Upload
arvind-mahandhwal
View
209
Download
7
Embed Size (px)
Citation preview
“Risk and Return Relashionship in securities Market.”
A report submitted to IIMT, Greater Noida on partial fulfillment of full time
Postgraduate Diploma in Business Management.
SUBMITTED TO: SUBMITTED BY:
Director Raman Chugh
IIMT ENR No. 11105
Greater Noida Batch: 11th
ISHAN INSTITUTE OF MANAGEMENT & TECHNOLOGY
2, KNOWLEDGE PARK-1, GREATER NOIDA
Website-www.ishanfamily.com
E-mail: [email protected]
ACKNOWLEDGEMENT
Here I take the opportuinity to express my gratitude to all
of them, who in some or the other way helped me to accomplish this
project. The project study cannot be completed without their
guidance, assistance, inspiration & cooperation.
For successful accomplishment of task apart from hard
work the most requisite is the right direction & guidance. And for a
student these become the major part for the study. In Sharekhan
this right direction and guidance is provided by my guide and all the
executives of the concerned department, in the form of necessary
information & exhibits that gave me a great help in completing my
work.
First of all I would like to thanks Dr. D.K.Garg, Chairman
Ishan Institute of Management & Technology and Mr. G.K.Sinha
Director and Head of Training & Placement Cell, Ishan Institute of
Management & Technology for giving me this opportuinity of doing
the summer training project in Sharekhan ltd.
A special thanks to Deepak Gupta (Asistannt Manager)
under whom I conducted this study, for his able guidance in getting
my project completed.
I am indebted to my parents because of whose help I have
been able to carry out this work successfully. I am also thankful to
my friends who directly or indirectly helped me lot.
The study has indeed helped me to explore more knowledgeable avenues related to my topic and I am sure it will help me in my future.
Raman Chugh
PGDBM (Finance) Enr. No. 11105 IIMT, Greater Noida.
DECLARATION
I, Raman Chugh student of PGDBM 2nd Semester in Ishan
Institute of Management & Technology, ENR. NO. – 11105, hereby
declare that, this Project Report under the title “Risk and Return
Relashionship in Securities Market” is the record of my original work
under the guidance of Mr. Deepak Gupta(Assistant Manager), Sharekhan
ltd. This report has never been submitted anywhere else for award of any
degree or diploma.
Place: Greater Noida Raman Chugh
Date PGDBM (Finance) Enr. No. 11105 IIMT, Greater Noida.
EXECUTIVE SUMMARY
Risk and Return in Securities market are two sides of one coin. Securities market is
totally based on this concept of risk and return, without risk returns can not be there,
because the share market is a type of speculation and speculation always carry risk with
it. It depends upon investors how they diversify the risk, how they minimize the risk by
creating an optimum portfolio of securities. RISK IN SECURITIES management process
is to maximize earnings in the context of an acceptable level of risk. With risk defined as
uncertainty, a key component of this process is delineating the possible range of
outcomes and their root causes. As new circumstances arise, a securities lender is
prompted to study existing data in order to achieve a greater comprehension of potential
situations a client may face.
The risks inherent in any securities lending program — including market, credit,
liquidity, operational, legal, and regulatory — are all relevant to the principal common
objectives shared by participants, namely stability of income and preservation of
principal. Of these, market and credit risk most readily lend themselves to quantification
and modeling due to the greater frequency and depth of the data available. The following
narrative focuses on the cash-collateralized securities lending transaction. The risks to
participants who accept non-cash collateral will be addressed in a later article that will
deal, at length, with credit risk modeling efforts.
The relationship of market value to purchase price is captured by its net
Asset value (NAV). The first Digest article noted that a security with a yield exceeding
the current market rate of interest for an investment with a similar maturity structure and
credit quality will be valued in excess of par, while a security with a yield lower than the
current market rate of interest for a similar investment will be valued below par. In the
latter scenario, an investment might have seemed attractive at the inception of the
transaction, but with an increase in interest rates, the original investment is now trading at
a value below the purchase price.
In effect, if the security is not sold prior to maturity, it will only incur an opportunity cost
or a foregone opportunity to earn the current yield. As such, NAV is an indicator of how
the portfolio will perform relative to the market going forward, and increased volatility of
this measure is suggestive of a build-up of risk.
