scholarship project paper 2016 risk and return in the ... · pdf fileunderstanding of the...

www.set.or.th/setresearch

Disclaimer: The views expressed in this working paper are those of the author(s) and do not necessarily

represent the Capital Market Research Institute or the Stock Exchange of Thailand. Capital Market

Research Institute Scholarship Papers are research in progress by

the author(s) and are published to elicit comments and stimulate discussion.

Scholarship Project Paper 2016

Risk and Return in The equity mutual fund industry: An unorthodox relationship and its application to new investment strategies

Suchaya Siamwalla

Assumption University

20 June 2016

Abstract In 1980, Bowman proposed the Bowman paradox, a negative relationship between the risk and return in 85 U.S. industries, a contradiction with the high risk-high return doctrine. Examining the open-end equity mutual funds in Thailand, this study documents the negative relationship that can occasionally be seen between risk and return in the industry during 2003-2012. The study further examines the factors that will affect the probability that a fund will deliver an outstanding low-risk, high-return performance using unbalanced panel logistic regression on a binary dependent variable. The results showed that funds with high non-systematic risk, also called idiosyncratic risk, and/or older funds, were more likely to deliver a low-risk, high-return performance, and the funds that were managed by the company that managed a high number of funds were less likely to deliver such performance. This study proposes a new performance evaluation tool called the “risk-return matrix.” This matrix suggested the funds with outstanding low-risk, high-return past performance.This study applied the results to three new investment strategies. All simulations demonstrated returns better than the industry’s average returns. JEL Classification: G11, G12 Keywords: Mutual Fund Performance, Bowman Paradox, Negative correlation, Risk and Return, Portfolio E-Mail Address: [email protected]

mailto:[email protected]

2016 Capital Market Research Institute, The Stock Exchange of Thailand

Content

Page

Chapter 1 Introduction 1

Chapter 2 Literature Review 4

Chapter 3 Research Framework and Hypotheses 8

Chapter 4 Research Methodology 13

Chapter 5 Statist ical Results 26

Chapter 6 Conclusion and Implicat ions 43

Bibliography 52


Table

1. Summary of the Hypotheses and their Expected Signs 9

2. Operationalization of the Independent and Dependent Variables 11

3. Descriptive Data of Sample Funds 26

4. The Correlation between the Risks and Returns of the Thai

Open-End Equity Mutual Fund Industry from 2005-2012 27

5. The Correlations between the Three-, Five-, and Ten-Year

Annualized Return and Standard Deviations. 28

6. The Distribution of Fund Performance during 2005-2012 29

7. The Risk-Return Matrix for 2005-2012. 29

8. The Logit Regression of the Characteristics of the

Low-Risk, High-Return Fund 32

9. The Summary of the Results of the Hypothesis Testing 34

10. The Portfolio of the Low-Standard Deviation Funds vs

the High- Standard Deviation Funds 35

11. The Return and Standard Deviation of the Portfolio of

the “Sweet Spot” Winner 40

12. Portfolio Emphasizing the Characteristics of the “Sweet Spot” Fund. 41

Figure

1. Research Framework 9

2. The Risk-Return Matrix 18

3. The Average Performance of Each Group during 2005-2012 30

4. The Scatter Plot between the Average Returns and Standard Deviations

of the Portfolios during 2006-2012 37

5. The Distribution of Funds with a “Sweet Spot” Performance (2005-2011) 39


Chapter1 Introduction

The conventional wisdom of economic theory suggests that one that expects a

higher return should be able to assume higher volatility (Merton, 1973). However, in 1980, one

of the studies in the management discipline affected this long-time widely accepted notion, a

publication in the Sloan Management Review entitled “A Risk/Return Paradox for Strategic

Management” (Bowman, 1980). It is known as the “Bowman paradox”.

Based on business administration and economic rationality, the managers of a firm

will expect high return from a high-risk investment (Bowman, 1980). The empirical study of

Bowman (1980) demonstrated a contradicting result. Using firm-level analysis of 85 U.S.

industries, Bowman found that the companies that had a profit higher than average tended to

have lower volatility of profit across the companies within industries. His study is also

applicable to financial study since Bowman defined risk as the volatility of the company profit.

He used the standard deviation of the return on equity as a proxy for risk.

There has been an increasing interest in extending the study of the Bowman

paradox, especially in the United States, Europe, and other Western countries (Baucus, Golec,

& Cooper, 1993; Fiegenbaum & Thomas, 1986; Guo & Whitelaw, 2006; Wiseman & Bromiley,

1991). Recently, Brockett et al. (1992) used chance constrained programing to study the

correlation between the risk and return of 383 U.S. open-end mutual funds during 1984-1988,

and Cooper et al. (2011) examined the quantitative evaluation of strategic performance of 120

U.S. equity mutual funds during 1993-1997. Both studies came to similar conclusions that the

Bowman paradox could appear in the mutual funds industry.

There are many reasons that support the conduct of this study. First, despite the

growth in the mutual fund industry in the emerging market, the mutual fund research of this

market is still limited (Ungphakorn, 2014). Second, the Bowman paradox was evidenced in the


2

U.S. mutual fund industry: during 1984-1988 (Brockett et al., 1992), and during 1993-1997

(Cooper et al., 2011). However, to the author’s best knowledge, a study of the Bowman

paradox using Thai mutual fund samples has not yet been documented, despite, the high

potential of the Thai financial market. The Thai stock market has had the highest daily trading

value among the ASEAN countries for three consecutive years and the Thai mutual industry

has had 23% growth in 2014 (Bank of Thailand, 2015; Deloitte, 2012; SET, 2014).

This study believes that the investigation of the paradox will be more useful for

investors if they are able to understand the factors that affect the low-risk, high-return

performance of mutual funds. The author was interested in finding out those factors and hopes

that the examination of the paradox in Thai equity mutual fund industry will increase the

understanding of the risk-return relationship in this industry, which may help investor enhance

their portfolio-returns.

1.1 Research Objective

This study has three objectives.

1. To investigate whether the paradox exists in the Thai equity mutual fund industry.

2. To find out the characteristic of the low-risk, high-return performance funds.

3. To find out whether, in the market where a negative risk-return relationship exists, investors

can capitalize on this phenomenon.

1.2 Expected Outcomes

1. This study expected to find that there is a negative relationship between risk and return in

the Thai open-end equity mutual funds industry.

2. This study expected that some of the organizational factors and fund characteristics, such

as the non-systematic risk, fund size, fund objective, type of parent company, fund age,


3

total number of funds, and the total assets under management within the same asset

management company, would have significant effect on the low-risk, high-return

performance funds.

3. This study hoped to offer some investment strategies that opens the opportunity for Thai

investors to gain benefit from the phenomenon. That is they can have a low-risk, high-

return investment.


Chapter 2 Literature Review

2.1 The Modern Portfolio Theory

Investors should maximize the expected return while minimizing risk (Markowitz,

1952). This notion of the modern portfolio theory of Markowitz pave the way to the Tobin’s

seperation theorem (Tobin, 1958), and the capital asset pricing model, the CAPM (Sharpe,

1964; Lintner, 1965; Mossin, 1966). The essence of the CAPM is that the asset can be priced

based on its co-variation with the systematic risk of the market. The non-systematic risk of the

assets will be eventually diversified, so need not be priced. If the CAPM is considered the first

breakthrough in the financial economic discipline during the 1960s, the arbitrage pricing theory

(APT) should be considered the second breakthrough during the 1970s. The CAPM indicates

that a single risk factor, beta, is accountable to the portfolio volatility, but APT of Ross (1976)

expanded the definition of systematic risk to cover several risk factors. Nevertheless, the

breakthrough from the arbitrage pricing theory (APT) in 1976 led to the development of the

multifactor models (Carhart, 1997; Chen, Roll, & Ross, 1986; Fama & French, 1993). During

the 1980s, various variables that were not previously part of the asset pricing model emerged.

2.2 The Bowman Risk-Return Paradox

A long generally-accepted notion that a high-risk investment should eventually deliver

a high-return became conspicuous in a study of Bowman (1980), later known as the Bowman

paradox. Bowman (1980) examined the average profit and volatilities of companies from 85

industries to cover 1,572 companies during 1968-1976. The results showed that 56 industries

supported the hypothesis of a negative risk-return relationship, 21 industries did not support it,

and 8 industries were tied. He concluded that the examination supported the negative

correlation hypothesis. Bowman paradox aroused much academic attention. Many studies

tried to explain the paradox from various angles, which can be categorized into three groups of

explanations (Andersen, Denrell, & Bettis, 2007).


5

The first group explained the paradox based on behavioral theory, such as prospect

theory, which holds that the risk-taking behavior of decision makers is influenced by the

expected outcome. Examples of researchers in this group are Fiegenbaum (1990),

Fiegenbaum and Thomas (1986, 1988), Jegers (1991), Johnson H. J. (1994), and Sinha

(1994). They believed that in a situation of low prospect, decision makers tend to take higher

risk.

The second group explained the paradox from the organizational management point

of view, such as Andersen et al. (2007), Bettis and Hall (1982), Bettis and Mahajan (1985), and

Miller and Chen (2003). This group believed that good management structure and practices

matter to the risk-return relationship.

The third group explained the paradox based on statistical artifacts, such as the

skewness of the return (Henkel, 2009), the choice of accounting data (Baucus, Golec, &

Cooper, 1993).

After three decades of continued study on the issue, a conclusion was not reached

as to why the paradox existed. Although a negative relationship between the risk and return of

investment was unorthodox, numbers of empirical studies confirmed that they existed, as

reviewed in the previous paragraphs. Recently the study of the Bowman paradox was

extended from industrial firms to the investment industry. Brockett et al. (1992) used the

chance constrained programming approach to study 830 U.S. mutual funds during 1984-1988.

Later, Cooper et al. (2011) examined 120 U.S. funds during 1993-1997 using a two-stage data

envelopment analysis (DEA) approach to evaluate the efficiency of fund strategic performances.

Both studies concluded that the paradox could happen in the mutual fund industry.


6

2.3 Performance Measurements of the Mutual Fund

Some investors use the mutual fund return to substitute for the mutual fund

performance. However, the Bank Administration Institute1 (BAI) required that the performance

measurement should include both risk and return elements (Bacon, 2008). There are several

types of performance measurements, which include both risk and return. For example the

return-based measurements and the holdings-based measurements. The example of the

return-based measurement are Treynor ratio, Sharpe ratio, Treynor-Mazuy measurement,

Henriksson and Merton measurement, Jensen alpha. The example of the holdings-based

measurements are Grinblatt-Titman measurement (GT) and Characteristic Selectivity (CS) of

Daniel, Grinblatt, Titman, and Wermers (1997).

2.4 Factors Affecting the Mutual Fund Performance

2.4.1 Organizational Factor and the Role of Fund Characteristics

There have been many studies related to the characteristics of mutual funds on

various subjects, such as performance (See & Jusoh, 2012; Huang & Shi, 2013), the growth

opportunities of a firm value (Kogan & Papanikolaou, 2013), fund survivorship (Carhart,

Carpenter, Lynch, & Musto, 2002), and fund disappearance (Cameron & Hall, 2003). All of

these studies agreed that characteristics such as size (Berk & Green, 2004; Chen, Hong,

Huang, & Kubik, 2004), age (Switzer & Huang, 2007), fund objective (Daniel et al., 1997; Fama

& French, 1993) the fund-expense ratio, and others were important predictors of fund attrition,

fund performance, etc.

1 The Bank Administration Institute is a non-profit organization, founded in 1924. It aims to

promote banking industry standards in bank operations and auditing academics, conferences,

and a research affiliate. It is the oldest and largest technical organization serving the U.S.

banking industry.


7

The mutual fund by itself is not a stand-alone organization. Each fund belongs to a

fund family, which is managed under an asset management company. As a result, the

characteristics of a fund management firm such as the type of parent company (Bogle, 2005;

Ferris & Yan, 2009), total numbers of fund (Bogle, 2010; Nanda, Wang, & Zheng, 2004; Pollet

& Wilson, 2008) affected fund performance. The total asset under management (Bhattacharya,

Lee, & Pool, 2013; Gasper, Massa, & Matos, 2006; Jong & Wingens, 2013) also affected how

the fund carries its risk or delivers its return.

2.4.2 Diversifiable Risk: Idiosyncratic Risk

According to modern portfolio theory, the systematic risk of any portfolio plays an

important role in forecasting expected return, whereas the idiosyncratic risk does not have that

role. The reason is that the idiosyncratic risk is diversifiable; it is independent of the market

and has zero expected value. Hence, the idiosyncratic risk should not, theoretically, relate to

the expected return. However, its role on expected return has been documented. The

explanation were the increment of institutional ownership (Malkiel & Xu, 2003), the firm’s

fundamental characteristics such as maturity (Wei & Zhang, 2006; Irvine & Pontiff, 2009), etc.

The role of idiosyncratic risk on an asset return has been documented more at the security

level than at the mutual fund level. Moreover, there is a lack of such study in the Thai context.


Chapter 3 Research Framework and Hypotheses

3.1 Relationship between Risk and Return in the Thai Equity Mutual Fund Industry

Conventional wisdom leads us to believe that investors will require a high rate of

return if they expect to face high uncertainty, and they will require a smaller return if they can

be more certain about it. However, the theoretical relationship between the two does not

indicate any certainty with regard to the sign of the relationship (Backus & Gregory, 1993). The

fact that Bowman (1980) found a negative relationship between corporate returns and their

standard deviations made it a valuable addition to the literature in the financial economics

discipline. Additionally, previous studies suggested that a negative risk-return relationship could

happen in the U.S. mutual fund industry (Brockett et al., 1992; Cooper et al., 2011).

Hence, this study offers the first null hypothesis to whether there is a significant

positive or no relationship between risk and return in the Thai open-end equity mutual funds

industry.

3.2 Characteristics of Funds with Low-Risk, High-Return Performance

As reviewed in Chapter 2, this study proposed to test whether the idiosyncratic risk

level of the Thai equity open-end mutual fund has a significant effect on the probability that the

mutual fund will deliver a low-risk, high-return performance.

Much of the literature has studied the effect of the organizational factors and fund

characteristics on fund performance, but to the author’s best knowledge, the study of their

effects on the probability that the fund will deliver a low-risk, high-return performance has not

been documented. This study also examined these relationships.

This study proposed the research framework in Figure 1 as follows.


9

Figure 1: Research Framework

Source: Developed for this study

* = Mutual fund performance is a binary variable, and it is equal to 1 if the fund delivers a low-

risk, high-return performance in the testing period and equal to zero otherwise.

