Equity Risk Premium in Emerging Markets

Master thesis

Equity Risk Premium in Emerging Markets

| Paul van Baars s1164201 FWM

Mobile:+316-3824 9442

Email: Paulvanbaars@hotmail.com

Starting date research: 29 November 2004

Internship: ING Asia Corporate Finance, Hong Kong

Professor: Dr. W. Westerman

Preface

The final stage of the study at the Faculty of Management and Organisation of the Rijksuniversiteit Groningen consists of a six month internship and a scientific research conducted during that period. For my master thesis assignment I preferred to go China, because the economical intensity really intrigues me. As connection to my Finance master program I decided to go to Hong Kong, in my opinion the financial heart of China.

My internship was held at ING Bank, the Corporate Finance division.

The time spend in Hong Kong was terrific, as well in social as professional point of view.

Living abroad enriches your view on life on various levels. The people at ING and people of Hong Kong opened my eyes and made me determined to return to South East Asia one day. Especially Eduard Wehry, Francis Ansvananda and Mandy Ho helped me a lot in settling in Hong Kong and, more important, getting my thesis into gear. I owe them much gratitude for that.

Regarding the finalising of the thesis I’m much obliged to Dr. Westerman. In line with his sharp and intriguing classes, he really helped me to structure the information and added a much needed focus to this thesis. Without this comments and suggestions the whole process would have lasted even longer.

Somehow relieved that the end of this chapter in my life is about to close, I’m looking forward to the next step. The fact that I’m pursuing a career in Corporate Finance, might even be the strongest compliment for my internship and master program.

Amsterdam, June 2006 Paul van Baars

Some

Executive Summary

Motivation

Financial institutions as well as academics are enrolled in a vivid discussion about the Equity Risk Premium (ERP), which has lead to various models that all have advocates and adversaries. ING, as any other financial institution, also faces the same issues when determining the ERP. Nowadays, an average of various outcomes of different approaches is taken, which is widely criticized by the financial sector as well as academics. Therefore ING would benefit from a model that enables her to get the right theory for the right country and therefore the right ERP, leading to correct valuations.

For ING Asia the urge for a well-defined model is even greater. Limited history, lack of market liquidity and the questionable validity of information, makes it hard to determine a correct premium for the Asian tigers.

Objective

This research has the objective to obtain the model that reflects the country’s Equity Risk Premium best, by analysis of various Cost of Capital models and the emerging markets of China, Taiwan, South Korea, India and the Philippines, in order to aid ING with the outcome. Crucial aspects of the model will be reliable data and the practical use of the model

In order to match the objectives criteria the following research question is defined:

“Which Cost of Capital model is best utilised to determine the Equity Risk Premium for certain markets in Asia?” Therefore the ING model will be analysed and compared to other ERP models.

ING Model

The ING model used is established by Damodaran. It follows the premise that one should use the ERP of a developed market and add a country risk premium. This premium is obtained by multiplying the country default spread with a ratio of the standard deviation of the equity market divided by the standard deviation of the country bond market.

Adding this to the base premium would lead to a local ERP.

Strong points of the model lie in the way country risk is incorporated. By using the

country credit rating, risk is reflected well and in combination with a US Treasury bond it

is counted only once. The fact that the model makes use of historical data is not very

suitable for emerging markets (developed ERP), but the forward looking measure

(country default spread) compensates for this. However the method of adapting this

model to emerging markets is too easy. Using a 10-year UST doesn’t comply with the

short economical cycles in emerging markets and using a 1.5x multiplier for each market

seems too arbitrary. Furthermore the model returns low ERPs for risky countries.

Other models

The following models have been analysed:

1. Identical Cost of Capital Model 2. World CAPM and Multifactor Model 3. Segmented/Integrated Model

4. Bayesian Approach (Ibbotson Associates) 5. Country Risk Rating (EHV) Model 6. CAPM with Skewness

7. Goldman-integrated sovereign yield spread Model 8. Goldman-segmented Model

9. Goldman-EHV hybrid Model 10. CSFB volatility ratio Model 11. Damodaran Model

12. Salomon Smith Barney Model

Qualitative and quantitative analysis has shown that not all models proved to be useful.

Either the model was based on arbitrary assumptions (Bayesian approach, Goldman Segmented and CSFB model), only useful for developed markets (World CAPM), the model was still under construction (CAPM with Skewness) or a country didn’t possess the needed historical data (China, India and the Philippines) which lead to false outcomes.

Conclusion

The Damodaran model used by ING proved to be very robust; it was applicable to all countries. However it often returned a very low ERP. The fact that it’s based on historical figures also makes it less suitable for every emerging market. When using Damodaran the perception occurs that countries are not very risky, although they are in reality. When the final selected models were compared even a further selection could be made on intuitive or non-financial factors, which lead to the following outcome:

- China: EHV Model

- Taiwan: EHV Model, Salomon Smith Barney Model

- South Korea: Goldman Integrated Model, Damodaran Model - Philippines: EHV Model

- India: Goldman Integrated Model, Damodaran Model

Remarkable is the fact that South Korea and India both are assigned to the same models, although they are of a very different nature. This is due to the fact that India better uses models in which the traditional beta-coefficient is absent and South Korea needs a model which treats the country data as relatively developed.

The practical use is also an important driver in the model selection process. Damodaran is fairly easy to use, but this counts for all selected models. Therefore all selected models can be used by people who only have access to widely available resources.

Each country has to be assessed individually before selection an ERP model. Emerging

markets have such a different nature, that not a single model applies to all emerging

countries.

Index

Preface... 2

Executive Summary ... 3

Index... 5

Introduction... 7

1. Methodology ... 9

1.1 Introduction... 9

1.2 Research Problem ... 9

1.2.1 Research objective ... 9

1.2.2 Research question ... 9

1.3 Sub-questions... 10

1.4 Restrictions ... 10

1.5 Conceptual Framework... 11

1.5.1 ING Model... 11

1.5.2 Other ERP Models ... 12

1.5.3 Strengths, Weaknesses and Model Selection... 12

1.5.4 Model Application ... 12

1.5.5 Outcome and Conclusion... 12

1.6 Preconditions ... 13

1.7 Research Methods... 13

1.8 Data collection methods... 13

1.9 Fundamental and practical research... 15

1.10 Literature... 16

1.11 Definitions ... 17

1.12 Timeframe... 19

2. ING Bank Model... 20

2.1 Model Description ... 20

2.2 Application... 22

2.2.1 China... 22

2.2.2 Taiwan ... 23

2.2.3 South Korea ... 23

2.2.4 Philippines ... 24

2.2.5 India ... 24

2.3 Criticism of the model ... 25

2.4 Conclusion ... 26

3. Cost of Capital in Emerging Markets Models... 27

3.1 Description General Assumptions ... 27

3.2 Description Emerging Market Models ... 29

3.3 Overview of Calculations ... 40

3.4 Conclusion ... 41

4 Conclusion ... 42

4.1 Selection... 42

4.2 Confrontation ... 45

4.3 Conclusion ... 47

4.4 Recommendation ... 48

Appendix 1 ING Model ... 50

Appendix 2 Interviews... 52

Appendix 3 Mechanics of the Models ... 60

Literature Thesis... 71

Introduction

A much argued topic in the financial sector is Equity Risk Premium (‘ERP’). The basic idea is simple; “Investors want higher compensation when they are subject to higher risk”.