Table of Contents
A. Company Profile
Chapter 1 About Sharekhan
Background of ShareKhan
4Ps in Sharekhan: Product Price Place Promotion
Marketing Strategy of Sharekhan
Competitors of Sharekhan: History of Company Financial Summary Balance Sheet Shareholding of company Top 5 mutual funds that own the company Management of the company Product of the company
Government policy
Taxation Aspect
National & International Image
Share Market Position
Management Theory in Sharekhan
Major Problems
Achievements
Future Prospects
B. Project Work
Chapter 2 Risk and Returns in Securities Market
Risk in Securities Market
Returns in Securities Market
Chapter 3 Diversification of Risk
Risk an Return- Portfolio Theory & Asset Pricing Model
Fixed Income – An Evolution in Risk and Return
Chapter 4 Related Concepts
Exploration Risk and Return in Global Equity Markets.
SEBI group on Securities Market Risk Management
Chapter 5
Calculation of Risk Beta
Nestle (I) Ltd. Jindal Steel Ltd. Sh. Cements
Findings and conclusion
Recommendation
Bibliography
Annexure
CHAPTER. 1
SHAREKHAN DEMAT
Dematerialisation and trading in the demat mode is the safer and faster alternative to the
physical existence of securities. Demat as a parallel solution offers freedom from
delays,thefts, forgeries, settlement risks and paper work. This system works through
depository participants (DPs) who offer demat services and the securities are held in the
electronic form for the investor directly by the
Depository. Sharekhan Depository Services offers dematerialisation services to
individual and corporate investors. Sharekhan have a team of professionals and the latest
technological expertise dedicatedexclusively to our demat department, apart from a
national networkof franchisee, making our services quick, convenient and efficient.At
Sharekhan, ourcommitment is to provide a complete demat solution which is simple,
Safe andsecure. Demat as a parallel solution offers freedom from delays, thefts,
forgeries,settlement risks and paper work. This system works through depository
participants (DPs) who offer demat services and the securities are held in the electronic
form. As per the guidelines of SEBI the trading through different stock exchanges can
bedone only when the shares are in DMat form. DMAT shares means the shares are notin
material form they are converted into electronic form. The transfer of shares in trading by
stock exchanges is done with the help of only DMAT accounts of the clients where the
shares of the clients are entered as in DMAT form. So with so much relevance of
DMATaccounts sharekhan also provide DMAT services to the clients
SHAREKHAN PRODUT OFFERING
The product offered by sharekhan ltd. is its DMAT Account and its services.
Sharekhan offers two types of DMAT accounts.
These are:
1. Speed Trade
2. Classic.
SPEED TRADE
The speed trade account is DMAT account which is mainly for the large scale share
trading Those persons whose main business is only of share trading, who are the dealers
in the share market and the the share trading is not for investments instead of that it is for
their operational activities.
Speed trade customers are different customers, so the services provided to them also
different as they are large clients for the company they are also provided many facilities,
they purchase and sell their shares in bulk and hence pay big amounts of brokerage to the
company so their brokerage charges are very less,
The company also take margin money cheques from these clients.Speed Trade clients are
different as per their sale and purchase of securities in market in bulk. The company
provide special research reports and assistance to the speed trade clients.
Speed trade clients can also access the relevant information regarding the trading from
the branch offices of the company relevant for them.
They can ask for the minimum brokerage charges, transaction charges to the company.
The major services available to them with the d mat account are as follows
Online share trading
Off line share trading
Researh Reports of the company.
Depositary services
Exposure
Portfolio services
Back up services
Derivatives investments.
Mutual Funds Services
Company information
Market inquiry
Price quoting
Major news affecting the market.
IPO Services
Technichal Services
Newsletters
Discussions
Fundamentals
Alerts
Commodities
Commodities Futures
Credit etc.
CLASSIC ACCOUNT
Classic is one of the major product offering by Sharekhan ltd. The company’s main
clients are of Classic DMAT account. These clients are mainly Individuals, HUFs ,
Corporates and others trade in shares just for investment and speculation apart from their
operating activities, i.e. whose main business and area of operations are not share trading.
The persons who want to open Classic Account in the share khan requires the following :
1. Two passport size photographs.
2. One photocopy of PAN card.
3. The Bank Pass Book First page photocopy
4. An agreement book containing agreement with Sharekhan, NSE, BSE and other
regulatory bodies as in prescribed format with stamp papers of Rs.20.