This study examined the effect of the organizational factors and fund characteristics

on the mutual fund performance using the second to the eighth hypotheses. This study

summarized the hypotheses and their expected outcomes in Table 1 as follows.

Table 1: Summary of the Hypotheses and Their Expected Signs

Hypotheses Expected outcome

𝑯𝟎𝟏: There is a significant positive or no relationship between risk and return in the Thai open-end equity mutual funds industry.

Reject the null hypothesis

𝑯𝟎𝟐: The idiosyncratic risk of a fund has no significant effect on the probability that a Thai open-end equity mutual fund will deliver a low-risk, high-return performance.

Reject the null hypothesis and find a positive significant effect

Risk specification factor Idiosyncratic risk (ѱ)

Organizational Factors and Fund Characteristics

Fund size (𝐿𝑜𝑔 𝑆𝑖𝑧𝑒) Fund objective (𝐷𝑂𝑏𝑗) Type of parent company (DType-parent) Fund age (𝐴𝑔𝑒) Total numbers of fund (No.Fund) Total Asser Under Management of company

(Log AUM)

Mutual fund performance*

(SS )

H2

H3-H8


10

Hypotheses Expected outcomes

𝑯𝟎𝟑: Fund size has no significant effect on the probability that a Thai open-end equity mutual fund will deliver a low-risk, high-return performance.

Reject the null hypothesis and find a negative significant effect

𝑯𝟎𝟒 : Fund objective has no significant effect on the probability that a Thai open-end equity mutual fund will deliver a low-risk, high-return performance.

Reject the null hypothesis and the growth objective is expected to have a negative coefficient -Other objectives are expected to have positive coefficients.

𝑯𝟎𝟓: The type of parent company of an asset management company has no significant effect on the probability that a Thai open-end equity mutual fund will deliver a low-risk, high-return performance.

Reject the null hypothesis and the bank-related type company is expected to have a negative coefficient -A non-bank-related type company is expected to have a positive coefficient.

𝑯𝟎𝟔 : Fund age has no significant effect on the probability that a Thai open-end equity mutual fund will deliver a low-risk, high-return performance.


𝑯𝟎𝟕: The number of funds under the management of an asset management company has no significant effect on the probability that a Thai open-end equity mutual fund will deliver a low-risk, high-return performance.

Reject the null hypothesis and find a negative significant effect

𝑯𝟎𝟖: The total assets under the management of an asset management company have no significant effect on the probability that a Thai open-end equity mutual fund will deliver a low-risk, high-return performance.




11

3.3 Operationalization of the Independent and Dependent Variables

The variables used in this study, their symbols, conceptual definitions, and their

operational definitions are summarized in Table 2 below.

Table 2: Operationalization of the Independent and Dependent Variables

Construct and Variable Name

Conceptual definition Operational definition

Mutual fund performance (SS)

A binary variable equal to 1 for the mutual fund that delivers a low-risk, high-return performance, and zero for other types of performance

A low-risk fund is a fund that is ranked in the lowest 30th percentile risk ranking. A high-return fund is a fund that is ranked in the highest 70th percentile return ranking.

Idiosyncratic risk of mutual fund (ψ)

The non-systematic risk of securities held by equity mutual funds that are not fully diversified

The square root of the sum of a regression residual squared The regression residual is based on the Fama-French three-factor model.

Log of fund size (𝐿𝑜𝑔 𝑆𝑖𝑧𝑒)

Net asset value of fund Log of the year-end net asset value

Fund objective (DObj)

Fund investment style reflects the main type of stocks the fund invests, whether they are the growth stocks, value stocks, or a mix of both.

A dummy variable. It is equal to one if the fund invests mainly in the growth stock and equal to zero if the fund mainly invests in other type of stocks. The assignment of fund objectives was inferred from the Morningstar® style box.

Type of parent company (DType-parent)

Type of parent company, whether they are bank-based or non-bank-based. It is inferred from the type of parent company that has the biggest share in an asset management company.

A dummy variable. It is equal to one for a bank-related parent company, and equal to zero for a for non-bank-related parent company. The information is gathered from the company’s public-disclosure information, annual reports, and websites.


12

Construct and Variable Name

Conceptual definition Operational definition

Fund age (Age) The age of a mutual fund Measured in years by counting from the month after the fund inception date to the cut-off month, and then divided by 12

Number of funds (No.Fund)

Total number of funds managed under an asset management company

Total number of all funds managed under an asset management company at the end of the year

Log of total assets under management (Log AUM)

Total assets under management

Log of year-end total net asset value of all funds managed by the firm



Chapter 4 Research Methodology

The sample selection is explained in 4.1. This study performed two separate statistical

tests. The first test, described in section 4.2, is the correlation procedure to test the first

hypothesis, 𝐻01, which is the risk and return relationship in a Thai open-end equity mutual

fund. The second test, presented in section 4.3, is the regression to test the hypotheses

𝐻02 − 𝐻08 , the factors that will affect the probability that a fund will deliver a low risk-high

return performance. The simulations in section 4.6 were tested to meet the third objective.

4.1 Sample Selection and Data Sources

This study selected all active equity funds in Thailand during 2003-2012. It excluded

international, closed-end, retirement (RMF) and long-term equity funds (LTF). Suppa-Aim

(2010) and Tangjitprom (2014) also employed this data collection method in their studies of

Thai equity mutual funds. The closed-end equity mutual funds in Thailand, which are mainly

triggered funds, RMF and LTF funds are managed under some regulations because they are

tax-benefited funds. This study excluded such funds because they have restricted risk patterns

and appetites. This study required funds with at least 36 months history for regression

purposes.

There were 118 funds under AIMC based on the AIMC2 definition of the equity fund

at the time of sample selection. After excluding 13 triggered funds, three index-tracked funds,

and 22 funds with age less than 36 months, this study examined a total sample of 64 funds in

2 AIMC is the Association of Investment Management Companies. Members consist of mutual

fund, private fund, and provident fund managers. AIMC is an association relating to and

representing investment management industry by the Office of the Securities and Exchange

Commission (SEC).


14

2003-2005, 70 funds in 2006, 78 funds in 2007, 80 funds in 2008-2009 and 83 funds in 2010-

2012.

4.2 The Risk and Return Relationship

This study tested the first hypothesis, 𝐻01, using the Pearson product moment

correlation coefficient, 𝜌𝑇𝑗 . This study correlated the return in excess of the risk free rate, 𝑟𝑛

𝑗 and

its standard deviation in one-year, three-year, five-year and ten-year timeframes. This study

assumed no transaction costs.

4.2.1 Return

This study used the total capital gain and dividend in calculating the return, as it

became a standard requirement in the latest 2010 Global Investment Performance Standards

(GIPS 2013) item 2.A.1.

(1). Calculate the monthly return by

(i) For a fund with no dividend pay-out policy, the 𝑟𝑛𝑗 reflects the capital gain as

follows.

𝑟𝑛𝑗

= {( 𝑁𝑎𝑣𝑒𝑛𝑑

𝑁𝑎𝑣𝑏𝑒𝑔𝑖𝑛) − 1} − 𝑟𝑓,𝑛

(ii) For a fund with a dividend pay-out policy, the 𝑟𝑛𝑗 reflects the capital gain and the

dividend income as follows.

𝑟𝑛𝑗 = {{(

𝑁𝑎𝑣𝑒𝑛𝑑

𝑁𝑎𝑣𝑥𝑑) . (

𝑁𝑎𝑣𝑥𝑑+𝐷𝑖𝑣 𝑝𝑒𝑟 𝑢𝑛𝑖𝑡

𝑁𝑎𝑣𝑏𝑒𝑔𝑖𝑛)} − 1} - 𝑟𝑓,𝑛

(2). Calculate the yearly return used the time-weighted rate of monthly return as its

measurement (Lawton & Jankowski, 2001), as follows

𝑟𝑌𝑗 = [∏ (1 + 𝑟𝑛

𝑗)𝑛

𝑖=1 ] − 1 (1)

(3). Annualize the three-year, five-year and ten-year return using geometric mean return.

𝑟𝑁𝑌𝑗 = √[∏ (1 + 𝑟𝑌𝑛

𝑗)𝑛

𝑖=1 ]𝑛

−1 (2)


15

П = The product of one plus monthly return from period 1 to n

𝑟𝑌𝑗 = Yearly return of fund j for the measurement year Y

𝑟𝑛𝑗 = Total return of fund j at month n, n=1-12

𝑟𝑓,𝑛 = Monthly risk free rate or T-bill rate.

𝑁𝑎𝑣𝑥𝑑 = Net asset value per unit at the end of the ex-dividend date.

𝑁𝑎𝑣𝑒𝑛𝑑 = Net asset value per unit of the last working day of the testing month.

𝑁𝑎𝑣𝑏𝑒𝑔𝑖𝑛 = Net asset value per unit of the last working day of the previous month.

𝑟𝑁𝑌𝑗 = The N years annualized return of mutual fund j.

𝑟𝑌𝑛𝑗 = The yearly return of mutual fund j at year n.

4.2.2. Risk

Following Fama and French (2012), this study employed standard deviation as its

risk measurement. This measurement is widely used by the mutual fund organization such as

Association of investment management companies (AIMC, 2016) and Morningstar®3

(Morningstar, 2016). The standard deviation is the square root of the volatility of monthly

returns of the testing year (Reilly & Brown, 2011) and multiplies by the square root of twelve to

make it annualized.

𝑆𝑇𝑗= √

1

𝑛−1∑ (𝑟𝑛

𝑗− 𝑟𝑗

𝑇 )

2𝑛𝑛=1 * √12 (3)

𝑆𝑇𝑗 = Annualized standard deviation of monthly return of mutual fund j for period T

𝑟𝑗𝑇

= Arithmetic mean of monthly return of mutual fund j for period T

𝑟𝑛𝑗 = Monthly return of mutual fund j at month n

The calculation used n = 36, 60 and 120 for the three-year, five-year, and ten-year

standard deviations, respectively.

3 Morningstar, Inc. is a provider of independent investment research in North America, Europe, Australia, and

Asia


16

This study also used Beta (β) as another proxy for risk since the slope angles of the

characteristic lines of the managed funds provide a refined measurement of the funds’ volatility

(Treynor, 1965), and it was widely used by many scholars (Black, Jensen, & Scholes, 1972;

Fama & MacBeth, 1973). Beta can be estimated from regression equation below.

𝑟𝑛𝑗= 𝛼 + 𝛽𝑇

𝑗 (𝑟𝑚𝑘𝑡,𝑛 -𝑟𝑓,𝑛) + 𝜀𝑛𝑗 (4)

𝑟𝑛𝑗 = The monthly return of mutual fund j at month n in excess of risk free rate

𝑟𝑚𝑘𝑡,𝑛 = Market return in month n

𝑟𝑓,𝑛 = Risk free rate in month n

4.2.3 Correlation between Risk and Return

This study used the Pearson product moment method because it is a suitable

method for the Interval or ratio scaled (Coakes & Ong, 2010; Ho, 2006). There is no theoretical

guideline as to how many years should be used to study the relationship between risk and

return. However, a study by the Investment Company Institute indicated that about 40 percent

of investors had an average of five years between the first purchase and redemption (ICI,

2001). As a result, this study performed the Pearson product moment correlation between the

60 monthly returns of mutual funds and their standard deviations in finding the correlation

between the risk and return. However, this study investigated three more frequencies of data

to cover wider scenarios. It cross-sectionally correlated the yearly, three-year, and ten-year

returns and standard deviations of sample funds to minimize the specific effect that resulted

from any individual year shock.

4.3 Measurement of Mutual Fund Performance

There are several performance measurements used in evaluating funds. Each

method has its own strength and is suitable in different contexts. The return-based

measurement such as the Treynor ratio or Sharpe ratio measures return per unit of risk. It is

easy to calculate but does not tell the actual level of risk to which the fund is exposed. The


17

regression-based performance measurement suffers from the choice of the benchmark (Roll,

1978), such as alpha. The holding-based measurement such as the Grinbatt-Tittman

measurement is good for the manager’s evaluation but overlooks the portfolio risk-adjusted

return (Reilly & Brown, 2011). The Characteristic Selectivity (CS) by Daniel et al. (1997)

measures the selection ability of the manager but requires a great deal of information about

portfolio holdings during each period, which is sometimes a difficult task for the general

investor. The percentile ranking of return is easy to understand but ignores the risk of the

mutual fund.

4.3.1 Methodology for New Performance Measurement

Led by the above constraints, this study proposed a new performance measurement,

which takes into account both level of risk and level of return. It is easy to use and, to a

certain level, reveals the ability of the fund manager. While some performance evaluations

require specific information and calculations that are achievable only by the financial analyst or

people in the field, the new performance tool in this study requires information that is accessible

by the general investor. This performance measurement is the 3x3 boxes of risk dimension

and return dimension. This study called it the “Risk-Return Matrix”. The procedure is to

independently rank the return and risk of funds and, then, to combine the characteristics of the

two dimensions.

The Return Ranking: In each year the return of all mutual funds were sorted in

ascending order from smallest to largest. This study calculated the position of the 30th and 70th

percentile using the following calculation (Anderson, Sweeney, & Williams, 1990).

𝑃𝑘 = (𝑘

100) ∗ 𝑛 (5)

𝑃𝑘 = The position in the sample set that represent the 𝑘𝑡ℎ percentile

k = The percentile point

n = Total number of observations


18

The position 𝑃30 and 𝑃70 were then marked for the next step. The funds that had

return ranked in the first 30th percentile were marked as “low return,” those in between the 30th

and 70th were marked as “medium return,” and those in the 70th percentile upward were marked

as the “high return” group.

The Risk Ranking: This study independently ranked funds from their standard

deviation level. The procedure in risk ranking was similar to the return ranking, except that it

used the level of risk in the order sorting. The result generated three other groups of funds: the

“low-risk,” the “medium-risk.” and the “high-risk” groups.

The Risk Return Matrix: The study developed a risk-return matrix based on the two

dimensions: risk and return. The product of the two dimensions was the nine groups of funds:

the low-risk, high-return group, the low-risk, medium-return group, the low-risk, low-return group,

the medium-risk, high-return group, the medium-risk, medium-return group, the medium-risk,

low-return group, the high-risk, high-return group, the high-risk, medium-return group, and the

high-risk, low-return group. Apparently, it is quite confusing and difficult to refer to these groups

in their full characteristics. For convenience, this study named these nine groups as shown in

the italic-bolded characters in the risk-return matrix in Figure 2.