The difficult point however is: how to measure this risk and allocate a premium to it? On this matter various theories are developed, which all have their advocates and adversaries.

Since ERP is essential for making investment decisions, the topic is much debated in the financial sector.

ING, as any other financial institution, also faces the same issues when determining the ERP. Nowadays, an average of various outcomes of different approaches is taken, which is widely criticized by the financial sector as well as academics. Therefore ING would benefit from a model that enables her to get the right theory for the right country and therefore the right ERP, leading to correct valuations.

For ING Asia the urge for a well-defined model is even greater. Limited history, lack of market liquidity and the questionable validity of information, makes it hard to determine a correct premium for the Asian tigers. The financial sector tends to use the Capital Asset Pricing Model (CAPM) for calculating the cost of capital. In emerging markets this is not appropriate, since the CAPM assumptions are violated, for instance the fact that emerging markets are not efficient capital markets as well as not perfect integrated (foreign investors can not freely invest in the local market and vice versa).

Conducting research, which provides ING with a well-founded ERP for a certain country, might seem ambitious, but in this thesis is not tried to show a certain theory to be right or wrong, but only relate a theory to a country. The theories are assumed to be correct, but often apply better to developed markets than emerging ones. The objective is to use these models and make the calculations for the selected countries. A reason to compare the theories and select the most appropriate one in a particular case, is because this is something which is not encountered in most of the literature.

This research comprises the Chinese, Taiwanese, Korean, Indian and the Philippine markets, since they all have a different nature, which could lead to interesting results.

In Chapter 1 an outset of the methodology used is given. The research objective and research question are defined. The latter is split into sub-questions to enhance the analysis.

As this is such a broad topic restrictions and preconditions are established. To give a graphical overview a conceptual framework is shown, which displays how the various factors are interconnected. The research consisted of various research and data collection methods, which are explained. A great deal of literature was used of which an overview is given. To conclude this chapter, a timeframe has been included.

Chapter 2 consists of an in-depth analysis of the model used by ING Bank. First the

elements of which this model consists are treated and afterwards the model is applied to

the selected countries. After calculation, these outcomes will be critically analysed and a

conclusion is drawn from this.

A purpose is to compare the ING model to other ERP models, which is done in Chapter 3.

First all selected models are analysed qualitatively and afterwards applied to the countries.

The chapter concludes with an overview of the outcome of the models. In Chapter 4 the outcomes are quantitatively analysed, which eventually leads to a selection of appropriate models. When the designated models are connected to the countries a further selection is made in order to select only those models which score high on all criteria. Afterwards these models will be compared to the ING model, in order to give a recommendation which model should be used. The chapter ends with some recommendations for further research.

The interviews and calculations are included in the Appendices

1. Methodology

1.1 Introduction

Methodology is defined as the study of methods or procedures used in a discipline so as to gain warranted knowledge (Gill & Johnson, 2002). This chapter will deal with the structure of the research.

First of all the research problem is defined, which is divided in a research objective and a research question. As this research question is to vast to cover on its own, it will be split up in various sub-questions to enhance the clarity and ability to answer the problem.

The topic is at first sight very broad, so it will be narrowed down in restrictions and preconditions.

By dealing with the sub-questions, the expectancy is to find the most appropriate way for ING to determine the Equity Risk Premium in emerging markets

1.2 Research Problem

“The research problem consists of an objective and a question. The question should be logically derived from the objective. Together they should cover exactly what will be researched and why” (Braster, 2000).

1.2.1 Research objective

Obtain the model that reflects the country’s Equity Risk Premium best, by analysis of various Cost of Capital models and the emerging markets of China, Taiwan, South Korea, India and the Philippines, in order to aid ING with the outcome.

Crucial aspects of the model will be:

• Reliable data

• Practical use of the model

Since the practical use of the model is highly important, not only scientific models will be analysed. Research has shown that scientific models tend to be very broad and complex and therefore have limited practical use. Professional models are thought to be user- friendly and are therefore included in this research.

1.2.2 Research question

Which Cost of Capital model is best utilised to determine the Equity Risk Premium for

certain markets in Asia?

1.3 Sub-questions

“The main research question will be split up in various sub questions, to enhance the clarity of the answer and give structure to the research” (Verschuren, 1999).

1. How does ING estimate the Equity Risk Premium in emerging markets?

• Of which elements is the ING model constructed?

• What is the motivation for use of this model?

• What are the pros and cons of this model?

• How specific is the focus on emerging markets?

This question will be answered in chapter 2

2. Which other methods are suitable for estimating the ERP in emerging markets?

• What are the pros and cons of each model?

• Which factors for determining the ERP are different in emerging markets than in the developed markets?

• Why is a certain model particularly suitable for emerging markets?

This question will be answered in chapter 3

3. Which model applies best to a certain country?

• Which are crucial elements for determining the Equity Risk Premium?

• Which elements are crucial for determining the Equity Risk Premium in emerging markets?

This question will be answered in chapter 4

1.4 Restrictions

Since Cost of Capital is such a broad topic, the examination of the theoretical findings will be limited to:

• The Equity Risk Premium such as generally referred to.

• The only risk that will be taken into account is Systematic Risk.

• Only direct effects on the ERP will be measured. Factors such as survivorship

bias, transfer of income from abroad (e.g. migrant workers) and the fact that some

emerging markets are becoming more developed, which leads to slower recovery

after a crisis are not taken into account.

The ERP models will only be applied to the following countries:

• China

• Taiwan

• South-Korea

• India

• Philippines

These countries are selected because they have a very different nature, so they might prove interesting to research.