5. The adequate fees.
The companies mainly clients are classic account holders, the company provide best
services to them. The company also take margin money cheques from these clients.The
brokerage etc. are charged by the company from the client are according to agreement
done with the client and hence it depends upon negotiating with the clients, the company
often charges minimum brokerage from the clients who trade in bulk.
The main services available to the cutomers of the classic account are
Online share trading
Off line share trading
Researh Reports of the company.
Depositary services
Exposure
Portfolio services
Back up services
Derivatives investments.
Mutual Funds Services
Company information
Market inquiry
Price quoting
Major news affecting the market.
IPO Services
Technichal Services
Newsletters
Discussions
Fundamentals
Alerts
Commodities
Commodities Futures
Credit etc.
The above are discussed as follows:
Sharekhan ltd. does its business mainly with the help of the online and so it does its
business mainy through the its website. www. Sharekhan.com . The company provide all
above services through the links available in the website the main links available there
are:
My sharekhan
Services
Trade now
Research
News
Market Corner
Derivatives
Knowledge centre
All above links have further links explaining one by one as:
MY SHAREKHAN
Under the link my sharekhan different links are abailable there these links are mainly
My portfolio
Discussion board
My alerts
My preferences
My newsletters
Sharekhan Seminars
My portfolio
My potfplio contains the information regarding the pms ie portfolio management
services, share khan provide the main services of portfolio and under that share khan
provide the information to the respective clients about the securities in therir portfolio,
their services, their prices, their reports about future position in market, their up an low of
the day, closing etc .
The information abailable in the potfolio link is regarding the combination of securities
purchased by the company in behalf of the client of a specific amount given by the client
for such portfolio and the link contains the main information of
no. of securities in the demat account of the client,
name of the company of the securities,
price of purchasing,
current position in the market
Profit /sales till date
Total amount invested
Balance
Broketage charges
All these informations are available to the client of the company by inserting a user name
and password provided by the company.
My alerts
My alert link is provided to the customer for the alerts to customer from time to time so
that customer can never remain uninformed regarding any thing before its occurrence,
under the link my alerts customers are informed about their requirements of
documentation for the clearance of trading, their securities buy and sale information,
expectation of the shares held by them in the market, any fees or charges due by them
like
Brokerage charges
Annual maintainence charges
Cheques dishonoured
Amount payable for the funding of purchase
Amount paid for selling of shares etc.
My Newsletters
My newsletters link contains the newsletters abailable to the clients about their
information necessary and impotant to them regarding the market positions of their shares
held by them and other important news, new guidelines , necessary documents required
by the regulatory bodies,
Changes in the brokerage charges, annual maintainance charges and other important and
relevant information.
ShareKhan Seminars
Sharekhan organizes seminars across the country from time to time in order to educate
investors in various subjects related to the stock, derivatives and commodities market.
This exclusive Seminars are organized by Sharekhan for its Online Trading Customers
for FREE.
These Seminar will equip and help you take an informed investment decision. If you wish
to built a healthy investment portfolio, ensure that you do attend our Seminar, which are
designed keeping in mind your requirements and organized by porfessionals.
All these Seminars are held by sharekhan from time to time as this is the part of its
services apart from one of its services, by seminars sharekhan can attract more new
Customers and can hold the existing customers, this is also the marketing strategy
Of the company.This can be taken as one medium of advertisement by the company
SERVICES
The another major link available in the website of the company is services.
Under the link services the major links available are
Online services
PMS
Commodity
DEMAT
Share shops
Mutual Funds
These all are discussed as one by one here
Online services
The major work by Sharekhan is done online. The online services are the reason of the
fast, fair and transperant services by the company, the online services mainly is online
share trading this is elaborated as
The client is provided with a unique customer id and a password made by himself and
after registering with the id and password he can get anytime access to the trading link of
the site, aby entering the password and id. After that he can take market watch BSE,
NSE, can place order for sale , purchase, modify the order for sale, purchase of shares
and can trade in any security listed on the stock exchange whether in Equity and whether
in futures and options of shares equity. The funds transfer take place from the saving
account of the customer.
This was regarding online trading of shares only but sharekhan also provide
online trading of commodities, in the same manner the clients can trade through different
commodity exchanges like national commodity exchange etc. They can place order
according to pre defined lot of commodity in the market through the exchange.
Apart from online trading services other online services are also available in sharekhan as
the IPO service are done online, research reports are provided online to the clients.