Figure 2: The Risk-Return Matrix

Risk Return

Low-risk (≤30th)

Medium-risk (>30th - <70th)

High-risk (≥70th)

High-return (≥70th)

Low-risk, High-return Sweet Spot

Med-risk, High-return Umami-2

High-risk, High-return Hot

Medium-return (>30th - <70th)

Low-risk, Med-return Umami-1

Med-risk, Med-return Salty

High-risk, Med-return Sour-2

Low-return (≤ 30th)

Low-risk, Low- return Bland

Med-risk, Low- return Sour-1

High-risk, Low-return Bitter Spot


According to this study, the best type of mutual fund performance is the “Sweet

Spot” in the upper left box, since it carries the lowest risk while delivers the highest return


19

compared to its peers. It is clear according to the definition that the “Sweet Spot” is the

preferable performance. This study was interested in examining the characteristic of the

“Sweet Spot” funds. It used the risk-return matrix for the performance evaluation and assigned

the dependent variable as a binary variable. In each calendar year, the funds that were placed

in the “Sweet Spot” group were assigned the value of one. The rest were assigned the value

of zero.

4.4 The Characteristics of the “Sweet Spot” funds

The Estimation of Idiosyncratic Risk (ѱ) : Following Ang et al. (2008), Fu

(2008), and Goyal and Santa-Clara (2003), this study computed the idiosyncratic volatility of

each mutual fund using Fama and French’s (1993) three-factor regression model. The

regression equation is as follows:

𝑟𝑛𝑗

= 𝛼𝑇𝑗

+ 𝛽𝑇𝑗(𝑟𝑚𝑘𝑡,𝑛 − 𝑟𝑓,𝑛) + 𝑠𝑇

𝑗𝑆𝑀𝐵 + ℎ𝑇

𝑗𝐻𝑀𝐿+ 𝜀𝑛

𝑗 (6)

𝑟𝑚𝑘𝑡,𝑛 = The market return at month n

𝑆𝑀𝐵 = The size premium measured by the return of small cap stocks minus the return of big cap stocks

𝐻𝑀𝐿 = The value premium measured by the returns of high book-to-market value stocks minus the returns of low book-to-market value stocks

ε = Regression residual

The measurement of the idiosyncratic risk of funds was the volatility of the regression

residuals (Goyal & Santa-Clara, 2003). The regression residuals were obtained from

regressing the equation (6) using historical 36-month data. The idiosyncratic volatility was a

standard deviation of regression residuals, 𝜀𝑗,𝑇. The calculation can be seen in equation (7)

and was annualized using equation (7.1).

𝜓𝑛𝑗

= √1

𝑛∑ 𝜀𝑗,𝑛

2𝑛

𝑗=1 (7)

𝜓𝑇𝑗

= 𝜓𝑛𝑗 * √12 (7.1)


20

𝜓𝑛𝑗 = The monthly idiosyncratic risk of fund j during the period T

𝜓𝑇𝑗 = The annualized idiosyncratic risk of fund j at time T

𝜀𝑗,𝑛 = Regression residual of fund j at month n

Fund Size (Log Size): The transformation of data into a logarithmic scale is

useful when we have great variations in the value of the data. The sample fund in this study

had sizes averaging from 4 million baht to 6,205 million baht. The author believed that using

the logarithms in the calculation of this variable was appropriate in order to minimize the great

variation in the data. This study used the logarithm of the fund size at the end of year T of the

sample fund as the fund-size variable as follows.

𝐿𝑜𝑔𝑆𝑖𝑧𝑒𝑇𝑗 = log (𝑁𝐴𝑉𝑇

𝑗 ) (8)

𝑁𝐴𝑉𝑡𝑗 = Net asset value of fund j at the end of year T

Fund Objective (DObj): This study used the dummy variable technique as a

measurement of fund objective (DObj) and used a Morningstar style box as the basis for

categorizing fund objectives (Morningstar, 2014). The assignment of the dummy variable for

the fund objectives was as follows.

The fund objective variable, DObj, takes a value of one if the fund is invested in the “growth

stock” and zero otherwise.

Type of Parent Company (DType-parent): The type of parent company

was a dummy variable. Following Nathapan (2010), this study divided the parent company of

asset management firms into two groups: bank-related and non-bank-related. Out of 83

funds, bank-related asset management companies managed 48 funds and non-bank-related

ones managed 35 funds. The study obtained this information from the websites of asset

management companies, or annual reports. The assignment of the dummy variables for type

of parent company was as follows.

The type of parent company variable, DType-parent, takes a value of one if the asset

management company is a subsidiary of a bank and zero otherwise.


21

Fund Age (Age): The fund age variable was measured in each year. The

counting in a full month began from the inception date of the funds to the observed month.

This study divided it by 12 in order to transform the unit into a year.

𝐴𝑔𝑒𝑇𝑗 =

(𝑁𝑜.𝑀𝑡ℎ𝑠)

12 (9)

𝐴𝑔𝑒𝑇𝑗 = Age of Fund j at time T

𝑁𝑜. 𝑀𝑡ℎ𝑠 = Number of months from the inception to the observed month

Total Number of Funds (No.Fund): This study obtained this variable by

using the number of all funds under supervision of an asset management company published

at the end of the year from www.aimc.or.th (AIMC, 2012).

Total Assets Under Management (Log AUM ): The Log AUM variable

was the logarithm of the total value of all assets under the management of the asset

management company. This study used the total asset size as at the end of the T period.

Log 𝐴𝑈𝑀𝑘,𝑇 = log (𝐴𝑈𝑀𝑇𝑘) (10)

𝐴𝑈𝑀𝑡𝑘 = Total assets under the management of the company k at the end of year T

The next section discusses the regression methodology used to test the second to the

eighth hypothesis, which were aimed to answer the second research question.

4.5 The Regression Model

This study aimed with its second objective to ascertain the characteristics of the

low-risk, high-return funds or the sweet spot funds. The methodology was set to examine

hypotheses two to eight. The dependent variable was the binary variable of fund performance

as explained in the section 4.3. The independent variables were the organizational factors and

characteristics of the mutual funds in section 4.4. The regression was set as the following:

Pr(𝑆𝑆𝑇𝑗

= 1) = 𝐿𝑜𝑔𝑖𝑡 (𝜇0 + 𝜇1 𝜓𝑇𝑗+𝜇2 𝐿𝑜𝑔 𝑆𝑖𝑧𝑒𝑇

𝑗+ 𝜇3 𝐷𝑂𝑏𝑗𝑗 + 𝜇4 𝐷𝑇𝑦𝑝𝑒 −

𝑝𝑎𝑟𝑒𝑛𝑡𝑘 + 𝜇5 𝐴𝑔𝑒𝑇𝑗

+ 𝜇6 𝑁𝑜. 𝐹𝑢𝑛𝑑𝑘,𝑇 + 𝜇7 𝐿𝑜𝑔𝐴𝑈𝑀𝑘,𝑇 + 𝜀𝑇𝑗

(11)

http://www.aimc.or.th/


22

𝑆𝑆𝑇𝑗 = The binary variable represents the performance of mutual

fund j at time T. It takes a value of one if the fund delivers a

low-risk, high-return performance and zero otherwise.

𝜓𝑇𝑗 = Idiosyncratic risk of fund j at time T

𝐿𝑜𝑔 𝑆𝑖𝑧𝑒,𝑇𝑗 = Log of size of fund j at the period T

𝐷𝑂𝑏𝑗𝑗 = Dummy variable for fund objective of fund j

𝐷𝑇𝑦𝑝𝑒 − 𝑝𝑎𝑟𝑒𝑛𝑡𝑘 = Dummy variable for the type of the parent company of the

asset management company k

𝐴𝑔𝑒𝑇𝑗 = Age of fund j at time T

𝑁𝑜. 𝐹𝑢𝑛𝑑𝑘,𝑇 = Total number of funds under management of an asset

management company k at time T

𝐿𝑜𝑔 𝐴𝑈𝑀𝑘,𝑇 = Log of the total assets under management of an asset

management company k at time T

𝜀𝑗,𝑡 = Regression residual

Since the sample funds in each year were not equal, ranging from 64 to 83 funds,

this study tested the hypotheses by performing the logistic regression of equation (11) using the

unbalanced panel data technique.

4.6 Enhancing Portfolio Performance with New Investment Strategies

The third objective of this study was to find out whether investors can gain some

benefit from this unorthodox relationship in their investments.

4.6.1 Low-Risk Strategy vs. High-Risk Strategy

This section tests whether the strategy of investing in a low-risk mutual fund delivers

returns comparable to the strategy of investing in a high-risk mutual fund, regardless of the

market condition. This section simulates two portfolios with different strategies. One was the

low-risk fund portfolio and the other one was a portfolio consisting of high-risk funds.


23

Each year the funds were ranked based on their risk. The funds ranked from the

30th percentile downward were selected to invest in the next calendar year. These funds were

classified as low-risk funds and were selected to be in the low-risk portfolio. The high-risk

portfolio was constructed in the similar manner except that the funds selected were the funds

where the risk was ranked from the 70th percentile upward. Then, this study calculated the

return of this low-risk portfolio and compared it with the high-risk portfolio using the following

equations.

Equation (12) is the calculation for the average return of the portfolio consisting of

low-standard deviation funds, and equation (13) is the calculation for the average return of the

portfolio consisting of high-standard deviation funds.

𝑃 𝑟𝑡𝐿𝑆𝐹 =

1

𝑛∑ 𝑟𝑖𝑡

𝐿𝑆𝑛𝑖=1 (12)

𝑃 𝑟𝑡𝐻𝑆𝐹=

1


𝐻𝑆𝑛𝑖=1 (13)

𝑃 𝑟𝑡𝐿𝑆𝐹= The average return of the portfolio consists of low-standard deviation funds.

𝑟𝑖𝑡𝐿𝑆 = Return in year t of fund i that is a member of Set LS, and Set LS = Set of funds

which have standard deviation in t-1, fall into the zero to 30th percentile.

𝑃 𝑟𝑡𝐻𝑆𝐹= The average return of the portfolio consists of high-standard deviation funds.

𝑟𝑖𝑡𝐻𝑆 = Return in year t of fund i that is a member of Set HS, and Set HS = Set of funds

which have standard deviation in t-1, fall into the 70th to 100th percentile.

Apart from the use of the total volatility of the returns as risk proxy in equation (12)

and (13), this study used beta as another proxy for risk in in equation (12.1) and (13.1), since

the CAPM concerns only the systematic risk of assets.

𝑃 𝑟𝑡𝐿𝐵𝐹 =

1


𝐿𝐵𝑛𝑖=1 (12.1)

𝑃 𝑟𝑡𝐻𝐵𝐹=

1


𝐻𝐵𝑛𝑖=1 (13.1)

𝑃 𝑟𝑡𝐿𝐵𝐹= The average return of the portfolio consists of low-beta funds.

𝑟𝑖𝑡𝐿𝐵 = Return in year t of fund i that is a member of Set LB, and Set LB = Set of funds

which have beta in t-1, fall into the zero to 30th percentile.


24

𝑃 𝑟𝑡𝐻𝐵𝐹= The average return of the portfolio consists of high-beta funds.

𝑟𝑖𝑡𝐻𝐵 = Return in year t of fund i that is a member of Set HB, and Set HB = Set of funds

which have beta in t-1, fall into the 70th to 100th percentile.

This study calculated and compared the 𝑃 𝑟𝑡𝐿𝑆𝐹and 𝑃 𝑟𝑡

𝐻𝑆𝐹 , as well as the 𝑃 𝑟𝑡𝐿𝐵𝐹

and 𝑃 𝑟𝑡𝐻𝐵𝐹 to verify whether this proposed investment enhanced the investment performance.

4.6.2 Investing in the “Sweet Spot” Perf ormers Strategy

This simulation followed the conclusion of Brown and Goetzmann (1995)’s study that

there was a performance persistence in the mutual fund industry (Brown & Goetzmann, 1995).

One easy way is to invest in past-winner mutual funds. This study proposed a simple way to

find out the winner funds using the feature of the risk-return matrix. The number of times each

fund appeared in the “Sweet Spot” group in the matrix was tallied during 2005-2011. The funds

that could make their way up to the ‘Sweet Spot” group the most often were selected as the

winners for this study. These funds were equally invested in the 2012-2014 simulation portfolio.

In order to evaluate the effectiveness of this strategy, the average return of this winner portfolio

was compared to the average industry return.

4.6.3 The strategy to invest in Funds with Characteristics of the

“Sweet Spot” Fund

This section used the results from the regression of this study. It examined whether

the characteristics, which showed a statistically-significant effect on the probability that the

funds will deliver a “Sweet Spot” performance, would help investors create a portfolio that would

yield returns higher than the industry average.

This study constructed portfolio S, which was the set of funds that had the following

conditions:

S = {𝐹𝑢𝑛𝑑𝑠| 𝑥𝑛 > 𝑥𝑛, 𝑖𝑓 𝜇𝑛 > 0, 𝑥𝑛 < 𝑥𝑛, 𝑖𝑓 𝜇𝑛 < 0, } (14)


25

𝑥𝑛 = Coefficient of the significant explanatory variables

𝜇𝑛 = Regression coefficient from equation (11)

𝑥𝑛 = Industry average value of variable 𝑥𝑛

Set S was the set of funds that had a value of 𝑥𝑛 higher than the industry average

value if its regression coefficient 𝜇𝑛, was significantly higher than zero. It also consisted of

funds that had a value of 𝑥𝑛 lower than the average industry value if its regression coefficient

𝜇𝑛 was significantly lower than zero. In order to determine the efficiency of this strategy, the

study compared the return of portfolio S with the industry average return.


Chapter 5 Statistical Results

5.1 Descriptive Data on the Sample Funds

Table 3: Descriptive Data of Sample Funds

Descr ipt ion 2005 2006 2007 2008 2009 2010 2011 2012

Average

(1) Number of Sample Funds

64 70 78 80 80 83 83 83 77.6

(2) Average fund s ize (mi l l ion)

564 478 602 347 465 545 483 718 525

(3) Max return 25.61% 8.33% 42.36% -37.11% 76.88% 59.69% 13.72% 71.06% 32.57%

(4) Min return 1.25% -10.76% 7.70% -50.24% 24.67% 20.31% -16.86% 20.57% -0 .42%

(5) Max-Min 24.36% 19.09 % 34.66% 13.13% 52.21% 39.38% 30.58% 50.49% 32.99% (6) Sample mean return 6.32% -5 .21 % 33.16% -43.45% 57.70% 41.62% -3 .38% 36.64% 15.43%

(7) STD 4.00% 4.44 % 6.44% 3.19% 9.91% 8.89% 6.99% 9.17% 6.62% (8 )Total stock market return 10.50% -0 .73 % 30.02% -42.92% 68.08% 44.13% 2.94% 39.21% 18.90%

:Developed for this study

5.2 The Relationship between Risk and Return of Thai Equity Mutual Funds

In order to meet the first research objective of this study, this section tested the first

hypothesis by examining the relationship between risk and return of the Thai open-end equity

mutual funds as shown in Table 4 and 5.