1.5 Conceptual Framework

Figure 1.1 Conceptual framework

1.5.1 ING Model

The ING model will be closely examined and explained. The drivers for choosing this model, developed by Damodaran, and country credit ratings as well as the decision to use a 10-year US treasury bond will be analysed. The practical use, reliability of the data, types of risk incorporated and whether this model is especially suitable for emerging markets are the other elements, which drive heavy in selecting the various models.

ING Model Other ERP Models

Strengths and Weaknesses

Outcome Model Selection

Model

Application

1.5.2 Other ERP Models

The selected other 12 models will be treated likewise, but with less in-depth analysis than with the ING model, because this would be too extensive. Special attention will be given to models that consist of different factors than the ING model

Analysis is made of the following models:

1. Identical Cost of Capital Model 2. World CAPM and Multifactor Model 3. Segmented/Integrated Model

4. Bayesian Approach (Ibbotson Associates) 5. Country Risk Rating (EHV) Model 6. CAPM with Skewness

7. Goldman-integrated sovereign yield spread Model 8. Goldman-segmented Model

9. Goldman-EHV hybrid Model 10. CSFB volatility ratio Model 11. Damodaran Model

12. Salomon Smith Barney Model

1.5.3 Strengths, Weaknesses and Model Selection

The models will be compared on their practicability for valuation purposes, because ING attaches much value to this factor. Furthermore the reliability of data will be examined.

This is closely related to Economical, Political and Historical risk. In developed markets, historical models use reliable data, but using historical data in emerging markets might be risky. Some models pay special attention to historical risk, therefore it will be explained which impact this has on the ERP. If a model is especially designed for emerging markets, this will be taken into account during the selection process.

When all models are compared, the ones will be selected, which have the highest score on the selected objectives.

1.5.4 Model Application

The selected models will be applied on the selected countries and the ERP will be calculated. It might be possible that a model, which scores high on all objectives is not applicable to every selected country, due to the different nature of each nation. For instance a model that uses historical data is not suitable for the Philippines as historical data here is widely regarded as unreliable, but for Taiwan this model might be very suitable. This is done on purpose to enhance the contrast between emerging markets.

After selection the ERP will be calculated and an overview is given.

1.5.5 Outcome and Conclusion

In this part the outcomes will be qualitatively be analysed. This will lead to a

recommendation for ING which model to use for which market.

1.6 Preconditions

In order to establish more focus, there are certain preconditions this thesis is subject to.

Product Preconditions

• This thesis has to meet the requirements set by the Faculty Management and Organization, Rijksuniversiteit Groningen, the Netherlands.

• The thesis will be written in English, in order to fit the purpose of the thesis.

• Decisions regarding the confidentiality of the thesis are yet to be made.

Process Preconditions

• The research is conducted by order of ING Asia Corporate Finance, to look from a new angle upon their methods of estimating the Equity Risk premium in emerging markets.

• Initially the research was set to take 6 months at the ING Asia Corporate Finance office, but the time frame has been expanded as the research proved to be more complex.

1.7 Research Methods Scientific research

The ERP is a much-debated topic in the financial scientific world; the key here was to make a distinction between relevant publications. The articles selected are either the most recent on a certain area, for instance Drew and Veeraraghavan (2003) on World CAPM, or the most recent version of a publication which initially caused a paradigm shift in thinking about ERP, such as Mehra and Prescott (2003) and Ibbotson (2001).

Financial Institutions research

The practical use of a certain model is of the utmost importance. Therefore the scope of this thesis is beyond scientific models and comprises professional models as well.

Interviews

To get better understanding of the use of the Equity Risk Premium, various fund managers and macro-economists are questioned

1.8 Data collection methods

Methods for collecting data as defined by Baarda en de Goede (2001):

1. Make use of existing information

2. Obtain data through written or oral interviews

3. Obtain data through observation

This thesis is primarily based on existing information, but the interviews held did prove to be helpful in getting a better understanding of the concept ERP and helped to recognise that the focus should be on Corporate Finance and not Financial Markets.

Making use of existing information

To get more acquainted with the concept of ERP, a great deal of scientific papers has been read. But as a gentle introduction the websites of Campbell Harvey and Aswath Damodaran proved to be most helpful. These sites contain links to various articles about the basics of ERP.

When the concept became clearer, other academic studies were used to identify the various models. A great deal of theories is constructed for developed markets and is therefore not fit for this thesis. The focus shifted somewhat to models especially designed for emerging markets. When a selection of articles and models was made, the collecting of actual data for the input could start. The Financial Times provided the current market rates, Moody’s the credit ratings and Damodaran the country spreads.

The theories and data used to construct the ING model was gathered through talks with employees and making use of valuation manuals of ING.

Interviews

The reason for conducting interviews was to get a better understanding of the practical use of the ERP. The ING Corporate Finance division used the ERP for valuation purposes, so the premium was of the utmost importance. During interviews with Ho Chee Hau and Michael Chiu from ING Investment management, the perspective emerged that financial markets use the ERP as a mere benchmark, which aides them in adjusting their portfolio positions. Therefore they did not calculate the ERP themselves, but used CSFB and Goldman Sachs reports. Somehow surprised by this outcome another interview was held with Jeroen Touw of ABN Amro Asset Management. Mr. Touw also stated the ERP was not that important to asset managers.

Since the ERP is calculated by macro-economists, interviews where held with emerging markets specialist Remco Vergeer of ABN Amro in Amsterdam. He did believe the ERP is important, but also stated that the importance for Corporate Finance is greater than for Financial Markets. Mr. Vergeer told that traditional CAPM models are not widely used for emerging markets, since these markets tend to violate CAPM’s assumptions:

• Efficient capital markets (don’t exist, especially in emerging markets)

• Perfect integration of markets (foreign investors can freely invest in local market and vice versa)

Structure data collection

Baarda and de Goede (2001) make the difference between structured research, i.e. one knows exactly what data is needed, and unstructured research, i.e. one does not know what data to look for and what the outcome of the research will be.

This research is a hybrid form of the two types. The knowledge of ERP and emerging

markets was quite basic before the research started. Interviews with ING employees and

standard textbooks provided more background information, which led to exploring new

articles. Due to the initial limited knowledge, at first the research was quite unstructured.

As soon as the matter became comfortable, the research could be expanded to more in- depth scientific articles on ERP and in specific ERP in emerging markets. When professional information and financial data was needed, the research became more structured by nature. ING was a great help in collecting financial data, because they have access to sources as Thomson, DataStream and other market data providers. Their knowledge of experts in this area also enabled the research to be more focussed and structured.