PMS
PMS is Portfolio management services as this is also one of the main services provided
by the sharekhan .The company charges different fees from the client for the PMS. As
many people are not able to decide which shares, equities are beneficial and they don’t
want to take risk so they hire the PMS of the company under that the experts employed
by the company make a portfolio of the securities of the amount given by the investor
and risk preference. My potfplio contains the information regarding the pms ie portfolio
management services, share khan provide the main services of portfolio and under that
share khan provide the information to the respective clients about the securities in therir
portfolio, their services, their prices, their reports about future position in market, their up
an low of the day, closing etc .
The information abailable in the potfolio link is regarding the combination of securities
purchased by the company in behalf of the client of a specific amount given by the client
for such portfolio and the link contains the main information of
no. of securities in the demat account of the client,
name of the company of the securities,
price of purchasing,
current position in the market
Profit /sales till date
Total amount invested
Balance
Brokerage charges
All these informations are available to the client of the company by inserting a user name.
DMAT:
As per the guidelines of SEBI the trading through different stock exchanges can be done
only when the shares are in DMat form. DMAT shares means the shares are not in
material form they are converted into electronic form. The transfer of shares in trading by
stock exchanges is done with the help of only DMAT accounts of the clients where the
shares of the clients are entered as in DMAT form.
So with so much relevance of DMAT accounts sharekhan also provide DMAT services to
the clients .Dematerialisation and trading in the demat mode is the safer and faster
alternative to the physical existence of securities. Demat as a parallel solution offers
freedom from delays, thefts, forgeries, settlement risks and paper work. This system
works through depository participants (DPs) who offer demat services and the securities
are held in the electronic form for the investor directly by the Depository.
Sharekhan Depository Services offers dematerialisation services to individual and
corporate investors. Sharekhan have a team of professionals and the latest technological
expertise dedicated exclusively to our demat department, apart from a national network
of franchisee, making our services quick, convenient and efficient.
At Sharekhan, commitment is to provide a complete demat solution which is simple Safe
and secure. Demat as a parallel solution offers freedom from delays, thefts, forgeries,
settlement risks and paper work. This system works through depository participants
(DPs) who offer demat services and the securities are held in the electronic form .
Commodity:
Commodity market in India is today developing very fast, the commodity market deals
with the actual commodity sale purchase through different stock exchanges available for
only commodity exchange, the system of commodity market is almost the same as to the
stock market.
Sharekhan securities provide the services to commodity as its different market segment
so its charges are different from of the share market.
In commodity also the same types of services like portfolio, research reports, futures,
exposure, credit, online trading, offline trading is also available as in equity.
Mutual Funds
Share khan also provide the services of mutual funds to the client on the payment of
prescribed fees, Mutual funds are the most selling securities today in the share market
and it is attracting people very much.
A mutual fund pools together sums from individual investors and invests it in various
financial instruments. Each mutual fund has its own investment objective, which broadly
falls into two categories: capital appreciation and current income.
Suppose a mutual fund sells one million units or shares (used as synonyms in this
context) at Rs 10 a share and collects a total Rs 10 million. If the fund objective stated
investment in blue-chip stocks, the fund manager would invest the entire proceeds (less
any commissions and management fee) of that sale in buying equity shares of companies
like Hindustan Lever, Reliance Industries, Hero Honda and so on. And each individual
who bought shares of the fund would own a percentage of the total portfolio only to the
extent of money invested. The value of the fund's portfolio would depend on how the
shares of these companies perform on the stock market (given their financial prospects).
If the total market value of these companies (as reflected in the fund) increases to Rs 12
million, then each original share of the fund would be worth Rs 12 (Rs 12 million divided
by one million shares). This per share value is what is known as the net asset value
(NAV) of the mutual fund. It equals the market value of all its assets (after adjusting for
commissions, expenses, and liabilities, if any) by the number of such units or shares
outstanding.
Mutual funds over direct investment in equities
As financial intermediaries, they do not come without risk. Also when defined in terms of
our chances of losing money, the risk in mutual funds is no different than that present in
other financial instruments. Still they are relatively safer and a more convenient way on
investing. They offer quick liquidity. Most private mutual funds can be redeemed in three
to four working days, unlike a fixed deposit that is more likely to be received a month
after its maturity, or an equity share after the end of its settlement period (or depending
up on our broker). This too cuts the overall risk associated with investing, often not so
visible and hence not accounted by many investors. But the market risk or the risk that
exists due to economy-wide factors remains. And there is always the possibility that a
fund fails to stick to its pre-determined objectives or invests in securities that alter its risk
profile. In which case, the blame goes straight to the fund manager and the Asset
Management Company (AMC), which manages the mutual fund. All said and done
mutual funds still have following advantage over direct investment in equities.