Table 4 shows two yearly results, the return correlated with the standard deviations

and the returns correlated with the beta. Table 5 shows the three-year, five-year, and ten-year

correlations between the returns and standard deviations.


27

Table 4: The Correlation between the Risks and Returns of the Thai Open-End Equity Mutual

Fund Industry from 2005-2012

Year 2005 2006 2007 2008 2009 2010 2011 2012

(1) Samples 64 70 78 80 80 83 83 83

(2) 𝜌𝐵,𝑟 – 𝑟𝑌

𝑗and Beta .058 -.306*** .817*** -.515*** .553*** .481*** -.718*** -.745***

(3) 𝜌𝑆,𝑟 - 𝑟𝑌𝑗and

Annualized STD -.428*** -.326*** .794*** -.613*** .637*** .256** -.671*** -.337***

(4)Total market return 10.50% -0.73% 30.02% -42.92% 68.08% 44.13% 2.94% 39.21%

(5) Average industry 6.32% -5.21% 33.16% -43.45% 57.70% 41.62% -3.38% 36.65%

*** Correlation is significant at the 0.01 level ** Correlation is significant at the 0.05 level

For convenience in using Table 4, the negative correlation numbers are bolded.

The table reveals the significant negative correlations in 2005, 2006, 2008, 2011, and 2012,

and the significant positive correlations in 2007, 2009, and 2010. Though a negative correlation

was not found in every period during 2005-2012, the Thai open-end equity mutual fund industry

did post more negative-correlation years than positive ones

The negative results in Table 4 could have accidentally happened due to some

economic shocks in that year. In order to reduce the effect of that shock, this study expanded

the correlation period into three, five, and ten years. Table 5 shows the results.

Both results in Table 4 and 5 documented that the open-end equity mutual fund

industry in Thailand occasionally posted a negative relationship between risk and return in

some individual years, as shown in Table 4, and some three-year and five-year intervals during

2003-2012, as shown in Panel A and B, Table 5. It also posts a significant negative

relationship in the annualized ten-year cumulative investment in Panel C, Table 5.


28

Table 5: The Correlations between the Three-, Five-, and Ten-Year Annualized Returns and

Standard Deviations

Panel A (3Y) 2003-05 2004-06 2005-07 2006-08 2007-09 2008-10 2009-11 2010-12 (1) 𝜌𝑠,𝑟 - 𝑟3𝑌

𝑗 and 𝑆3

𝑗

.188 64

-.409*** 64

-.054 64

-.541*** 70

.052 78

.223** 80

.256** 80

-.503*** 83

Panel B (5Y) 2003-07 2004-08 2005-09 2006-10 2007-11 2008-12 - -

(2) 𝜌𝑠,𝑟 - 𝑟5𝑌

𝑗 and 𝑆5

𝑗

0.15 64

-.482*** 64

-.260** 64

.259** 70

0.151 78

-.402*** 80

- -

Panel C (10 Y) 2003-12 - - - - - - -

(3) 𝜌𝑠,𝑟 - 𝑟10𝑌

𝑗 and 𝑆10

𝑗

-.299** 64

- - - - - - -

Note: The italic numbers are the number of samples

*** Correlation is significant at the 0.01 level ** Correlation is significant at the 0.05 level

The results from Table 4 and 5 led the author to reject the null hypothesis 𝐻0 1,

which stated that there is a positive or no relationship between risk and return in Thai open-end

equity mutual funds industry. This result agreed with the conclusion of Brockett et al. (1992)

and Cooper et al. (2011), that there could be a negative relationship between risks and returns

the mutual funds industry.

5.3 Fund Performance as Dependent Variable

Using the risk-return matrix methodology as discussed in Chapter 4, for every year,

this study classified the fund performance into nine groups. Table 6 distributes the samples

into these nine groups.

Table 6 shows that during the study period, the industry had 71 fund-years with

“Sweet Spot” performance and 550 fund-years with other types of performance.


29

Table 6: The Distribution of Fund Performance during 2005-2012

For each year, the samples are categorized into nine types of fund according to their characteristic of the level of risk and return. The numbers in the column under each year are the distribution of the sample funds in those years into the nine types of fund. The total number of samples in each year is presented in the last row. The total numbers in the far right column are a summation of each type of fund from 2005 to 2012. This study had 621 fund-year samples.

Risk Return 2005 2006 2007 2008 2009 2010 2011 2012 Total Performance

Low Risk

High 10 5 7 13 0 9 17 10 71 Sweet Spot

Medium 8 7 3 9 9 8 7 14 65 Umami-1

Low 2 9 14 2 15 8 1 1 52 Bland

Medium Risk

High 3 7 4 7 13 2 5 12 53 Umami-2

Medium 7 10 18 23 11 17 22 16 124 Salty

Low 14 11 8 2 8 14 6 5 68 Sour-1

High Risk

High 7 9 13 4 11 14 3 3 64 Hot

Medium 9 11 9 0 12 8 4 3 56 Sour-2

Low 4 1 2 20 1 3 18 19 68 Bitter

Total 64 70 78 80 80 83 83 83 621

Table 7: The Risk-Return Matrix for 2005-2012 The table combines the sample distribution funds during 2005-2012 into nine types of fund performance. The italic bolded numbers are the number of funds that fell into that category. The average return of each group is shown in percentage number. The average standard deviation is shown in parentheses

Risk

Return Low-Risk Medium-Risk High-Risk

High Return “Sweet Spot” = 71 25.46% (15.97%)

“Umami-2” = 53 25.59% (17.56%)

“Hot” = 64 25.77% (18.68%)

Medium Return “Umami-1” = 65 17.66% (16.21%)

“Salty “ = 124 17.53% (17.55%)

“Sour-2” = 56 17.55% (19.05%)

Low Return “Bland” = 52 11.39% (16.52%)

“Sour-1” = 68 11.92% (17.62%)

“Bitter Spot” = 68 11.66% (18.74%)


30

Table 7 shows that out of the total sample of 621 observations during 2005-2012, 71

observations exhibited a “Sweet Spot” performance with average return of 25.46% and standard

deviation of 15.97%.

Table 7 also added other evidence—that Thai open-end equity mutual fund did not

have a positive relationship during 2005-2012. The 621 observations are scattered over the

nine boxes rather than clustered around the “Bland,” “Salty,” and “Hot” boxes. If we expect a

high-risk, high-return relationship, the observations should form to demonstrate an upward slope

from low-risk, low-return, to medium-risk, medium-return, and high-risk, high-return. Table 7

does not reveal such a pattern.

In the following diagram, this study demonstrates and discusses the risk and return of

each group.

Figure 3: The Average Performance of Each Group during 2005-2012

High Return Medium Return Low Return Risk


Focusing on the first three bars, the “Sweet Spot”, the “Umami-1”, and the “Bland”

represent low risk groups since their risks are within the first 30th percentile. Within this group,

Sweetspot

Umami-1

BlandUmam

i-2Salty Sour-1 Hot Sour-2

BitterSpot

Return 25.46% 17.66% 11.39% 25.59% 17.53% 11.92% 25.77% 17.55% 11.66%

Risk 15.97% 16.21% 16.52% 17.56% 17.55% 17.62% 18.68% 19.05% 18.74%

Low Risk Medium Risk High Risk

14.00%

14.50%

15.00%

15.50%

16.00%

16.50%

17.00%

17.50%

18.00%

18.50%

19.00%

19.50%

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%Average Performance of Each Group


31

the “Sweet Spot” has the lowest risk of all at 15.97% versus 16.21 % and 16.52 %, despite its

highest return at 25.46 % versus 17.66 % and 11.39 %. This information emphasizes the

outstanding performance of the “Sweet Spot’ group when compared to peers.

5.4 Logistic Regresssion

In order to test the second hypothesis to the eighth hypothesis for the second

research objective, which aimed to find out the characteristics of the “Sweet Spot” fund, this

study performed a logistic regression using unbalance panel data analysis.

5.4.1 The Hausman Test

In the process of performing the Hausman test to decide whether to use the fixed-

effect model or the random-effect model, this study found that the fixed-effect model dropped

370 observations. A possible explanation for this occurrence is multicollinearity. The model in

this study contained two dummy variables, the DObj and the DType-parent. The mechanism of

the fixed-effect model was to create dummy variables for those fixed effects. This was done in

order to control for the differences across groups; hence, the model relied on a reasonable

amount of within-group variation. Having a dummy variable as a predictor reduces the within-

group variation.

Due to this significant limitation in the data for the fixed-effect model, this study used

the random-effect model in performing the logistic regression.

5.4.2 The Regresssion

The statistical results from the logistic regression using the random-effect model are

shown in Table 8. Though it is more applicable to use the random-effect model than the fixed-

effect model, this study shows the result of both models in Table 8 for comparison purposes.


32

Table 8: The Logistic Regression of the Characteristics of the Low-Risk, High-Return Fund

𝑃𝑟(𝑆𝑆𝑇𝑗

= 1) = 𝐿𝑜𝑔𝑖𝑡(𝜇0 + 𝜇1 𝜓𝑇𝑗

+ 𝜇2 𝐿𝑜𝑔 𝑆𝑖𝑧𝑒𝑇𝑗

+ 𝜇3 𝐷𝑂𝑏𝑗𝑗 + 𝜇4 𝐷𝑇𝑦𝑝𝑒 − 𝑝𝑎𝑟𝑒𝑛𝑡𝑘 +

𝜇5 𝐴𝑔𝑒𝑇𝑗

+ 𝜇6 𝑁𝑜. 𝐹𝑢𝑛𝑑𝑘,𝑇 + 𝜇7 𝐿𝑜𝑔 𝐴𝑈𝑀𝑘,𝑇 + 𝜀𝑇𝑗 )

Variables Model (1):Fixed-effect Model (2):Random-effect

Coef. z P>|z| Coef. z P>|z| dy/dx

𝝍 .299 1.97 0.048 ** .419 3.71 0.000*** .0262

𝑳𝒐𝒈 𝑺𝒊𝒛𝒆 -.244 -0.80 0.425 -.042 -0.26 0.798 -.0026

𝑫𝑶𝒃𝒋 (omitted) - - -.045 -0.10 0.921 -.0028

𝑫𝑻𝒚𝒑𝒆 − 𝒑𝒂𝒓𝒆𝒏𝒕 (omitted) - - -.239 -0.48 0.634 -.0153

Age .236 2.12 0.034** .101 2.54 0.011** .0063

𝑵𝒐. 𝑭𝒖𝒏𝒅 .007 0.64 0.524 -.014 -2.07 0.038** -.0009

𝑳𝒐𝒈 𝑨𝑼𝑴 -1.802 -2.42 0.016** .294 0.92 0.356 .0184

Constant - - - -11.083 -1.35 0.177

No. obs 251 621

Chi2(df) 14.04(5) 25.13(7)

Prob>Chi2 0.0154** 0.0007***

*** = significance at the 0.01 level.

** = significance at the 0.05 level

From Table 8, the idiosyncratic risk of fund, 𝜓, shows its significant effect at the 99

% confidence level on fund performance, as its p-value =0.000. Fund age, (Age), and the total

number of funds, (No.Fund,) are significant at the 95 % confidence level with p-values = 0.011

and 0.038, respectively. Other variables do not significantly affect the probability that a fund

will deliver sweet spot performance.

The coefficient of 0.419 of the idiosyncratic risk (𝜓) indicates that, holding other

variables at fixed value, the fund with higher idiosyncratic risk is more likely to deliver “Sweet

Spot” performance than a fund with lower idiosyncratic risk. The dy/dx value of 0.0262

indicates the marginal effect of the variable. The one unit increases in idiosyncratic risk


33

significantly increases the probability that that the fund will deliver “Sweet Spot” performance by

2.62 %.

The Age coefficient of 0.101 indicates that holding the other variables at fixed value,

an older fund is more likely to deliver “Sweet Spot” performance than a younger fund. The

dy/dx value of 0.0063 indicates that, as a fund gets older by one year, the probability that the

fund will deliver “Sweet Spot” performance will increase by 0.63%.

The No.Fund coefficient of -0.014 indicates that holding the other variables at fixed

value, the fund under the supervision of an asset management company that manages a large

number of funds is less likely to deliver “Sweet Spot” performance compared to funds under the

supervision of an asset management company that manages a smaller number of funds. The

dy/dx value of -0.0009 indicates that if the company increases the number of funds under its

management by one, it will decrease the probability that the fund under its supervision will

achieve “Sweet Spot” performance by 0.09%.

Table 8 shows the probability of getting greater chi-square is significant at 0.0007. It

leads this study to conclude that the overall model is statistically significant.

From the regression results, this study concluded that the idiosyncratic risk of funds

and the age of funds have significant positive effects on the probability that a fund will deliver

low risk-high return performance. The total number of funds under the management of an

asset management company has a significant negative effect on the probability that a fund will

deliver “Sweet Spot”, low risk-high return performance.

5.5 Summary of the Results of the Hypothesis Testing

Table 9 summarizes all of the hypotheses of this study and their findings.


34

Table 9: The Summary of the Results of the Hypothesis Testing

The Hypotheses Findings

𝐻01: There is a significant positive or no relationship between risk and return in the Thai open-end equity mutual funds industry.

Rejected the null hypothesis and found a significant negative relationship from time to time

𝐻02: The idiosyncratic risk of a fund has no significant effect on the probability that a Thai open-end equity mutual fund will deliver a low-risk, high-return performance.

Rejected the null hypothesis and found a significant positive relationship

𝐻03: Fund size has no significant effect on the probability that a Thai open-end equity mutual fund will deliver a low-risk, high-return performance.

Failed to reject the null hypothesis

𝐻04 : Fund objective has no significant effect on the probability that a Thai open-end equity mutual fund will deliver a low-risk, high-return performance.


𝐻05: The type of parent company of an asset management company has no significant effect on the probability that a Thai open-end equity mutual fund will deliver a low-risk, high-return performance.


𝐻06 : Fund age has no significant effect on the probability that a Thai open-end equity mutual fund will deliver a low-risk, high-return performance.