Baarda and de Goede (2001) also distinguish direct and indirect data collection. When the information or data sought after is threatening or tacit, one rather not wants to use a direct approach. This only occurred during the interviews. As the most information is obtained when an interviewee can talk freely, open and objective questions were asked at first.

After getting this information, more direct and critical questions were asked to retrieve more subjective statements to acquire a better image of the interviewee’s stance towards ERP.

1.9 Fundamental and practical research

“The central issue with a theoretical or fundamental research is a knowledge problem.

This means that a certain theory will be examined or a new one will be developed”

(Braster, 2000). In this particular case only existing theories will be examined. If an

adequate answer to the research question has to be found, then there initially has to be an

extensive qualitative research on the theoretical background of ERP. Therefore the first

phase will consist of fundamental research.

Figure 1.2 Research model

When one has to solve a practical problem, Braster defines this as practical or applied research, “here fundamental research can play a part, but only in use of existing theories”

(2000). This is what happens in the second phase.

1.10 Literature

In this paragraph is outlined which literature will be used in which paragraph. As this concerns a scientific research, the fundaments of the literature will consist of academical studies concerning the ERP in emerging markets. However, these scientific models many times prove to be complex and are therefore not suitable to use for valuation purposes.

Therefore there is room for non-scientific studies, mostly performed by financial institutions.

Literature research

Select models

Link model- country

Make calculations

Fundamental research

Analyze outcomes

Assign model to country

Practical research

Determine ERP for selected emerging

market

In Chapter 2 the ING model will be analysed. This model consists of a 10-year US Treasury Bond, a multiple given by Damodaran and credit ratings by Standard & Poor’s (S&P). The choice for using a 10y US bond will be analysed and Damodaran’s stand will be explained by use of his articles "Estimating Equity Risk Premiums"(1999) and

“Country Risk and Company Exposure: Theory and Practice”(2003). In addition there will be short overview of the credit rating system of Moody’s and S&P

In Chapter 3 there will be an in depth analysis of the various models covering ERP in emerging markets. Campbell Harvey wrote an article “12 ways to identify the International Cost of Capital” (2005). This article gives a clear overview of various models, but is quite concise. I will use some of the models mentioned, but will elaborate on this using other theories by Drew and Veeraraghavan” Beta, Firm size, Book-to- market equity and Stock returns”(2003), regarding world CAPM, “The Shrinking Equity Premium”(1999) by Jeremy Siegel and regarding the Bayesian ERP, “Modelling emerging market risk premia using higher moments”(1999) by Soosung Hwang and Stephen E. Satchell regarding CAPM with Skewness.

For analysing the Goldman and Salomon Smith Barney models I will use non-scientific literature, since these are professional studies. Nonetheless they are included, as they made an impact on the ERP discussion, the studies used are “Discounting a location”(1999) by Rameen Soltani and “Emerging Market Discount Rates”(1999) by Mariscal and Hargis. The CSFB model emerged from research by Lucia Hauptman and Stefano Natella in “The cost of equity in Latin America: The eternal doubt”(1997), but is not timely anymore, therefore Harvey’s analysis (2005) of this model will be used.

In Chapter 4 the focus will be on factors that are of paramount importance in emerging markets, therefore literature by Salomons and Grootveld. “The Equity Risk Premium:

Emerging versus Developed Markets” (2003), Domowitz, Glen and Madhavan, “Country and currency Risk Premia in an emerging Market”(1998) and Drew and Veerarghavan

“A closer Look art the size and value Premium in Emerging Markets: Evidence from the Kuala Lumpur Stock Exchange”(2002) is used.

1.11 Definitions

The definitions used throughout this thesis are derived from the standard finance textbook

“Corporate Finance” by Ross, Westerfield and Jaffe (2002) and from Campbell Harvey’s financial glossary (www.biz.yahoo.com/f/g). Definitions are included as some might be multi-interpretable.

Beta-coefficient

The Beta-coefficient is a measure of the sensitivity of a security’s return to movements in an underlying factor. It’s a measured systematic risk.

Brady bond

A Brady bond is a bond issued by an emerging country under a debt reduction plan.

Capital Asset Pricing Model (CAPM)

The Capital Asset Pricing Model is an equilibrium asset pricing theory that shows that equilibrium rates of expected return on all risky assets are a function of their covariance with the market portfolio.

R

= R

+ β(R

- R

), where R

= return on equity R

= risk-free rate β = beta-coefficient

R

= expected market return Cost of Equity Capital

The Cost of Equity Capital is the required return on the company’s common stock in capital markets. It is also called the equity holders’ required rate of return because it is what equity holders can expect to obtain can in the capital market. It is a cost from the firm’s perspective.

Equity Risk Premium

The Equity Risk Premium (ERP) is the excess return on the risky asset that is the difference between expected return on risky assets and the return on risk-free assets.

Ex post / ex ante

Ex post is the use of historical returns, whereas ex ante means the use of forecasted returns.

Geometric / Arithmetic

The geometric mean multiplies all numbers is a series and the takes the n

root of this product; where n is the amount of numbers in the series.

The arithmetic mean is the sum of all numbers in a series, divided by total amount of numbers.

ING

Throughout this thesis, ING is used for ING Asia Pacific, the Corporate Finance division, unless stated otherwise.

Efficient capital markets

An efficient capital market is a market in which there are never any arbitrage opportunities, whereby arbitrage is defined as the simultaneous buying and selling of a security at two different prices in two different markets, resulting in profits without risk.

Perfectly efficient markets present no arbitrage opportunities. Perfectly efficient markets seldom exist, but arbitrage opportunities are often precluded because of transactions costs.

Perfect integrated markets

Perfect integrated markets are markets where locals can freely invest abroad and

foreigners can invest unlimited in this market.

Risk–free rate

The risk-free rate is defined as ‘the rate earned on a risk-less asset, which is commonly defined as short-term obligations of the U.S. government’. However risk-free doesn’t mean free of risk. There is a chance the issuer of the bond will default.

Skewness

Negative skewness means there is a substantial probability of a big negative return.

Positive skewness means that there is a greater-than-normal probability of a big positive return.

Systematic Risk

Systematic risk is any risk that affects a large number of assets, each to greater or lesser degree.

Systematic Risk Principle

The systematic risk principle is that only the systematic portion of risk matters in large, well-diversified portfolios. Thus, the expected returns must be related only to systematic risk.

Volatility

Volatility is a statistical measure of the tendency of a market or security to rise or fall sharply within a period set.