Affordable
Almost everyone can buy mutual funds. Even for a sum of Rs 1,000 an investor can
invest in a mutual fund.
Professional Management
This is the biggest advantage mutual fund have over direct investment over equity. For an
average investor, it is a difficult task to decide what securities to buy, how much to buy
and when to sell. By buying a mutual fund, we acquire a professional fund manager who
manages our money. This is the person who decides what to buy for us, when to buy it
and when to sell. The fund manager takes these decisions after doing adequate research
on the economy, industries and companies, before buying stocks or bonds. Most mutual
fund companies charge a small fee for providing this service which is called the
management fee.
Diversification
According to finance theory, when our investments are spread across several securities,
our risk reduces substantially. A mutual fund is able to diversify more easily than an
average investor across several companies, which an ordinary investor may not be able to
do. With an investment of Rs 5000, you can buy stocks in some of the top Indian
companies through a mutual fund, which may not be possible to do as an individual
investor.
Liquidity
Unlike several other forms of savings like the public provident fund or National Savings
Scheme or real assets, you can withdraw your money from a mutual fund on immediate
basis.
Transparency
Regulations for mutual funds have made the industry very transparent. We can track the
investments that have been made on our behalf and the specific investments made by the
mutual fund scheme to see where our money is going. In addition to this, we get regular
information on the value of our investment.
Tax Benefits
Mutual funds have historically been more efficient from the tax point of view. A debt
fund pays a dividend distribution tax of 12.5 per cent before distributing dividend to an
individual investor or an HUF, whereas it is 20 per cent for all other entities. There is no
dividend tax on dividends from an equity fund for individual investor.
MUTUAL FUND INVESTING - RISK VS REWARDS
Having understood the basics of mutual funds the next step is to build a successful
investment portfolio. Before we can begin to build a portfolio, one should understand
some other elements of mutual fund investing and how they can affect the potential value
of our investments over the years. The first thing that has to be kept in mind is that when
we invest in mutual funds, there is no guarantee that we will end up with more money
when we withdraw our investment than what we started out with. That is the potential of
loss is always there. The loss of value in our investment is what is considered risk in
investing.
Even so, the opportunity for investment growth that is possible through investments in
mutual funds far exceeds that concern for most investors. Here’s why.
At the cornerstone of investing is the basic principal that the greater the risk we take, the
greater the potential reward. Or stated in another way, we get what we pay for and we get
paid a higher return only when we're willing to accept more volatility.
Risk then, refers to the volatility -- the up and down activity in the markets and individual
issues that occurs constantly over time. This volatility can be caused by a number of
factors -- interest rate changes, inflation or general economic conditions. It is this
variability, uncertainty and potential for loss, that causes investors to worry. We all fear
the possibility that a stock we invest in will fall substantially. But it is this very volatility
that is the exact reason that we can expect to earn a higher long-term return from these
investments than from a savings account.
Different types of mutual funds have different levels of volatility or potential price
change, and those with the greater chance of losing value are also the funds that can
produce the greater returns for we over time. So risk has two sides: it causes the value of
our investments to fluctuate, but it is precisely the reason we can expect to earn higher
returns.
We might find it helpful to remember that all financial investments will fluctuate. There
are very few perfectly safe havens and those simply don't pay enough to beat inflation
over the long run.
Investing in mutual fund – choosing the right scheme/option
Choosing a mutual fund is not an easy task with so many funds. Rarely do investors-
normal investors, who do something else for a living-have a systematic checklist of
things that they should evaluate about a fund, which they are considering buying. Here's
my blueprint for a structured approach to fund selection. There are four basic areas that
one must evaluate in a fund to decide whether it's a good investment.
Performance:
Performance comparisons must be used only to compare the same type of fund. They are
meaningless otherwise. Only when used within the same category of funds performance
numbers tell anything at all..
Risk:
Almost all investing is risky, at least those investments that get any meaningful returns.
In general it is said that the riskier a fund, the more its potential for earning high returns,
at least most of the time. However, this is a simplified view that implies that a given
amount of risk always gets the same returns. This is simply not true because not all funds
are equally well-run.