Rejected the null hypothesis and found a significant positive relationship

𝐻07: The number of funds under the management of an asset management company has no significant effect on the probability that a Thai open-end equity mutual fund will deliver a low-risk, high-return performance.

Rejected the null hypothesis and found a significant negative relationship

𝐻08: The total assets under the management of an asset management company have no significant effect on the probability that a Thai open-end equity mutual fund will deliver a low-risk, high-return performance.



35

5.6 Alternative investment Strategies

The statistical tests in this section are provided in order to meet the third objective

of this study. They were to find out whether investors can capitalize on this unorthodox

relationship and apply the results of this study to enhance their portfolio return. This study

offered three simulations below.

5.6.1 Low-Risk Strategy vs High-Risk Strategy

Two types of risk were tested.

A. The low-standard deviation strategy vs the high standard deviation

strategy: Table 10 shows that in Panel A, B, and C , the simulation results indicate the

superior returns of the low standard deviation strategies compared to the high standard

deviation strategies in 2006, 2008, 2011, and 2012, and the inferior performance in 2007, 2009,

and 2010. However, when looking at the total period, the arithmetic mean of all seven years

from 2006-2012 indicates the superior performance of the low-standard deviation strategies

over the high- standard deviation strategies in all three panels.

Table 10: The Portfolio of the Low-Standard Deviation Funds vs. the High- Standard Deviation Funds

𝑃 𝑟𝑡𝐿𝑆𝐹 =

1


𝐿𝑆𝑛𝑖=1 (12)

𝑃 𝑟𝑡𝐻𝑆𝐹 =

1


𝐻𝑆𝑛𝑖=1 (13)

STD is the standard deviation of fund returns. In Panel A, the equal-weighted return of low STD funds, in row

(1), was computed using equation (12), where 𝑟𝑖𝑡𝐿𝑠 was the return of fund i in Set LS, which is the set of funds

that have risk in t-1 falling into 0th - 30th, when ranked by the standard deviation. The number in the parentheses

under the average return of each year is the annualized standard deviation, and the italic number represents

the number of funds selected. Row (2) Panel A reports an equal-weighted return of the high risk portfolio. It

was computed using equation (13), where 𝑟𝑖𝑡𝐻𝑠 was the return of fund i in Set HS, which was the set of funds

that had risk in t-1 falling into 70th - 100th when ranked by the standard deviation. Row (3) is the difference

between the two strategies. Panel B follows the same demonstrations process, but changes the percentile cut-

off to 0th - 15th and 85th - 100th. Panel C is the return of the portfolio of the ten lowest and ten highest STD funds.

Panel D shows the industry average return in row (10) and the annualized total market stock return in row (11).


36

Return/Risk 2006 2007 2008 2009 2010 2011 2012 Average Cumulative

Return

A. Low VS High STD Strategy at 30th percentile and 70th percentile

(1) Equal-weighted return of low STD (30th) LS-30

-2.08%

(16.36%)

20

33.38%

(19.41%)

21

-40.21%

(33.27% )

24

53.51%

(17.93%)

24

37.31%

(16.43%)

24

2.77%

(19.59%)

25

45.81%

(12.67%)

25

18.64%

(19.38%)

147%

(2) Equal-weighted return of high STD (70th) HS-30

-5.73%

(17.18%)

20

36.69%

(19.81%)

21

-46.08%

(36.79%)

24

60.22%

(22.60%)

24

47.67%

(17.94%)

24

-3.91%

(22.38%)

25

28.69%

(13.48%)

25

16.79%

(21.46%)

103%

(3) = (1) - (2)

3.66%

-3.31%

5.87%

-6.71%

-10.35%

6.67%

17.12%

1.85%

43.4%

B. Low vs. High STD Strategy at 15th percentile and 85th percentile

(4) Equal-weighted return of low STD (15th) LS-15

-0.64% (15.97%)

10

33.92% (19.07%)

11

-40.30%

(33.42%)

12

47.79%

(16.86%)

12

36.11%

(15.53%)

12

6.97% (18.84%)

13

50.94%

(13.20%)

13

19.26%

(18.98%)

158%

(5) Equal-weighted return of high STD (85th) HS-15

-4.86% (17.17%)

10

34.78% (19.32%)

11

-47.15%

(37.90%)

12

61.17 %

(21.89%)

12

44.35%

(17.86%)

12

-6.26% (23.71%)

13

27.60%

(13.46%)

13

15.66%

(21.61%)

88.58%

(6) = (4) - (5) 4.22% -0.85% 6.85% -13.38% -8.24% 13.22% 23.34% 3.60% 69.44%

C. Ten Lowest vs Ten Highest STD Funds Strategy (7) Equal-weighted return of low STD (Lowest 10) LS-10

-0.64% (15.97%)

10

33.85% (18.88%)

10

-40.67% (33.55%)

10

47.12% (16.89%)

10

35.95% (15.05%)

10

7.87% (18.68%)

10

52.32% (13.21%)

10

19.40%

(18.89%)

159%

(8) Equal-weighted return of high STD (Top 10) HS-10

-4.86% (17.17%)

10

34.20% (19.23%)

10

-47.29% (38.09%)

10

59.59% (21.43%)

10

43.94% (18.03%)

10

-5.71% (13.45%)

10

28.11% (13.29%)

10

15.43% (21.53%)

86.74%

(9) = (7) –(8) 4.22% -0.35% 6.62% -12.47% -7.99% 13.58% 24.21% 3.97% 72.57%

D. Mutual Fund Industry and Total Stock Market

(10) Average return of industry

-5.21% (16.59%)

70

33.16% (19.27%)

78

-43.45% (34.62%)

80

57.70% (20.18%)

80

41.62% (17.56%)

83

-3.38% (22.04%)

83

36.65% (13.26%)

83

16.73% (20.50%)

110%

(11) Total stock market return

-0.73% 30.02% -42.92% 68.08% 44.13% 2.94% 39.21% 20.10% 155%


37

The 2006-2012 cumulative return in the far right column also reveals the superiority

of the low-risk strategies over the high-risk strategies in all three panels. The cumulative return

of investing in the funds with the lowest 30th percentile risk (LS-30) in row (1) is higher than the

cumulative return of investing in the funds with the highest 70th percentile risk (HS-30) in row

(2) by 43.40%. The cumulative return of investing in the funds with the lowest 15th percentile

risk (LS-15), in row (4), is higher than the cumulative return of investing in the funds with the

highest 85th percentile risk in (HS-15), in row (5), by 69.44%. The cumulative return of

investing in the ten funds with the lowest risk (LS-10) in row (7) is higher than the cumulative

return of investing in the ten funds with the highest risk (HS-10) in row (8) by 72.57%.

In summary, Table 10 demonstrates that the three investment strategies which

invested in the low-standard deviation funds, selected from t-1, during 2006-2012, yielded a

higher return than investing in the high-standard deviation funds during the same period. Using

the information from Table 10, Figure 4 shows the scatter plots of the average return and

standard deviation of each strategy during 2006-2012.

Figure 4: The Scatter Plot between the Average Returns and Standard Deviations of the Portfolios

during 2006-2012

The markers in the graph represent the average return and average standard deviation

of the different portfolios as follows:

LS 30

HS 30

LS 15

HS 15

LS 10

HS 10

Industry Avg

15.00%

15.50%

16.00%

16.50%

17.00%

17.50%

18.00%

18.50%

19.00%

19.50%

20.00%

18.50% 19.00% 19.50% 20.00% 20.50% 21.00% 21.50% 22.00%

Ave

rage

Ret

urn

Average Standard Deviation

STD Strategy- Average Return and Standard deviation


38

LS-30 = The portfolio of the low-standard deviation funds that were ranked in the 30th

percentile and below

LS-15 = The portfolio of the low-standard deviation funds that were ranked in the 15th

percentile and below

LS-10 = The portfolio of the ten lowest standard deviation funds

HS-30 = The portfolio of the low-standard deviation funds that were ranked in the 70th

percentile and above

HS-15 = The portfolio of the low-standard deviation funds that were ranked in the 85th

percentile and above

HS-10 = The portfolio of the ten highest standard deviation funds

Figure 4 illustrates an interesting outcome with its downward pattern. The seven-

year average returns of low-standard deviation strategies (LS) are all higher than the seven-

year average returns of the high-standard deviation strategies (HS). The points LS-30, LS-15,

and LS-10 are all on the upper left of the graph, while the points HS-30, HS-15, and HS-10 are

all at the lower right of the graph. All “LS” strategies have an average return ranging from

18.34% - 19.40% with a standard deviation of 18.89% -19.38%, while the “HS” strategies have

an average return ranging from 15.44% - 16.79% with a standard deviation of 21.46% -21.61%.

It can be inferred from Figure 4 that the “LS” strategy of this study is superior to the “HS”

strategy of this study.

B. The low-beta strategy vs the high-beta strategy: The methodology of this

section was the same as the methodology in previous section, but the ranking criteria were

changed to the beta of funds. The results showed that the cumulative return of all three types

of portfolio with low-beta funds were higher than that of high-beta funds by 31%-66%. To save

space, this study will not demonstrate its empirical result here. However, the results confirmed

the superiority of low risk strategy over the high-risk strategy during the studying period.


39

5.6.2 Investing in the “ Sweet Spot” Performers

This second simulation requires some data preparation on the part of the investor.

Prior to the construction of the portfolio, investors need to do a simple analysis of the risk and

return level of funds using the Risk-return Matrix proposed by this study. Each year the matrix

offers “Sweet Spot” funds. This study tallied the frequency at which each fund fell into the

“Sweet Spot” performance during 2005-2011. The simulated portfolio would invest in these

funds which, by definition of this study, are the best performers. The results are shown in

Figure 5.

Figure 5: The Distribution of Funds with a “Sweet Spot” Performance (2005-2011)


Figure 5 demonstrates that during the seven-year period, from 2005-2011, there were

five funds that reached the “Sweet Spot” four times in seven years. Four funds were rated

“Sweet Spot” three times in seven years, and four others that were rated the “Sweet Spot” two

out of seven years. Eight funds were able to make it to this group one time, while there were

43 funds that had never reached the “Sweet Spot” performance at all.

43

8

4

4

5

0 10 20 30 40 50

0

1

2

3

4

Number of Funds

Numb

er o

f Tim

es in

"Swe

et Sp

ot"

Funds with "Sweet spot" performance during 2005-2011


40

This study tested whether investing in funds with a strong history of a “Sweet Spot”

performance would yield a favourable return. The results are shown in Table 11.

Table 11: The Return and Standard Deviation of the Portfolio of the “Sweet Spot” Winner

The table presents a 2012-2014 portfolio of five funds selected from the “Sweet spot” winner, evaluated during the period 2005-2011. The return and risk of the five selected funds are in rows (1) to (5). The average return and risk of the selected funds are in row (6) and the industry averages are in row (7). Row (8) is the difference between the two.

Fund Code

2012-2014 Return

2012-2014 Risk

(1) Fund # 4 79.43% 17.24% (2) Fund # 5 71.87% 12.43% (3) Fund # 7 73.32% 12.33% (4) Fund # 8 72.58% 12.25% (5) Fund # 11 64.46% 12.79% (6) Equally-weighted of the selected five

72.33%

13.41%

(7) Industry average 58.11% 15.16%

(8) = (6)-(7) 14.22% (1.75)%

Table 11 demonstrates the aggregated return of investment during the 2012-2014.

Rows (1) to (5) are the three-year aggregated return investment of the funds with the highest

frequency into the ‘Sweet Spot” performance during 2005-2011. The results show the superior

return of this selected portfolio compared to the industry average. The average return of these

five “Sweet Spot” winners, in row (6), is 72.33% while for the industry, in row (7), it is at

58.11%. The average standard deviation of these five “Sweet Spot” winners is 13.41% while

for the industry, in row (7), it is 15.16%. The difference in row (8) shows that, during 2012-

2014, the simulated portfolio delivered a return higher than the industry average by 14.22%,

while carried lower risk.


41

Using the risk-return matrix proposed earlier, this study offers a second alternative

investment strategy, which shows that the ability to select high frequency “Sweet Spot”

performers may enhance the portfolio return by not necessarily assuming higher risk.

5.6.3 The Portfolio Emphasizing the Characteristics of the Sweet Spot

Fund

This third strategy utilized the regression results from the regression. This section

simulates the portfolio consisting of funds with these characteristics called “Portfolio S” as

shown below.

𝑆 = {𝐹𝑢𝑛𝑑𝑠|ѱ: ѱ > ѱ, 𝐴𝑔𝑒: 𝐴𝑔𝑒 > 𝐴𝑔𝑒 , 𝑁𝑜. 𝐹𝑢𝑛𝑑: 𝑁𝑜. 𝐹𝑢𝑛𝑑 < 𝑁𝑜. 𝐹𝑢𝑛𝑑 }

This study filtered those three variables using the marginal effects, dy/dx, as their

criteria. As a result, Portfolio S consisted of funds that had an idiosyncratic risk level higher

than the industry average, were older than the industry-average fund age, and were managed

by an asset company that had a lower number of funds under management than the industry

average, at t-1. This study simulated the investment of portfolio S at time t. The results can be

seen in Table 12.

Table 12: Portfolio Emphasizing the Characteristics of the “Sweet Spot” Fund Portfolio 2006 2007 2008 2009 2010 2011 2012 Average Cumulative

return (1) Funds with ѱ higher than average

-3.70% 33.93% -42.17% 54.63% 35.44% -1.66% 36.59% 16.15% 110%

(17.06%) (18.63%) (33.60%) (18.72%) (17.37%) (22.70%) (13.59%) (20.24%)

28 18 30 32 30 26 44

(2) Funds with ѱ and Age higher than average.

-2.68% 33.60% -41.20% 50.25% 32.47% -2.06% 43.08% 16.21% 113%

(16.81%) (18.36%) (32.86%) (17.94%) (16.33%) (21.89%) (13.18%) (19.62%)

18 14 14 14 14 10 21

(3) Funds with ѱ and Age higher than average and No.Fund lower than average

-0.19% 33.60% -41.20% 48.27% 31.45% 3.04% 46.14% 17.30% 130%

(16.95%) (18.36%) (32.86%) (17.56%) (16.15%) (20.34%) (13.05%) (19.32%)

10 14 14 12 12 6 16

(4) Industry average -5.21% 33.16% -43.45% 57.70% 41.62% -3.38% 36.65% 16.73% 110%

The results in Table 12 show that the portfolio return conditioning on these

characteristics performed better than the industry average. Row (1) shows the portfolio


42

consisting of funds that had an idiosyncratic risk higher than the industry average. Its average

return was 16.15%, the average standard deviation was 20.24%, and the cumulative return was

110%. In Row (2), the portfolio added one more filter. It took out the fund with an age lower

than the industry average. This process slightly increased the portfolio return to 16.21% while

decreased its volatility to 19.62%. Row (3) shows the results after another filter, the funds that

were managed by the company with a high number of funds under management, were taken

out. The cumulative return of the selected funds in row (3) is 20 % higher than the industry

average, in row (4), at 130%-110%.