1.12 Timeframe

Most of the practical research is conducted at ING Asia, whereas the literature study primarily was conducted in the Netherlands. Since some problems emerged during the process, there is an exceptional long time spent on the Proposal and the Literature study.

Figure 1.3 Timeframe

Proposal Literature study Research Writing Thesis

Sep 04 Dec 04 Mar 05 Jun 05 Sep 05 Dec 05 Mar 06 Jun 06

2. ING Bank Model

ING uses the model established by Damodaran (1999, 2003) for estimating the ERP. In this chapter, first some elements of cost of equity (risk-free rate, ERP and Country risk premium), will be treated, because ING has a different look on these elements than other models that are used throughout this thesis. Afterwards the actual ERP for each country will be calculated. The outcomes of the calculations will be analysed and the chapter ends with a conclusion of ING’s methodology. The insights that enabled Damodaran to construct this model will be dealt with in here and in Chapter 3, because there an overview of all strengths and weaknesses is given. In paragraph 2.4 the answer to sub- question 1 ‘How does ING estimate the Equity Risk Premium in emerging markets?’ will be given.

2.1 Model Description

In this part all elements which establish this model are explained and mentioned on which assumptions they are based.

Damodaran Model

The model used by ING is established by Damodaran (1999, 2003). It follows the premise that one should use the ERP of a developed market and add a country risk premium. This premium is obtained by multiplying the country default spread with a ratio of the standard deviation of the equity market divided by the standard deviation of the country bond market. Adding this to the base premium would lead to a local ERP.

Equity Risk Premium = Base premium for Mature Equity market + Country Premium σ

Country Risk Premium = Country Default Spread *

σ

Risk-free rate

For risk-free rate, ING uses the 3-month average yield for the 10-year US Treasury Bond (UST). In specific a US dollar denominated long-term local bond is not used, as the country default risk has already been incorporated in the Country risk premium

Equity Risk Premium

According to ING; ‘The Equity Risk Premium is measured as the extra return that equity holders expect to achieve over risk free assets on average. The consensus method of estimating the ERP is to use historical data, assuming that the best indication of future market developments is actual historic performance, taken over a long period of time to reduce the impact of short term market fluctuations’.

ING, as many other financial institutions, considers the US stock market is having the

most detailed and reliable historical information. Therefore the S&P 500 is used as the

benchmark stock market and as a benchmark risk-free rate the 10-year US Treasury bond yields are being used.

For calculating the ERP for the US stock market, the following table, provided by Damodaran is used.

Period Stocks -/- T. Bonds

1928 – 2000 5.51%

1962 – 2000 4.52%

1990 – 2000 7.09%

Table 2.1 Historical geometric risk premia in US market Source: ING research, Damodaran (New York University)

As stated before, ING is aiming for a long period of data, to filter short-term fluctuations;

therefore the period 1928 – 2000 is used, with a corresponding ERP of 5.51%.

Country Risk Premium

To estimate the required local ERP, ING uses the country credit ratings by Moody’s as starting point. This credit rating is used to obtain a multiple for the long-term local currency issuer credit spread over the 10-year US Treasury bond. To obtain the country risk premium, the following formula is established:

Country default spread above the US risk-free rate is:

(10y US Treasury bond * multiple) – 10y US Treasury bond

In fact this comes down to a number of basis points added to the US risk-free rate and thus establishes a local risk-free rate.

The ERP of a country is expected to be greater than its default spread; therefore the country default spread should be multiplied by the relative equity market volatility for that market. Following Damodaran (2003), this is obtained by:

Standard deviation in country equity market Equity market volatility =

Standard deviation in country bond market

ING uses the average equity market volatility for emerging markets, which is 1.5 times.

Moody’s Country Credit Rating

An essential part of this calculation is Moody’s credit rating of a specific country.

Moody’s defines these rating as follows on www.moodys.com: “Moody's long-term

obligation ratings are opinions of the relative credit risk of fixed-income obligations with

an original maturity of one year or more. They address the possibility that a financial

obligation will not be honoured as promised. Such ratings reflect both the likelihood of

default and any financial loss suffered in the event of default.’ For each country will be

described what the rating implies, the added numbers (1, 2, 3) indicate how high a rating

ranks in it’s A to C rating category, e.g. A1 is less risky then A2.

The rating reflects a country’s default risk, but modifies this by other elements, for instance, political and currency stability, its budget and trade balances (see Appendix 2, interview Remco Vergeer).

2.2 Application

In this paragraph the country risk premia for our sample will be calculated. To convert this premium to ERP, it needs to be augmented with the ERP for mature markets, which is shown in paragraph 3.3. The credit rating used is retrieved from Moody’s as per 3 March 2006. In Appendix 1 the calculation made by Aswath Damodaran of New York University (NYU) is shown for selected countries. ING uses this exact model, only they explicitly mention a multiplier. This multiplier can be seen as a substitute for beta- coefficient and is also easy to derive from Damodaran.

2.2.1 China

In this paragraph the country risk premium for China will be calculated. Decisive factor here is the China government bond rating, as this makes the premium country specific.

For China the A2 rating indicates that it’s considered upper-medium grade and subject to low credit risk.

Since Moody’s rating is applicable to bonds, the outcome needs to be adjusted for the equity markets. Equity markets have the tendency to be more volatile then bond markets.

In this particular case, Damodaran estimates the emerging equity market to be 1.5 times more volatile then the bond market. In the following table is shown how China’s country risk is calculated by following Damodaran.

CHINA COUNTRY RISK PREMIUM

ASSUMPTION

a) 10 –year US Treasury 4.8% 29/03/2006

b) China Gov't bond rating A2 Moody's

c) A2 / UST Yield Multiplier 1.17x Average spread for A2 over UST - A. Damodaran (NYU)

d) China country default spread 0.80% (a) * (c) - (a) e) Equity market volatility 1.5x

Average for emerging markets - A. Damodaran (NYU)

China country risk premium 1.20% (d) * (e)

Table 2.2 China country risk premium

2.2.2 Taiwan

In line with the China calculation, here the Taiwan government bond rating makes the premium country specific. Apparently investing in Taiwan is not riskier then investing in the US, hence the country risk premium is zero. The Aaa rating indicates that Taiwanese bonds are judged to be of the highest quality, with minimal credit risk. This would imply Taiwan is not an emerging market. However Moody’s does make a small distinction, Taiwan is rated Aaa3, so it’s considered to be in the lower range of the highest grade. In the next table is stated how the country risk for Taiwan is established by using Damodaran.