The true measure of risk is whether a fund is able to give the kind of returns that justify
the kind of risk it is taking. Evidently, this is not as easy to measure as returns. There are
a wide variety of statistical techniques that can be used to measure this. When someone
says that a fund has a good performance, it means that the fund, compared to similar
funds, performed better, given its risk level.
Portfolio:
Unlike performance and risk, portfolio is one of the 'internals' of a fund. It is internal in
the sense that the result of good, bad or ugly portfolios is already reflected in the first two
measures and it's perfectly OK to choose funds on the basis of those two measures alone
without actually bothering about what they own. My basic analysis of portfolios
measures whether a fund (I am talking about equity funds here) holds mostly large,
medium or small companies. It also looks at whether a fund prefers companies that may
be overpriced but which are growing fast or whether it prefers low-priced stocks
belonging to companies that are growing at a more gentle pace. For fixed income funds,
an analogous analysis tells one whether a fund prefers volatile but potentially high return
long-duration securities or stable and low return short-duration securities. Also, one can
analyze whether a fund prefers safer (lower returns) securities or riskier (higher returns)
securities.
Management:
Fund management is a fairly creative and personality-oriented activity. This may not be
true of some types of funds like shorter-term fixed-income funds and, of course, index
funds, but equity investment is more of an art than a science. When we are buying a fund
because we like its track record (and unless we can foresee the future, that's the only way
to buy a fund), what we are actually buying is a fund manager's (or sometimes a fund
management team's) track record. What we need to make sure is that the fund manager
who was responsible for the part of the fund's track record that we are buying into is still
there. A high-performance equity fund with a new manager is a like a new fund.
Cost:
Funds are not run for free and nor are they run at an identical cost. While the difference
in different funds' cost is not large, these can compound to significant variations,
especially for fixed income funds where the performance differential between funds is
quite small to begin with. Even for equity funds, it may not be worth buying a higher cost
fund that appears to be only slightly better than a lower cost one. There is no reason for
one AMC to have much higher costs than others, apart from the fact that it wants to have
a higher margin, or that it wants to spend more on things like marketing, which are of no
relevance to you. If an AMC wants higher returns from its business, then it must justify it
by giving you higher returns on your investments.
Cost of investing in mutual funds
In addition to loads a mutual fund also charges asset management fees, and certain other
expenses. These charges imposed by mutual funds are meant to compensate the fund for
the expenses it incurs in managing assets, processing transactions and paying brokerages.
For instance every redemption request involves not only administrative processing costs
but also other costs associated with raising money to pay off the outgoing investor.
Mutual Funds cannot, however, be arbitrary in the imposition of these charges. For
instance, regulations stipulate that the difference between the repurchase and the resale
price cannot exceed 7 per cent of sale price, and that recurring expenses cannot exceed
2.5 per cent of average weekly net assets.
Loads:
Entry Load/Sale Load
It is the charge imposed on the investor at the time his entry into the fund. Thus, the
investor has to pay for the value of the units plus an additional charge. This additional
charge is called the entry/sale load.
Exit Load/Repurchase Load
It is the charge imposed on the investor at the time of his exit from the scheme.
Operationally, therefore, the mutual fund will pay back to the investor the value of the
units reduced by the charge levied on exit.
Contingent Deferred Sales Charge
A mutual fund may not want to charge an exit load in all the cases. In such a case the
mutual fund may impose charges based on the time of withdrawal. Thus, a fund desirous
of long-term investors may stipulate that the exit charge will keep reducing with duration
of investment. Such a charge is called Contingent Deferred Sales Charge. The asset
management company is entitled to levy a contingent deferred sales charge for
redemption during the first four years after purchase, not exceeding 4% of the redemption
proceeds in the first year, 3% in the second year, 2% in the third year and 1% in the
fourth year. In order to charge a CDSC the scheme has to be a no load scheme as per the
regulation laid down by SEBI. The idea behind charging CDSC is the recovery of
expenses incurred on promotion or distribution of the scheme
Switchover/Exchange Fee
It is the fees charged by a fund when the investor decides to switch his investment from
one scheme of the fund to another scheme from the same fund family.
