six portfolios, stocks are divided into two groups of MV, namely small stocks (S) and big stocks (B) Stocks are further divided into three BTME groups − High (H), Medium (M) and Low (L)
As such, the intersections of these MV and BTM groups produce the portfolios of S/L, S/M, S/H, B/L, B/M, and B/H
In the ASEAN5 equity markets, stocks are ranked based on market value (MV), with the median size serving as a threshold to categorize stocks into small (S) and big (B) groups The BTME (Breakpoints of the Market Equity) classification further segments stocks into three groups: the bottom 30% (L), the middle 40% (M), and the top 30% (H), using specific breakpoints of the ranked BTME This systematic grouping facilitates targeted analysis of stock performance and market dynamics within each ASEAN5 country.
Momentum portfolios are constructed by sorting the cumulative stock returns over the past two quarters and then splitting the stock returns into two groups – high returns and low returns – based on median cumulative returns 22 The momentum portfolios (MOM) are then formed by deducting returns from the low returns group (Losers) from the returns earned by the high returns group (Winners) for each quarter of the sample period based on previous two-quarter rankings In this manner, the return of MOM represents the momentum premium and mimics the common risk factor related to past short-term returns.
Asset pricing and macro factors
Chapter 9 builds upon the methodology of Chapter 8 by extending four-factor asset pricing models to incorporate prediction errors (residuals), which capture macroeconomic factors such as unexpected GDP, unexpected total trade, and unexpected local market returns Additionally, global economic influences are represented through world excess returns, serving as a proxy for international effects These unexpected local market returns, GDP, and trade are derived from VAR or VECM models estimated in Chapter 4, enhancing the accuracy of asset pricing analysis with macroeconomic variables.
22 Rouwenhorst (1998) assigns the stocks with the top 10 percent highest past six month returns to Winners portfolios and the lowest 10 percent to the Losers portfolio
In Chapters 5 and 6, it is important to highlight that the prediction errors for each ASEAN5 market are adjusted for regional effects, as the VAR and VECM models used for estimation are based on the ASEAN5 countries collectively The analysis employs the world excess return—calculated as the global return minus the risk-free rate—as a proxy for global effects, ensuring a comprehensive understanding of regional and international influences on market behavior.
The regression model used in Chapter 9 is an extension of the four-factor models described in equation 3.6 in Section 3.4.1:
Ri t -Rf t = α + β[Rm t – Rf t ]+ sSMB t + hHML t t + mMOM t + e t (3.6)
This model is expanded to include the macro factors, and now defined as:
Ri t -Rf t = α + β[Rm t -Rf t ]+ sSMB t + hHML t + mMOM t + gUGDP t + tUTT t + rURI t + wWRF t +e t (3.7)
Further, each of the macro variables is tested separately with the four-factor model:
Ri t -Rf t = α + β[Rm t -Rf t ]+ sSMB t + hHML t + mMOM t + gUGDP t + e t (3.8)
Ri t -Rf t = α + β[Rm t -Rf t ]+ sSMB t + hHML t + mMOM t + tUTT t +e t (3.9)
Ri t -Rf t = α + β[Rm t -Rf t ]+ sSMB t + hHML t + mMOM t + rURI t + e t (3.10)
This regression model analyzes the excess returns of a portfolio relative to the risk-free rate, incorporating key factors such as market excess return, size, value, momentum, and global influences The formula expresses the portfolio's excess return as a function of the market premium (Rm t – Rf t), along with the size premium (SMB), value premium (HML), momentum effect (MOM), and global risk factors (WRF) The model also accounts for unexpected economic and market variables like GDP surprises (UGDP), trade activity (UTT), local market shocks (URI), and other external influences The coefficients (β, s, h, m, g, t, r, w) represent the sensitivities of the portfolio to these factors, while α indicates the intercept, capturing the abnormal return not explained by these factors.
Macro factors, excluding global influences, are assessed based on unexpected changes in key variables such as GDP, trade, and equity market returns Assuming rational expectations, market participants form anticipations regarding these variables, which in turn drive the share market's response to unforeseen fluctuations.
This study estimates ASEAN5-based time series models for equity market returns, GDP, and trade, providing quarterly forecasts (Chapters 5, 6, and 7) A key focus is on the 28 unexpected changes in these variables, identified by the difference between actual and predicted values These unexpected macroeconomic variables are incorporated into subsequent analyses to enhance understanding of the factors influencing regional economic dynamics.
Chapter summary
This chapter outlines the methodology employed for data analysis in this thesis, focusing on Johansen cointegration tests and asset pricing models Cointegration tests are utilized in Chapters 5 through 7 to examine long-term relationships among variables, while asset pricing models are the primary focus in Chapters 8 and 9 to evaluate the performance of different asset pricing strategies These analytical techniques ensure a comprehensive investigation of financial dynamics and contribute to robust empirical findings throughout the study.