TAIWAN COUNTRY RISK PREMIUM

ASSUMPTION

a) 10 –year US Treasury 4.8% 29/03/2006

b) Taiwan Gov't bond rating Aaa3 Moody's

c) Aa3 / UST Yield Multiplier 1.00x Average spread for Aaa3 over UST - A.

Damodaran (NYU) d) Taiwan country default spread 0.00% (a) * (c) - (a)

e) Equity market volatility 1.5x Average for emerging markets - A. Damodaran (NYU)

Taiwan country risk premium 0.00% (d) * (e) Table 2.3 Taiwan country risk premium

2.2.3 South Korea

Here the calculations for South Korea are made. The A3 rating implies that South Korean government bond ratings are in the low end of the upper-medium grade and are subject to low credit risk. The next table shows how a country risk premium for South Korea can be derived by using Damodaran’s model.

SOUTH KOREA COUNTRY RISK PREMIUM

ASSUMPTION

a) 10 –year US Treasury 4.8% 29/03/2006

b) South Korea Gov't bond rating A3 Moody's

c) A3/ UST Yield Multiplier 1.19x Average spread for A3 over UST - A. Damodaran (NYU)

d) South Korea country default spread 0.90% (a) * (c) - (a) e) Equity market volatility 1.5x

Average for emerging markets - A. Damodaran (NYU)

South Korea country risk premium 1.35% (d) * (e)

Table 2.4 South Korea country risk premium

2.2.4 Philippines

For the Philippines the country credit rating is B1, which implies that the government bonds are considered speculative and are subject to high credit risk. This becomes clearer, when looking at the following table, which results in an eventual country risk premium of 6%, which is very high.

PHILIPPINES COUNTRY RISK PREMIUM

ASSUMPTION

a) 10 –year US Treasury 4.8% 29/03/2006

b) Philippines Gov't bond rating B1 Moody's

c) B1 / UST Yield Multiplier 1.83x Average spread for B1 over UST - A. Damodaran (NYU)

d) Philippines country default spread 4.00% (a) * (c) - (a)

e) Equity market volatility 1.5x Average for emerging markets - A. Damodaran (NYU)

Philippines country risk premium 6.00% (d) * (e) Table 2.5 Philippines country risk premium

2.2.5 India

India is included in this thesis, because it is located outside South East Asia and it is one of the fastest growing countries in the world. After extensive analysis of the selected countries some differences may emerge. The credit rating of India is Baa and is therefore regarded by Moody’s as subject to moderate credit risk. The risks are considered medium-grade and as such may possess certain speculative characteristics. According to Damodaran, the following table reflects the country risk premium of India.

INDIA COUNTRY RISK PREMIUM

ASSUMPTION

a) 10 –year US Treasury 4.8% 29/03/2006

b) India Gov't bond rating Baa3 Moody's c) Baa3 / UST Yield Multiplier 1.56x

Average spread for Baa3 over UST - A.

Damodaran (NYU) d) India country default spread 2.70% (a) * (c) - (a) e) Equity market volatility 1.5x

Average for emerging markets - A. Damodaran (NYU)

India country risk premium 4.05% (d) * (e)

Table 2.6 India country risk premium

2.3 Criticism of the model

In this paragraph the assumptions on which the ING model is based will be analysed. The outcomes will not be analysed until chapter 4, because then all models will be compared.

Each factor that establishes the model will be dealt with separately. Before analysing the various factors that establish the model, an overview of the ERP per country is given, when using the ING model.

Country Country risk premium Equity Risk Premium

China 1.20% 6.71%

Taiwan 0.00% 5.51%

South Korea 1.35% 6.86%

Philippines 6.00% 11.51%

India 4.05% 9.56%

Table 2.7 Country risk and equity risk premium

(1) ERP in a mature market is 5.51% as mentioned in 2.1

Risk-free rate (a)

ING uses the 3-month average yield for the 10-year UST, since the country default risk has already been incorporated in the country risk premium.

Using a UST is correct, as a local dollar denominated bond would lead to double counting the country risk. This risk is already reflected in the credit rating and the average emerging market volatility.

However, the fact that a long-term bond is chosen seems odd. Through interviews held with various fund managers the thought emerged that Asia has very different economical cycles than developed countries. In Asia, it seems, every 2 to 3 years a crisis occurs, whether an earthquake in Kobe, SARS in Hong Kong or bird flu in various countries.

Therefore it’s naïve to use a long-term bond. A 2 or 3 year UST would reflect the risk better. At rates per 29 March 2006 this would lead to a minor change (4.79% in stead of 4.78%). How small the difference may be, for valuation purposes it might have some impact, especially since the effect becomes greater due to the multiplier used.

Government bond rating and default spread (b), (c), (d)

Moody’s is a reliable and up-to-date source for retrieving country credit ratings. The practical use of this factor is very high, since country default risk is already reflected in the rating.

By insertion of the multiplier(c) the models becomes clearer, but it doesn’t add anything.

Since Damodaran gives the amount of basis points one has to add as country default spread, the multiplier as such is useless.

By using Damodaran’s figures, two other pitfalls emerge:

1. The data is not as up-to-date as Moody’s

2. The number of basis points added may not reflect the default risk entirely correct

Ad 1. Moody’s is a commercial enterprise whose quality of data is key to survive, for

Damodaran it’s not. Although his models are widely used and his academic studies

critically acclaimed, the urge to update his tables is less stringent. Therefore it seems

risky to rely on this.

Ad 2. The constitution of the number of basis points is not entirely random or arbitrary, but might not be accurate enough. Every country that has B1 rating gains 400 bp and every B2 rating gains 500 bp. A 100 bp difference for a one level lower rating is harsh, because two countries with the same rating are not interchangeable. For instance China and Cyprus both have an A2-rating, therefore these countries both get an 80bp spread added.

Equity market volatility (e)

According to Damodaran the 1.5x multiplier is to correct bond market for equity volatility. Bond markets tend to be less volatile than equity markets, therefore the credit rating, which is for bonds, needs to be adjusted, so it is compatible with the equity risk premium.

The correction Damodaran carried through is well justified in general. ING considers it to be only a multiplier for emerging markets, which Damodaran does not. The fact that ING only uses it for emerging markets, is due to the fact she thinks that emerging stock markets tend to be more volatile in respect to their local bond markets than in developed ones. Although this is a well thought assumption, it is a too general thought to expect the 1.5 times multiplier will be applicable to every emerging stock market. The gap between the default spreads of South Korea and the Philippines is huge and therefore it seems too easy to use an average multiplier.

2.4 Conclusion

In this paragraph will be outset what the stronger and weaker elements of the ING model are.