The simulation approach in section 5.6 reveals some useful results. Section 5.6.1

simulated the simplest strategy. Investors just selected and invested in the low-risk fund

compared to their peers and still got a higher return. The simulation in section 5.6.2 added a

ranking analysis of the fund performance. This strategy was trying to capitalize from the funds

suggested by risk-return matrix. The result revealed that if investors believe in a “Sweet Spot”

winner, evaluated by the matrix, and invest in these funds, they should get higher portfolio

performance than the industry average. The third experiment in section 5.6.3 used the

significant characteristics suggested by the regression results. This study screened funds at t-1

based on these characteristic criteria and invested in the subsequent year. This portfolio gave

a return higher than the industry average.

Section 5.6 offered three new investment strategies. This section was set to meet the

third objective of this study. The results from Section 5.6 found that, in the mutual fund market

where a negative relationship between risk and return exists, there could be strategies that

assist investors in managing their mutual fund investment such that they will receive a higher

return while not assuming a higher risk.


Chapter 6 Conclusion and Implications

6.1 Overview of the Study

Bowman (1980) introduced controversial results concerning the negative relationship

between risk and return in his examination of the correlation between the return on equity ratio

of a firm (ROE) and the standard deviation of ROE. The study implied that a high-return

company has lower risk. Bowman (1980) offered the Bowman paradox as another angle on

asset pricing study, which traditionally believed that a high return could be expected from the

high-risk investment. The issue brought about many subsequent studies. Those studies

offered various hypotheses to explain the paradox. Andersen (2007) concisely summarised the

direction of that literature into three groups. The first group was interested in the behaviour of

the investment decision maker and used behavioural theory, such as the prospect theory, to

explain the paradox. The second group was interested in the organizational factors that

explained the paradox through the fundamental factors of the firm. The third group suspected

that the paradox occurred from risk misspecification. Most of these studies examined the

paradox using the industrial sector as their sample rather than the financial sector, such as

banks or mutual funds. However, the two recent works of Brockett et al. (1992) and Cooper et

al. (2011) studied the paradox using the U.S. mutual fund as their sample.

This study has extended the paradox into the Asian context for various reasons.

First, the Asian market is a high-potential investment market with assets under management of

129.2 billion USD in 2013 and the Bowman paradox is an argument against the high-risk, high-

return notion of the asset pricing subject. Knowing more about the paradox in the Asian

context will help our understanding of the asset-pricing subject better. Second, the literature

on the mutual fund in the Asian context is limited, and this study will certainly be beneficial for

future investigation of the mutual fund industry in Asia. This study used the Thai market as its


44

sample because the Thai stock market is considered structured—its capitalization in 2014 had

the highest trading value within ASEAN for three consecutive years (SET, 2015).

This study is sought answers to three research questions. First, it wanted to find out

whether there was a negative relationship between risk and return in Thai open-end equity

mutual fund industry. Second was to find out which characteristics of funds would affect the

probability that the funds can deliver a low-risk, high-return performance. Third was to find out

whether the results of this study can be used to enhance the investment return.

This study set the first hypothesis under the first research question. It hypothesised

that there is a negative relationship between the risk and return of Thai open-end equity mutual

funds. This study used the Pearson product moment correlation between the return versus the

standard deviation and the return versus the beta of funds as its methodology to answer this

question. The discussion of the findings is in part (1) of section 6.2.

In order to address the second research question, this study constructed seven

hypotheses to find out the effect of seven variables on the probability that a fund will deliver a

low-risk, high-return performance. This study used unbalanced panel data logistic regression

as its methodology for this objective. A 95 percent confidence level was used to determine the

significance of the tests. The findings are discussed in part (2) of section 6.2.

In order to answer the third research question, this study simulated three types of

portfolios. The first simulation was the comparison of the return of the low-risk fund portfolio

versus the high-risk fund portfolio. The second type was the portfolio consisting of the fund

suggested by the risk-return matrix, the performance measurement proposed in this study. The

third portfolio consisted of funds filtered according to the results of the regression of this study.

The findings are discussed in part (3) of section 6.2.

6.2 Summary of the Findings

This section summarises the findings that answered the three research questions.


45

(1) Does the Thai open-end equity mutual fund industry have a negative relationship

between risk and return?

This study suggests that there could be a negative risk-return relationship in the Thai

open-end equity mutual fund industry from time to time. The Pearson product moment

correlation between risk and return during the study period of 2003-2012 showed that there was

a significant negative relationship with a correlation of -0.299 between the risk and return of the

sample funds during this ten-year period. When examining different frequencies, one-year,

three-year, and five-year periods, the results documented the presence of a negative

relationship between risk and return in the industry as follows: in the eight one-year period tests

of the correlation between the returns and standard deviations, there were five significant

negative relationships and three significant positive relationships. In the eight three-year period

tests of the correlation, there were three significant negative relationships, two significant

positive relationships, and three non-significant ones.

In the six five-year period tests of such correlation, there were three significant

negative relationships, one significant positive relationship, and two non-significant ones.

Although the results showed mix types of correlations, there were more negative-correlation

years/intervals than positive ones from all tests.

The fact that the ten-year period showed a negative relationship and the other types

showed more negative relationships than positive ones led this study to conclude that the Thai

open-end equity mutual fund industry does not always have a positive, high-risk, high-return

relationship; there can occasionally be a negative risk-return relationship in the industry.

(2) What organizational factors or characteristics of a fund have a significant effect on the

probability that a fund will deliver a low-risk, high-return performance?

This study explored seven organizational factors and fund characteristics that

affected fund performance. They were non-systematic risk, fund size, fund objective, type of

parent company, fund age, total number of funds, and the total assets under management


46

within the same asset management company. The results showed that the fund with high

idiosyncratic risk and/or an older fund was more likely to deliver a low-risk, high-return

performance. The fund that belongs to the asset management company that manages too

many funds is less likely to deliver a low-risk, high-return performance.

This study concludes that the idiosyncratic risk of a fund and the age of a fund have

positive effects on the probability that the fund will deliver a low-risk, high-return performance.

The total number of funds managed by the asset management company was seen to have an

adverse effect on the probability that funds will deliver such performance.

(3) Can the results of this study be used to enhance the investment portfolio return?

This study offers three new investment strategies as alternatives. The simulation

experiments revealed that the results of this study can be used to help investors enhance their

investment portfolio return, as discusses below.

(3.1) The first simulation portfolio showed that, under the study period, investing in

low-standard deviation mutual funds delivered a return comparable to investing in high-standard

deviation mutual funds. The results showed that investors that invest at time t, in the mutual

funds that were classified as low-standard deviation at t-1 will accumulate wealth better than

those that choose to invest in high-standard deviation funds. Such results persisted when this

study changed its risk proxy from the standard deviation to the beta.

(3.2) The use of the risk-return matrix, the performance measurement proposed by

this study, is helpful and can be a convenient tool for investors to implement. This performance

measurement is based on the percentile ranking procedure and can suggest funds that have

excellent performance during any testing period. This study simulated the portfolio which

invested in 2012-2014 in the funds that fell into the “Sweet Spot” type in the past the most

often, and the results were impressive. During 2012-2014, this “Sweet spot” fund portfolio

could deliver a return higher than the industry average by 14.22 %, while carrying lower risk.


47

(3.3) Using the results of the regression, this study constructed a portfolio

consisting of funds that had characteristics of “Sweet Spot” funds. These characteristic were

obtained from the regression results of this study. The results in Chapter 5 shows that, from

2006-2012, investing in an older-than-average fund with an idiosyncratic risk higher than

average, and that was managed by a company with a number of funds less than industry

average, yielded an average higher return and lower volatility than the industry average. This

study suggests that these three characteristic can be used as a guideline to select funds with a

high performance.

6.3 Conclusion

The study provides further empirical support that there is not always a positive risk-

return relationship in the mutual fund industry. It also provides a new performance evaluation

technique, the risk-return matrix, which is easy to prepare, uses data that are accessible by

general investors, and yet is proved to be useful in helping investors evaluate fund

performance. The study regressed the proposed model and found the characteristics of these

winner funds. The results showed that the level of specific risk of funds, fund age, and number

of funds managed by a company can indicate the probability of the winner funds. The study

also showed how new investment strategies, which do not necessarily follow the high-risk, high-

return notion, can assist investors in obtaining a higher return while being exposed to lower

risk.

6.4 Implications of the Research

Meeting the first objective, which was to find the relationship between the risk and

return of the Thai open-end equity mutual fund industry, this study documented that there are

times when the funds with high-standard deviation deliver low returns and funds with low-

standard deviation deliver high returns. The phenomenon implies that, in the Thai open-ended

equity mutual fund industry, the market information is not always efficient; some fund managers

may be able to obtain some information that other managers cannot, hence helping them to


48

make better investment decisions. Furthermore, the study also found that, from time to time,

the funds with a high beta deliver low returns and the funds with a low beta deliver high returns.

This shows that the beta of the Thai open-end equity mutual fund does not behave according to

the modern portfolio theory and that the beta can be responsible for this unorthodox

relationship. The CAPM is a single factor model which is based on the beta of the asset. The

irregular behaviour of the betas in these mutual funds also implies that a single factor model

might not be a suitable model for estimating the expected return for the Thai equity mutual

funds.

The results that satisfied the second objective also suggest the characteristics of the

Thai-equity mutual fund industry. The positive significant effect of the idiosyncratic risk of funds

on fund performance in this study emphasizes the role of the diversification effect as a third

dimension of performance evaluation, as pointed out by Gibson and Sidoni (2013). In their

book, the authors suggested that a performance evaluation using only two dimensions, return

and volatility, might not be enough in the present investment environment. Gibson and Sidoni

(2013) suggested that the high total volatility of funds comes from the inability of managers to

diversify the diversifiable risk.

However, the results from this study—that the high level of idiosyncratic risk is

associated with the high performer fund—implies that the high level of idiosyncratic risk may

come from the intention not to diversify to the full level rather than the inability to diversify, as

pointed out by Gibson and Sidoni (2013). Nevertheless, this study agrees with Gibson and

Sidoni (2013)—that perhaps the third dimension of performance evaluation could be useful.

Adding idiosyncratic risk, apart from total volatility and return, as the third dimension could

enhance the performance evaluation of mutual funds.

The significance of the Age variable from the regression was interesting. As a fund

ages its management team can also gain experience about the market, hence can manage the

fund better. This implies that the human capital that is inherited in the fund may play a

significant role in the success of the fund. Another advantage of aged fund versus younger


49

funds is that the unitholder of the aged fund may be more stable compared to the unit holder of

the younger fund. The fluctuation of fund inflow and outflow in a turbulent time could be less

with aged funds compared to younger funds, and this would help the fund manager manage

better. These characteristics embedded in the older funds increase the probability that these

funds will perform better than younger funds.

The significant negative impact of the No.Fund variable on the fund performance has

implication about managements as well. The efforts that the management teams have to put

into managing more funds could limit their abilities to perform well. Companies with too many

funds were documented to divert their resources from their focus on managing the funds to

marketing and administration activities (Bogle, 2010).

The results of the three simulations in meeting the third objective, which tested

whether the results from this study could be applied to the enhancement of investment return,

also offer suggestions concerning the characteristics of the Thai open-end equity mutual fund

industry.

The first simulation results uniformly indicated that over the seven-year period, all

types of low-risk strategies could help investors accumulate their wealth better than the high-

risk strategies. This implies that risk is manageable, especially in the hands of professional

management. Funds with low risk can deliver a return higher than funds with high risk.

The second simulation suggest the best fund to invest in using the risk-return

matrix. The matrix revealed the top five funds that can perform the highest number of times at

the “Sweet Spot” as candidates. The return of these suggested funds beat the average

industry. One interesting fact is that these five funds were all managed by the same company.

This implies that the organization variable may play an important role in the fund’s performance.

This implication is in line with the regression results that indicated that fund age and the

number of funds in the company are significant factors in the ability of the fund to deliver a low-

risk, high-return performance.


50

In the third simulation where the portfolio consisted of funds that passed the

characteristic criteria, the average return of the portfolio was higher than the industry average.

This implies that the organizational factors and the characteristic of funds can be a useful

indicator of fund performance.

The results of this study imply that perhaps the fundamental multifactor model could

be explored further to see whether it will have better predictive power on mutual fund

performance than the single factor model.

6.5 Limitations of the Research and Recommendations for Future Studies

The mutual fund market in Thailand has a rather short history compared to the

western market; hence, at the time of this research, there were limited numbers of literature

concerning the Thai market. To make the study more meaningful, the author of this study

expanded its literature review section to cover both Thai and foreign studies.

There were limitations regarding the sample used in this study. Although this study

concluded that, the Bowman paradox could exist in the mutual fund industry, which was in line

with the studies of Brockett et al. (1992) and Cooper et al. (2011), such a conclusion is based

on samples from the Thai mutual fund industry only. In order to strengthen the evidence of

such a phenomenon, this study encouraged the examination of the paradox using samples from

other countries.

Another limitation concerning the sample was that the Thai mutual fund industry is

considered to be in its youth, since the industry opened to retail investment only in 1996

(Thaimutualfund, 2014). The rather short history of the industry led to the necessity of applying

the panel data analysis. The advantages of this technique are that it increases the degree of

freedom and hence will increase the variability of the data and can capture greater capacity for

the complexity of data than single cross-section or time-series data. The panel methodology

has advantages as well as limitation. The challenge of applying this technique is to be cautious

regarding the impact of unobserved heterogeneity. In order to overcome this challenge, this


51

study verified the suitability of the fixed-effect model versus the random-effect model using the

specification test. The Hausman specification test suggested the random-effect model for this

study.

In order to strengthen the accuracy of the results, this study encourages similar

testing in the future, when the Thai mutual fund industry is more mature. If, in the future, the

negative risk-return relationship is still present in the absence of the above limitations, its time

and space invariance results will be most valuable to the academic society.

Despite the above limitations, the sample of this study is still valid enough to support

its results. The ten-year period sample in this study is already longer than the five-year period

used in Bowman (1980), in Brockett et al. (1992), and by Cooper et al. (2011), and the Thai

mutual fund industry is considered a good representative of the emerging market (Suppa-Aim,

2010).

This study attempted to explore the factors affecting the negative risk-return trade-

off. In this very first attempt, it used the idiosyncratic risk and other fund characteristics as the

explanatory variables, since they were supported by various literature in terms of having effects

on the fund performance. The results have already revealed three factors that will affect the

probability that the funds will deliver low-risk, high-return performance: idiosyncratic risk, fund

age, and the total number of funds managed by an asset management company. As

elaborated previously, there are sound reasons why these three factors explain the paradox.

However, this study encourages more research on other factors that will have an effect on such

performance.

The last recommendation for future study concerns the risk-return matrix, the

performance measurement proposed in this study. This study showed that this two-dimension

matrix is an efficient tool to be used in mutual fund investment; however, the results of this

study also open the opportunity to explore the third dimension of performance evaluation.


52

Bibliography

AIMC. (2012, December 31). Mutual fund market share. Retrieved December 15, 2014, from Association of Investment Manangement Companies: http://http://oldweb.aimc.or.th/21_infostats_nav.php

AIMC. (2013). Fund investment consultant. Bangkok: Association of Investment Management Companies.

AIMC. (2014a, May 19). Mutual fund statistic. Retrieved December 11, 2014, from Association of Investment Management Companies: www.aimc.or.th/2011/21

AIMC. (2014b, May 19). Industry overview. Retrieved December 11, 2014, from Association of Investment Management Companies: http://oldweb.aimc.or.th/21_overview_detail.php?nid=6&subid=0&ntype=2

AIMC. (2015, March 20). Fund market share. Retrieved January 18, 2015, from Association of Investment Management Companies: http://oldweb.aimc.or.th/en/21_infostats_nav.php

Aldrich, J. H., & Nelson, F. D. (1984). Linear probability, logit, and probit models. Newbury Park: Sage Publications Inc.

Amihud, Y., & Goyenko, R. (2013). Mutual fund's r squared as predictor of performance. Review of Financial Studies, 26(3), 667-694.

Andersen, T. J., Denrell, J., & Bettis, R. A. (2007). Strategic responsiveness and bowman's risk-return paradox. Strategic Management Journal, 28(4), 407-429.

Anderson, D. R., Sweeney, D. J., & Williams, T. A. (1990). Statistics for business and economics (4th ed.). St. Paul: West Publishing Company.

Ang, A., Hodrick, R. J., Xing, Y., & Zhang, X. (2006). The cross-section of volatility and expected returns. Journal of Finance, 61(1), 259-299.

Ang, A., Hodrick, R. J., Xing, Y., & Zhang, X. (2008). High idiosyncratic volatility and low returns: International and further U.S. evidence. Journal of Financial Economic, 91(1), 1-23.

Argyrous, G. (2005). Statistics for research (2nd ed.). London: SAGE Publications Ltd.

Backus, D. K., & Gregory, A. W. (1993). Theoretical relations between risk premiums and conditional variances. Journal of Business and Economics Statistics, 11(2), 177-185.


53

Bacon, C. (2008). Practical portfolio performance measurement and attribution. (2nd, Ed.) Chichester, West Sussex, England: John Wiley & Sons, Ltd.

Bailey, J. V., Richards, T. M., & Tierney, D. E. (2009). Evaluating portfolio performance. In P. Lawton, & T. Jankowski, Investment performance measurement (p. 11). Hoboken, New Jersey: John Wiley & Sons, Inc.

Bain, W. G. (1996). Investment performance measurement. Cambridge: Woodhead Publishing.

Bank of Thailand. (2015). Asset management companies: Assets and liabilities. Bangkok: Bank of Thailand. Retrieved September 23, 2015, from https://www.bot.or.th/Thai/Statistics/FinancialInstitutions/AMC/Pages/default.aspx

Banz, R. W. (1981). The relationship between return and market value of common stocks. Journal of Financial Economics, 9(1), 3-18.

Baucus, D. A., Golec, J. H., & Cooper, J. R. (1993). Estimating risk-return relationships: An analysis of measures. Strategic Management Journal, 14(5), 387-396. Retrieved from http;//www.jstor.org/stable/2486824

Bellman, E. (2012, November 20). New investment avenues in southeast asia. Retrieved October 10, 2014, from The Wall Street Journal: http://www.wsj.com/articles/SB10001424127887324352004578130693768238324

Bennett, P. C. (2013). Financial economics of index annuities: An analysis of investor returns. Boca Raton, Florida, USA: Universal-Publishers.

Berk, J. B., & Green, R. C. (2004). Mutual fund flows and performance in rational markets. Journal of Political Economy, 112(6), 1269-1295.

Berk, J. B., & Stanton, R. (2007). Managerial ability, compensation, and the closed-end fund discount. Journal of Finance, 62(2), 529-556.

Bettis, R. A., & Hall, W. K. (1982). Diversification strategy, accounting determined risk, and accouting determined return. Academy of Management Journal, 25(2), 254-264.

Bettis, R. A., & Mahajan, V. (1985). Risk/return performance of diversified firms. Management Science, 31(7), 785-799.

Bhattacharya, U., Lee, J. H., & Pool, V. K. (2013). Conflicting family values in mutual fund families. Journal of Finance, 68(1), 173-200.


54

Black, F., Jensen, M. C., & Scholes, M. (1972). The capital asset pricing model: Some empirical tests. In M. C. Jensen, Studies in the theory of capital markets. Praeger Publishers Inc.

Bodie, Z., Kane, A., & Marcus, A. J. (2011). Investment and portfolio management (9th ed.). New York, NY, United States: McGraw-Hill.

Bodie, Z., Kane, A., Marcus, A. J., & Jain, R. (2014). Investments. New York: McGraw-Hill/Irwin.

Bogle, J. C. (2005). The mutual fund industry 60 years later: For better or worse? Financial Analysts Journal, 61, 15-24.

Bogle, J. C. (2010). Common sense on mutual funds. Hoboken: John Wiley & Sons, Inc.

Bowman, E. H. (1980). A risk-return paradox for strategic management. Sloan Management Review, 21(3), 17-31.

Brockett, P. L., Charnes, A., Cooper, W. W., Kwon, K.-H., & Ruefli, T. W. (1992). Chance constrained programming approach to empirical analyses of mutual fund investment strategies. Decision Sciences, 23(2), 385-408.

Brown, S. J., & Goetzmann, W. N. (1995). Performance persistence. Journal of Finance, 50(2), 679-698.

Burkey, M. L. (2014). Statistics/Econometric. Retrieved September 20, 2014, from Burkeyacademy: http://www.burkeyacademy.com/home/statistics-econometrics

Burmeister, E., Roll, R., & Ross, S. A. (1994). A practitioner's guide to arbitrage pricing theory. In A practitioner's guide to factor models (pp. 1-30). Research Foundation Publication.

Burmeister, E., Roll, R., & Ross, S. A. (2003). Using macroeconomic factors to control portfolio risk. Retrieved May 26, 2014, from http://www.ftse.com/Analytics/BIRR/Documents/Using_Macroeconomic_Factors.pdf

Callin, S. C. (2008). Alpha, beta, and alpha-beta seperation. In S. C. Callin, Portable alpha theory and practice (pp. 95-133). Hoboken, New Jersey, U.S.A.: John Wiley & Sons, Inc.

Cameron, A. C., & Hall, A. D. (2003). A survival analysis of australian equity mutual funds. Australian Journal of Management, 28(2), 209-226.


55

Campbell, J. Y., Lettau, M., Malkiel, B. G., & Xu, Y. (2001). Have individual stocks become more volatile? An empirical exploration of idiosyncratic risk. Journal of Finance, 56(1), 1-43.

Capon, N., Fitzsimons, G. J., & Prince, R. A. (1996). An individual level analysis of the mutual fund investment decision. Journal of Financial Services Research, 10(1), 59-82.

Carhart, M. M. (1997). On persistence in mutual fund performance. Journal of Finance, 52(1), 57-82.

Carhart, M. M., Carpenter, J. N., Lynch, A. W., & Musto, D. K. (2002). Mutual fund survivorship. Review of Financial Studies, 15(5), 1439-1463.

Chan, L. K., Chen, H.-L., & Lakonishok, J. (2002). On mutual fund investment styles. Review of Financial Studies, 15(5), 1407-1437.

Chan, L. K., Hamao, Y., & Lakonishok, J. (1991). Fundamentals and stock returns in Japan. Journal of Finance, 46(5), 1739-1789.

Chandler Jr., A. D. (1962). Stategy and structure: Chapters in the history of the american industrial enterprise. Cambridge: Bread Books.

Chen, J., Hong, H., Huang, M., & Kubik, J. D. (2004). Does fund size erode mutual fund performance? The role of liquidity and organization. American Economic Review, 94(5), 1276-1301.

Chen, N.-F., Roll, R., & Ross, S. A. (1986). Economic forces and the stock market. Journal of Business, 59(3), 383-403.

Clark, A. (2013, February 4). Lipper Insight. Retrieved September 4, 2014, from Thomson Reuters: http://lipperinsight.thomsonreuters.com/2013/02/idiosyncratic-risk-fund-returns-and-fund-flows/

Coakes, S. J., & Ong, C. (2010). SPSS version 18 for windows. Australia: John Wiley & Sons, Inc.

Connor, G. (1995). The three types of factor models: A comparison of their explanatory power. Financial Analysts Journal, 51(3), 42-46.

Cooper, W. W., Ruefli, T. W., & Wilson, C. L. (2011). Strategy and implementation: Quantitative evaluation of strategic performance with application to mutual funds (November 1, 2011). McCombs Research Paper Series No. IROM-09-11, 1-44. Retrieved from http://ssrn.com/abstract=1963927


56

Daniel, K., Grinblatt, M., Titman, S., & Wermers, R. (1997). Measuring mutual fund performance with characteristic-based benchmarks. Journal of Finance, 52(3), 1035-1058.

Deloitte. (2012). Growth opportunities for financial services in Thailand: Driving the long-term potential. Deloitte Touche Tohmatsu Limited, Center for Financial Services. New York: Deloitte Development LLC.

Dielman, T. E. (2005). Applied regression analysis. Mason: South-Western Cengage Learning.

Dietz, P. O. (1965). Evaluating the investment performance of noninsured pension funds. Journal of Finance, 20(4), 720-721.

Dietz, P. O. (1966). Pension fund investment performance: What method to use when. Financial Analysts Journal, 22(1), 83-86.

Donnelly, A. S. (1985). The three rs of investing: Return, risk, and relativity. Homewood: Dow-Jones-Irwin.

Elton, E. J., Gruber, M. J., & Blake, C. R. (2012). Does mutual fund size matter? The relationship between size and performance. Review of Asset Pricing Studies, 2(1), 31-55.

Elton, E. j., Gruber, M. J., Brown, S. J., & Goetzmann, W. N. (2011). Modern portfolio theory and investment analysis. Hoboken, NJ: John Wiley & Sons, Inc.

Encyclopaedia Britannica. (2013). Henry Briggs. Retrieved June 6, 2014, from Encyclopedia Britannica: http://global.britannica.com/EBchecked/topic/79504/Henry-Briggs

Evans, J. L., & Archer, S. H. (1968). Diversification and the reduction of dispersion: An empirical analysis. Journal of Finance, 23(5), 761-767.

Evans, R. B. (2010). Mutual fund incubation. Journal of Finance, 65(4), 1581-1611.

Fabozzi, F. J. (1998). Active equity portfolio management. New Hope: FJF associates.

Fama, E. F., & French, K. R. (1993). Common risk factors in the returns on bonds and stocks. Journal of Financial Economics, 33, 3-56.

Fama, E. F., & French, K. R. (1998). Value versus growth: The international evidence. Journal of Finance, 53(6), 1975-1999.

Fama, E. F., & French, K. R. (2012). Size, value, and momentum in international stock returns. Journal of Financial Economics, 105(3), 457-472.


57

Fama, E. F., & MacBeth, J. D. (1973). Risk, return, and equilibrium: Empirical tests. Journal of Political Economy, 8(3), 607-636.

Ferris, S. P., & Yan, X. S. (2009). Agency Costs, Governance, and Organizational Forms: Evidence from the Mutual Fund Industry. Journal of Banking and Finance, 33, 619-626.

Fiegenbaum, A. (1990). Prospect theory and the risk-return association: An empirical examination in 85 industries. Journal of Economic Behavior and Organization, 14(2), 187-203.

Fiegenbaum, A., & Thomas, H. (1986). Dynamic and risk measurement perspective on bowman's risk return paradox for strategic management: An empirical study. Strategic Management Journal, 7(5), 395-407.

Fiegenbaum, A., & Thomas, H. (1988). Attitudes toward risk and the risk-return paradox: Prospect theory explanations. Academy of Management Journal, 31(1), 85-106.

Frumkin, D., & Vandegrift, D. (2009). The effect of size, age, beta and disclosure requirements on hedge fund performance. Journal of Derivatives and Hedge Funds, 15, 241–251.

Fu, F. (2009). Idiosyncratic risk and the cross-section of expected stock returns. Journal of Financial Economics, 91, 24-37.

Gasper, J.-M., Massa, M., & Matos, P. (2006). Favoritism in mutual fund families? Evidence on strategic cross-fund subsidization. Journal of Finance, 61(1), 73-13.

Gibson, R. C., & Sidoni, C. J. (2013). Asset allocation: Balancing financial risk (5th ed.). New York, USA: Mc Graw Hill.

GIPS. (2013). Standards and interpretation. Charlottesville, Virginia: CFA Institute. Retrieved October 16, 2013, from Global Investment Performance Standards 2010: http://www.cfapubs.org/doi/pdf/10.2469/ccb.v2010.n5.1

Gitman, L. J., Joehnk, M. D., & Smart, S. B. (2011). Fundamentals of investing (11th International ed.). Boston: Pearson Education Inc.

Goetzmann, W. N., Ingersoll Jr., J., & Ivkovic, Z. (2000). Monthly measurement of daily timers. Journal of Financial and Quantitative Analysis, 35(3), 257-290.

Goodman, S. (2008). A dirty dozen: twelve p-value misconceptions. Seminars in Hematology, 45(3), 135-140.


58

Goyal, A., & Santa-Clara, P. (2003). Idiosyncratic risk matters. Journal of Finance, 58(3), 975-1003.

Graham, B. (2003). The intelligent investor. New York: Harper Collins Publishers Inc.