Strong points of the model are in the way country risk is incorporated. By using the country credit rating, risk is reflected well and in combination with a UST it is counted only once. The data used is sometimes very reliable (10-year UST, Moody’s credit rating), but also sometimes a bit questionable (Damodaran).

Also, the method of adapting this model to emerging markets is too easy. Using a 10-year UST doesn’t comply with the short economical cycles in emerging markets and using a 1.5x multiplier for each market seems too arbitrary.

When obtaining the actual ERP, ING uses the geometrical risk premium over the period 1928 – 2000. By using a longer horizon, the data should become more reliable. For emerging markets it not always appropriate to use local historical data, as this often is unreliable or not available. Since the US ERP is adjusted for country specific risk, it is correct to use this data. Using a local ERP augmented with a country risk premium would inevitably lead to double counting risk.

The fact that this model is hybrid of historical values (US ERP) and a forward looking

measure (country credit rating), makes it useful for emerging markets.

3. Cost of Capital in Emerging Markets Models

Since Cost of Capital is such an important topic in finance, a vast array of models and theories have been constructed through the years. Some of them proved to be very robust, others did not. Often the initial models were revised through time by the original author, some were used as a base by new authors.

In this chapter various theories will be analysed and afterwards the most adequate models for this thesis will be selected. First some general comments, especially suitable for developed markets, will be dealt with. However the core of this chapter are the models developed especially for emerging markets. Therefore, in paragraph 4.4 the answer to sub-question 2 ‘Which other methods are suitable for estimating the ERP in emerging markets?’ will be given.

3.1 Description General Assumptions

The purpose of this paragraph is to identify some contrasts between developed and emerging market models. The objections that appear here are also valid for other models in the next paragraph and therefore mentioned in advance. Two approaches are chosen, as their flaws also can be seen in certain models in paragraph 4.2.

Movements in developed markets are relatively easy to forecast. The predictability of these markets is due to:

- Reliable historical data

- Predictable economical cycles - Transparency of local market - Political stability

When one wants to estimate the ERP in developed markets, a simple model base on historical values can be used. As will be shown in the following analysis, the models primarily differ on timeframe used and how the mean value is established.

Damodaran (1999, 2003), Mehra & Prescott (2003) constructed the fundaments of modern ERP models with the following theoretical models:

- Historical Premium Approach - Modified Historical Premium

These models will be dealt with shortly as they are not widely used in practice, but could not be omitted since they were the fundaments on which modern theories were based.

Historical premium approach

Damodaran (2003) considers historical premiums only applicable to developed markets.

When investors earned, say 6%, by investing in stocks rather than bonds for the past

years, it is reasonable to think this will also happen in the future, especially if the 5% is

earned every year over a long period. Damodaran shows this by an example regarding the US market.

Annualised Std deviation in stock returns Standard error in the Risk Premium =

√ Number of years of data in sample

For the US the standard deviation in stock returns is 20% and in this example 75 years of reliable able data is available. This leads to a standard error of (20%/ √75) = 2.31%, where 5% is the threshold to consider a calculation valid

In emerging markets the standard error rapidly grows, due to high volatility the standard deviation might be larger than 20%. But the real impact is made by the lack of reliable data. Where some academics even consider the US data to be reliable up to 100 years, for some emerging markets only 10 years are available. Even with a short calculation where the input is not very drastic. When using the same volatility as in the US, 20%, but only 10 years of data, the standard error becomes 6.32%; therefore this approach should be rejected. With higher volatility the standard error only becomes larger, which makes the model useless for emerging markets.

Modified Historical Premium

Since the Historical premium approach doesn’t apply to emerging markets, the model needs to be modified. For this the ERP is calculated as follows:

ERP= Base Premium for Mature Equity Market + Country ERP

The Country’s ERP can be calculated by looking at the volatility of the country’s equity market. Volatility can be measured by looking at the standard deviation of an equity market. For emerging markets you use a standard deviation relative to the US market:

Standard Deviation Country X Relative Standard Deviation Country X =

Standard Deviation US

When the relative standard deviation is multiplied with the US ERP, the total ERP for the market is calculated. After subtraction of the US ERP, the country ERP remains.

According to Damodaran (2003) there are two objections:

1. Firstly, two markets with different structures and liquidity are compared on volatility.

This is considered dangerous, because very risky emerging markets can have low standard deviations, due to the illiquidity of the market. This leads to severe understating of that market.

2. As Second, but less important, objection is the fact that two different currencies are

compared. The relative standard deviation will only give a true figure if the emerging

market currency is converted in US dollar.

The first argument implies that this model should not be used. Therefore in the next paragraph a look will be taken on models especially designed to overcome these problems.

Concluding from this paragraph can be said that a short time frame isn’t likely to provide valid results in emerging markets, which makes historical data using models often not valid. Also, markets are not perfect interchangeable and therefore cannot be easily compared with each other.

3.2 Description Emerging Market Models

In this paragraph there will be an extensive analysis of emerging market ERP models.

They often use forecasted data and consist of more elements, because emerging markets are far more complex than developed ones. As a guideline for this paragraph, Harvey’s article “12 Ways to Identify the International Cost of Capital” (2005) is used. This article provides a proper overview, but it is not extensive enough and therefore it is augmented with other models.

It must be noted in advance that Harvey describes 12 models to calculate the Cost of Capital, the outcome is that his International Cost of Capital and Risk Calculator (IICRC) returns the most valid rates. This model is not included in the list as it’s not widely available, so the practical use is negligible. Another argument for not attaching much value to Harvey’s overview, is the fact that the IICRC is a real cash cow for Harvey, therefore he might be somewhat biased in his findings.

Each model will be analysed on its fundaments, strengths and weaknesses and the level of being especially useful for emerging markets. Comments by Harvey (2005) will be mentioned when useful. Finally a recommendation will be given for whether or not to use this model for the selected emerging markets. This choice will be elaborated on in paragraph 4.3.

1. Identical Cost of Capital

In fact the Identical Cost of Capital is not a specific model but basically it’s the proposition to calculate an emerging market ERP the same way as for developed markets.

This might score high in practical use when dealing with a lot of different countries, but is a too simple method, as countries are far from identical at all. One cannot consider risk to be equal in every country, as this would lead to false valuations.

Strengths and Weaknesses

Harvey (2005) explains with a simple example that this method would destroy value. If

every country is perceived risky, a high discount rate is used and less risky countries

would be valued too low and therefore an investment decision would be rejected. When a

low discount rate is used, i.e. all countries have low risk, a very risky project would be

valued far too high and it thus would lead to a bad investment decision.