GrantThornton. (2014, July 27). Thailand: The business growth environment. Retrieved from GrantThornton: http://www.grantthornton.co.th/publications/IBR2014_Thailand_focus.pdf

Gremillion, L. (2005). Mutual fund industry handbook: A comprehensive guide for investment professionals. Hoboken: John Wiley & Sons.

Grinblatt, M., & Titman, S. (1989). Mutual fund performance. Journal of Business, 62, 393-416.

Grinblatt, M., & Titman, S. (1993). Performance measurement without benchmarks: An examination of mutual fund returns. Journal of Business, 66(1), 47-68.

Gruber, M. J. (1996). Another puzzle: The growth in actively managed mutual funds. Journal of Finance, 51(3), 783-810.

Guo, H., & Savickas, R. (2008). Average idiosyncratic volatility in g7 countries. Review of Financial Studies, 21(3), 1259-1296.

Guo, H., & Whitelaw, R. F. (2006). Uncovering the risk-return relation in the stock market. Journal of Finance, 61(3), 1433-1463.

Hamilton, L. C. (2004). Statistics with stata. Toronto: Thomson Learning.

Han, Y., Hu, T., & Lesmond, D. (2014, June 23). Liquidity biases and the pricing of cross-sectional idiosyncratic volatility around the world. AFA 2012 Chicago Meetings Paper. Chicago. Retrieved from http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1787361##

Haugen, R. A. (2001). Modern investment theory (5th ed.). New Jersey: Pearson Education.

Hausman, J. A. (1978). Specification tests in econometrics. Econometrica, 46(6), 1251-1271.

Henkel, J. (2000). Risk-return fallacy. Schmalenbach Business Review, 52, 363-373.

Henkel, J. (2009). The risk-return paradox for strategic management: Disentangling true and spurious effects. Strategic Management Journal, 30(3), 287-303.

Henriksson, R. D., & Merton, R. C. (1981). On market timing and investment performance II: Statistical procedure for evaluating forecasting skills. Journal of Business, 54(4), 513-533.


59

Higgins, J. J. (2004). Introduction to modern nonparametrics statistics. Pacific Grove: Thompson Learning, Inc.

Ho, R. (2006). Handbook of univariate and multivariate data analysis and interpretation with spss. Boca Raton: Chapman & Hall/CRC.

Hosmer, D. W. (2000). Applied logistic regression. New York: John Wiley & Son.

Huang, X., & Shi, Q. (2013). Explaining the performance of chinese equity funds. 21st Pacific-Basin Finance, Economics, Accounting, and Management Conference 4-5 July 2013, (p. 19).

ICI. (2001). Redemption activity of mutual fund owners. Fundamentals, 10(1), 1-8.

ICI. (2006). Investment company fact book. Retrieved 3 August, 2015, from Investment Company Institute: http://www.icifactbook.org/2006/06_fb_glossary.html

ICI. (2015, September 30). Research and statistics. Retrieved November 5, 2015, from Investment Company Institute: https://www.ici.org/research/stats/worldwide/ww_q2_15

InterActiveCorp. (2014). Dictionary. Retrieved February 1, 2014, from Investopedia: www.investopedia.com/terms/m/morningstarinc.asp

Irvine, P. J., & Pontiff, J. (2009). Idiosyncratic return volatility, cash flows, and product market compettition. Review of Financial Studies, 22(3), 1149-1177.

Jegers, M. (1991). Prospect theory and the risk-return relation: Some belgian evidence. Academy of Management Journal, 34(1), 215-225.

Jensen, M. C. (1968). Problems in selection of security portfolios: The performance of mutual funds in the period 1945-1964. Journal of Finance, 23(2), 389-416.

Johnson, H. J. (1994). Prospect theory in commercial banking industry. Journal of Financial And Strategic Decisions, 7(1), 73-89.

Jong, F. d., & Wingens, L. (2013). Do firm characteristics influence mutual fund performance? An empirical study for europe mutual funds. Journal of Financial Perspectives, 1(1), 1-14.

Judge, O. G., Hill, R. C., Lutkepohl, H., & Lee, C. H. (1985). The theory and practice of econometrics. New York: Wiley.

Kacperczyk, M., Sialm, C., & Zheng, L. (2005). On the industry concentration of actively managed equity mutual funds. Journal of Finance, 60(4), 1983-2011.


60

Kacperczyk, M., Sialm, C., & Zheng, L. (2008). Unobserved actions of mutual funds. Review of Financial Studies, 21(6), 2379-2416.

Khorana, A. (2001). Performance changes following top manager turnover: Evidence from open-end mutual funds. Journal of Financial and Quantitative Analysis, 36(3), 371-393.

Kmenta, J. (1986). Elements of econometrics. (2nd, Ed.) New York: Macmillan.

Kogan, L., & Papanikolaou, D. (2013). Firm characteristics and stock returns: The role of investment-specific shocks. Review of Financial Studies, 26(11), 2718-2759.

Koks, D. (2006). Explorations in mathematical physics: The concepts behind an elegant language. New York: Springer.

KResearch. (2013, February 1). K-econ analysis. Retrieved November 11, 2013, from Kasikorn Research: www.kasikornresearch.com/EN/K-EconAnalysis/pages/ViewSummary

Lawton, P., & Jankowski, T. (2001). Investment performance measurement. Hoboken, New Jersey: John Willey & Sons, Inc.

Lintner, J. (1965). The valuation of risk assets and the selection of risky investments in stock portfolios and capital budgets. Review of Economics and Statistics, 47(1), 13-37.

Malkiel, B. G., & Xu, Y. (2003). Investigating the behavior of idiosyncratic volatility. Journal of Business, 76(4), 613-644.

Markowitz, H. (1952). Portfolio selection. Journal of Finance, 7(1), 77-91.

McMillan, M. G., Pinto, J. E., Pirie, W. L., & Van de Venter, G. (2011). Investments: Principles of portfolio and equity analysis. New Jersey: John Wiley & Sons, Inc.

Merriam-Webster. (1991). Webster's ninth new collegiate dictionary. Springfield, Massachusetts, U.S.A.: Merriam-Webster Inc.

Merton, R. C. (1973). An intertemporal capital asset pricing model. Econometrica, 41(5), 867-887.

Miller, K. D., & Chen, W.-R. (2003). Risk and firm's costs. Strategic Organization, 1(4), 355-382.

Morningstar. (2014). Morningstar style box. Retrieved February 1, 2014, from Morningstar: www.morningstar.com/InvGlossary/morningstar_style_box.aspx

Mossin, J. (1966). Equilibrium in a capital asset market. Econometrica, 34(4), 768-783.


61

MSCI. (2014). Barra equity model. Retrieved May 30, 2014, from MSCI inc: www.msci.com/products/portfolio_management_analytics/equity_models/

Nanda, V., Wang, Z. J., & Zheng, L. (2004). Family values and star phenomenon. Review of Financial Studies, 17(3), 667-698.

Nathaphan, S. (2010). Thai mutual fund industry: Challenge and opportunity in the next decade, CMRI research paper 4/2553. Retrieved April 1, 2014, from https://www.set.or.th/setresearch/files/symposium/2553_paper1_FullPaper.pdf

Nathaphan, S., & Chunhachinda, P. (2012). Determinants of growth for thai mutual fund Industry. International Research Journal of Finance and Economics(86), 120-131.

NESDB. (2011). The eleventh national economic and social development plan. Bangkok: National Economic and Social Development Board. Retrieved June 1, 2014, from http://www.nesdb.go.th/ewt_news.php?nid=5748&filename=develop_issue

Novy-Marx, R. (2007). An equilibruim model of investment under uncertainty. Review of Financial Studies, 20(5), 1461-1502. doi:10.1093/revfin/hhm01

Ooi, J. T., Wang, J., & Webb, J. R. (2009). Idiosyncratic risk and reit returns. Journal of Real Estate Finance and Economics, 38(4), 420-442.

Oulton, N. (2007). Ex post versus ex ante measures of the user cost of capital. Review of Income and Wealth, 53(2), 295-317.

Park, H. M. (2010). Hypothesis testing and statistical power of a test. Working paper, Indiana University, The University Information Technology Services (UITS) Center for Statistical, Bloomington. Retrieved June 5, 2014, from http://www.indiana.edu/~statmath/stat/all/power/index.html

Pastor, L., Stambaugh, R. F., & Taylor, L. A. (2014). Scale and skill in active management (BFI Working paper 2014-03). The University of Chicago. Chicago: The Becker Friedman Institute For Research in Economics.

Pawabutr, P., Charoenrook, A., & Nataphan, S. (2011). Fund manager survey program. Bangkok: Capital Market Research Institute.

Peterson, P. P., & Fabozzi, F. J. (1999). Analysis of Financial Statement. New Hope: FJF Associates.

Podesta, F. (2002). Recent development in quantitative comparative methodology. Brescia, Italy: University of Brescia.


62

Pollet, J. M., & Wilson, M. (2008). How does size affect mutual fund behavior? Journal of Finance, 63(6), 2941-2969.

Priceton University. (2007). Data and statistic service. Retrieved August 3, 2014, from Princeton University: http://dss.princeton.edu/online_help/stats_packages/stata/panel.htm

Qin, N. (2013). Idiosyncratic risk and mutual fund performance persistence. Midwest Finance Association Annual Meeting, 2013. Chicago: Finance Management Association. Retrieved September 15, 2014, from http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1974805

Reilly, F. K., & Brown, K. C. (2011). Analysis of investments and management of portfolios. Canada: South-Western.

Roll, R. (1978). Ambiguity when performance is measured by the securities market line. Journal of Finance, 33(4), 1051-1069.

Ross, S. A. (1976). The arbitrage theory of capital asset pricing. Journal of Economic Theory, 13, 341-360.

Roubini, N. (2011). Crisis economics. London: Penguin Books Ltd.

SAS. (2014). Knowlwdge base. Retrieved July 9, 2014, from SAS Institute Inc.: http://support.sas.com/documentation/cdl/en/etsug/63348/HTML/default/viewer.htm#etsug_panel_sect039.htm

Schmidheiny, K. (2014). Panel data: Fixed and random effects. Short guides to microeconometrics, pp. 1-16. Retrieved August 31, 2014, from http://kurt.schmidheiny.name/teaching/panel2up.pdf

SEC. (2013). MRAP. Retrieved August 27, 2013, from The Securities Exchange Commission: http://market.sec.or.th/public/mrap/MRAPView.aspx?FTYPE=M&PID=0357&PYR=2547

See, Y. P., & Jusoh, R. (2012). Fund characteristics and fund performance. International Journal of Economics, 1(9), 31-43.

SET. (2014, January 5). Annual Report 2014: The stock exchange of thailand. Bangkok: Stock Exchange of Thailand. Retrieved January 20, 2015, from Stock Exchange of Thailand: http://www.set.or.th/en/about/annual/files/annual_report_2014_eng_v2.pdf

SET. (2015). Stock exchange of thailand shows excellence performance in 2014. Retrieved January 5, 2015, from Stock Exchange of Thailand: http://www.set.or.th/set/newsdetails.do


63

Sharpe, W. F. (1964). Capital asset prices: A theory of market equilibrium under condition of risk. Journal of Finance, 19(3), 425-442.

Sharpe, W. F. (1966). Mutual fund performance. Journal of Business, 39(1), 119-138.

Sinha, T. (1994). Prospect theory and the risk return association: Another look. Journal of Economic Behavior and Organization, 24, 225-231.

Skrinjaric, T. (2013). Market timing ability of mutual funds with tests applied on several croation funds. Croatian Operational Research Review, 4, 176-186.

Stambaugh, R. F., Yu, J., & Yuan, Y. (2015). Arbitrage asymmetry and the idiosyncrstic volatility puzzle. Journal of Finance, 70(5), 1903-1948.

Statman, M. (1987). How many stocks make a diversified portfolio? Journal of Financial and Quantitative Analysis, 22(3), 353-363.

Suppa-Aim, T. (2010). Mutual fund performance in emerging markets: The case of Thailand ((Unpublished doctoral dissertation) ed.). West Midlands, U.S.

Switzer, L. N., & Huang, Y. (2007). How does human capital affect the performance of small and mid-cap mutual funds? Journal of Intellectual Capital, 8(4), 666-681.

Tangjitprom, N. (2014). The effect of fund size on performance: The evidence from active equity mutual funds in Thailand. Research Journal of Finance and Accounting, 5(10), 82-86.

Thaimutualfund. (2014, May 15). History of mutual fund. Retrieved July 30, 2015, from www.thaimutualfund.com: http://www.thaimutualfund.com/AIMC/aimc_about.jsp

The Maine Office of Securities. (2013, June 17). Investing Basics. Retrieved June 1, 2014, from The Office of securities State of Maine: http://www.maine.gov/pfr/securities/investing_basics.htm

Tobin, J. (1958). Liquidity preference as behavior towards risk. Review of Economic Studies, 25(1), 65-86.

Treynor, J. L. (1965). How to rate management of investment funds. Harvard Business

Review, 43(1), 63-75.

Treynor, J. L., & Mazuy, K. K. (1966). Can mutual fund outguess the market? Harvard

Business Review, 44, 131-136.

Ungphakorn, T. (2014). Mutual fund flow behavior in Thailand. Bu Academic Review, 13(2),

1-15. Retrieved from

http://www.bu.ac.th/knowledgecenter/epaper/july_dec2014/pdf/ac01.pdf


64

Wang, K.-C. (2005). Business life cycle and five elements theory. Ph.D Dissertation.

Retrieved April 8, 2014, from systemicbusiness.org/pubs/2005_isss_016_Wang.pdf

Wei, S. X., & Zhang, C. (2006). Why did individual stocks become more volatile? Journal of

Business, 79(1), 259-292.

Wiseman, R. M., & Bromiley, P. (1991). Risk return associations: Paradox or artifact? An

empirically tested explanation. Strategic Management Journal, 12(3), 231-241.

Wooldridge, J. M. (2002). Econometric analysis of cross section and panel data. Cambridge:

MIT Press.

World Bank. (2012). Economy rankings. Retrieved November 9, 2013, from World bank

doing business project: www.doingbusiness.org/rankings

Zhou, X. (2004). Two essays on mutual funds. Dissertation, University of North Carolina,

Finance, Chapel Hill.

Zhu, K., Zhou, X., Xu, S., Sun, D., Ren, W., Zhou, K., & Yang, G. (2014). The loss of taste

genes in cetaceans. BMC Evolutionary Biology, 14(218), 1-10. Retrieved October 30,

2015, from https://dash.harvard.edu/handle/1/13454766