As show in the previous paragraph, using developed markets models for emerging markets the results are not valid, as they are based on assumptions that don’t hold for developing markets.

2. World CAPM and Multifactor Model

The World CAPM and the Multifactor Model are put together under one heading as they are based on the same assumptions. The Multifactor Model by Drew and Veeraraghavan (2003) is a slightly elaborated version of the CAPM.

The CAPM is widely acclaimed for it high practicability. Developed by Sharpe in 1964, it’s still common used. The model was initially developed for the US market and appeared to hold. For other developed countries, it has also proved to be a robust measure for risk that was not able to diversify through a well-diversified portfolio.

CAPM = R

= R

+ β(R

- R

)

The difference between the market return R

and risk-free rate R

is the ERP. β is used to make the risk specific for a company. Where β is greater than 1, this means that the stock of a company would rise sharper than its index and vice versa.

The expected market return is defined by the volatility of the market, further explained by the Sharpe ratio. This ratio of an equity investment measures the excess return (above a risk free rate) earned per unit of risk taken and is given by

Sharpe Ratio = Excess Return / σ (σ = volatility as a measure of risk)

In an efficient global market where rational investing occurs, the excess return per unit of risk taken should not differ for two different equity investments. The expected excess return of a market is its ERP thus for UK and US markets

ERP

/ σ

= ERP

/ σ

Æ ERP

= ERP

*( σ

/ σ

).

The relative volatilities of a market can provide a country risk premium. For US take the historical spread between 10-yr stock market returns and government bond yield. This method was, and still is, used for valuations concerning developed markets.

From this point on, two scientific directions appeared in order to make this model applicable to every single country in the world. First it will be dealt with ‘World CAPM’, and second it will be elaborated on the use of CAPM in emerging markets.

Strengths and Weaknesses

First, when applying CAPM to the world market, two things are highly important:

1. (R

- R

) is the expected return of a well-diversified portfolio with stocks from all over the world in excess of the risk-free rate

2. β is here the measure to make the ERP country specific

Harvey (2005) indicates that the second assumption does not hold when applied to the world market. His evidence showed that not beta-coefficient, but variance explained the return across world markets.

Still Harvey believes in the fact that there is something as a global ERP. During interviews with portfolio managers (see Appendix 2) the belief originated that this does not exist, because unreliable and inadequate historical data, liquidity issues and irrelevant history make it impossible to compare completely different markets and pool them together.

Second, other evidence against the use of the World CAPM as well as the following Multifactor Model, when applied to emerging markets, is incorporated in CAPM’s assumptions of efficient capital markets and complete integration of markets (foreign investors can freely invest in local market and vice versa).

These assumptions are strongly violated in emerging markets. First of all, efficient capital markets, seldom, if not never, exist. If capital markets were efficient, there would never be any arbitrage opportunities, or, in other words, all investors would have exactly the same information at the same time. Although this is very rare to happen, it’s more likely to occur in complex emerging markets than in developed ones.

The second assumption is even more applicable to emerging markets. When a market is perfect integrated, foreign investors can freely invest in the local market and vice versa.

Examples from China and Taiwan show this is not correct. In both countries the maximum share a foreign investor can hold in the local markets is capped at 25% or respectively 50%, therefore this assumption doesn’t hold.

A stronger point of World CAPM is the mean risk premium. As seen in the definitions overview, arithmetic or geometric mean can be used. Siegel (1999) criticises CAPM, because it uses an arithmetic mean. Siegel states that for long time horizons, the geometric mean needs to be used. However, as stated in chapter 2, the cycles in emerging markets are relatively short compared to developed ones, so for this research a short time horizon is desirable. Although this works in favour of World CAPM, it does not outweigh the weaknesses

The Multifactor Model is based on the same assumptions as World CAPM. Therefore the criticism is the same and only the different factors of this model will be analysed in the following part.

Multifactor Model

The Multifactor Model is explained by Drew and Veeraraghavan (2003). According to

their article, a single risk factor (i.e. beta-coefficient) is not sufficient to explain stock

returns. They pledge for making the distinction between small-mid and large cap stocks,

as well as the book-to market equity. They conclude for several emerging markets (Hong

Kong, Malaysia, Philippines) that small firms, with high book-to market equity, generate

higher returns than large firms, with low book-to-market equity. Although they state that

the Multifactor Model captures a premium, which is ignored by CAPM, they don’t reject CAPM before the Multifactor Model is thoroughly tested.

R

– R

= a

+ b

(R

– R

) +s

SMB

+ h

HML

+ ε

R

is the average return of a certain portfolio (S/L, S/M and S/H; B/L, B/M and B/H). R

is the risk-free rate observed at the beginning of each month. SMB is the difference each month between the return on a portfolio of small stocks and the portfolio of big stocks;

HML is the difference each month between the return on a portfolio of high book-to- market equity stocks and the return on a portfolio of low book-to-market equity stocks.

The factor loadings b

, s

and h

are the slopes in the time-series regression.

Strengths and Weaknesses

The Multifactor Model is based on the same assumptions as CAPM and even with the additional factors it’s not especially designed for emerging markets. Moreover, the authors also don’t consider the model to be useful yet.

3. Segmented/Integrated Model

Bekaert and Harvey (2003) describe the Segmented/Integrated model, which is especially designed for emerging markets, but also holds in developed ones.

Emerging market returns are highly non-normal and highly volatile. The current liberalization process in emerging markets had led to a small increase in correlation with developed market indices and a small decrease in dividend yield. The latter could lead to a decrease in cost of capital or an improvement in growth opportunities. The model is as follows:

CC= w[world CC] + (1-w)[local CC]

Where CC is Cost of Capital, and w is determined by variables that proxy for degree of integration, like size of trade sector and equity market capitalization to GDP. To calculate world Cost of Capital, a global risk-free rate plus β

x world risk premium is used and for the local Cost of Capital, a local risk-free rate plus β

x local risk premium is used. So in fact it’s a combination of 2 CAPM models.

Strengths and Weaknesses

Since this model is based on CAPM, the same criticism regarding efficient and integrated capital markets is applicable. However, the correction in CAPM makes it more suitable than the World CAPM and Multifactor Model. In this thesis the model gains weight, as Taiwan and China don’t have perfectly integrated capital markets.

4. Bayesian Approach (Ibbotson Associates)

The following model by Ibbotson and Chen (2001) is very much influenced by the fact

that investors can expect the same or just slightly lower returns over time in respect to