Recurring Expenses:
Apart from loads, mutual funds also charge some other expenses. Even here regulations
stipulate the ceiling on each head. Some of the fees charged by the fund are:
o Investment Management & Advisory Fees - As the name explains this is
meant to remunerate the asset management company for managing the investor's money.
o Trustee Fees - is the fees payable to the trustees for managing the trust.
o Custodian Fees - is the fees paid by the fund to its custodians, the
organization which handles the possession of the securities invested in by the fund.
o Registrar and Transfer Agents Charges - is the fees payable to the registrar
and the transfer agents for handling the formalities related to the transfer of units and
other related operations.
TRADE NOW
Trade now is one of major and important link in the website of the sharekhan. By this
link the customer who have a genuine ID and password can sale and purchase their
securities through online.
There are mainly two segments of Trade now:
1. Fast Trade
2. Classic
RESEARCH
Sharekhan ltd. also provide research reports to its clients about the postion of market
For example which securities can fall and which can be up in the market , at what time
Which security should sale or purchase etc.Sharekhan understand that every investors
needs and goals are different. Hence it provide a comprehensive set of research reports,
so that customers can make the right investment decisions regardless of their investing
preferences.Its research mainly includes:
1. Fundamental
2. Technichal
3. Mutual funds
NEWS
This is the important service of Sharekhan limited , as when the customer logon to the
website of the company he don’t require to go another site for the news section,
Sharekhan’s experts edit the whole day news update the section for the whole day and
Present in simpler form in front of the investors. The major sections of news link are as
follows:
Top Stories
Sectoral news
Economy
Finance
Press Digest
Mutual Fund
IPO
Market Today
MARKET CORNER
In the link market corner all information regarding
Charting, Announcements,FIs activities, Reseults, Price Watch,Company information,
IPO, Mutual Fund etc.
DERIVATIVES
Derivatives services by sharekhan limited includes the following
Equity Futures and Options
Commodity Futures
KNOWLEDGE CENTRE
The link knowledge centre includes the following further links which are provided for
Increase in knowledge of the clients and prospective clients of the organsiation. The
Main links are here as follows;
School
Stock Trivia
Opinion polls
Book Reviews
FAQs
PRICE
The price is very important for any organization specially which there is tough
competition in the market. The prices charged by sharekhan are not so low so that
thebrand image in mind of customers and potential clients do not fall , and it is not so
Much high that it can be easily beatable by the competitors.
DMAT OPENING CHARGES
Online and Offline Services
The maximum charges under this are Rs. 750 [including the fees payable to NSE, BSE
and other regulatory bodies]The minimum charges under this are Rs. 375 [including the
fees payable to NSE, BSE and other regulatory bodies] this is in case of corporate deal.
With the cheque of Rs. 5000 as margin money the fees for this is charged as Rs. 500.
These are for life time.
Offline Services Only
For the offline services only the charges are cheque of Rs. 460. These are for life time.
BROKERAGE CHARGES
Maximum Brokerage Charges are on intra day 10 paisa.
Minimum Brokerage Charges are on intra day 5 paisa.
Maximum Brokerage Charges are on delievery 50 paisa.
Minimum Brokerage Charges are on delievery 25 paisa.
MAINTENANCE CHARGES
No maintenance charges for first year, from 2nd year it is Rs. 300 p.a.
PLACEShare Shops
Get everything you need at a Sharekhan outlet!
Customers have to do is walk into any of Sharekhan 588 share shops across 213 cities in
India to get a host of trading related services – the friendly customer service staff will
also help them with any account related queries they have.
Share Khan ltd. has many branches all over the India through where the company
operates in different areas and The company has a long chain of frenchisee also so the
network of sharekhan ltd has spread across all over the country. Share khan shops are
available many places in India and customer can take assistance from any of these for the
product and service availing. Sharekhan outlet offers the following services:
Online BSE and NSE executions (through BOLT and NEAT terminals)
Free access to investment advice from Sharekhan's research team
Sharekhan ValueLine (a fortnightly publication with reviews of recommendations, stocks to watch out for etc
Daily research reports and market review (High Noon, Eagle Eye)
Pre-market Report (Morning Cuppa)
Daily trading calls based on technical analysis
Cool trading products (Daring Derivatives, Trading Ring and Market Strategy)
Personalised advice
Live market information
Depository services: Demat and Remat transactions
Derivatives trading (Futures and Options)
Internet-based online trading: SpeedTrade, SpeedTradePlus
The places where Shops of sharekhan are mainly
Agra
Ahmedabad
Allahbad
Ambala
Amritsar
Anand
Banda
Bangalore
Baroda
Belgaum
Bhopal
Bhubneshvar
Bijapur
Bhopal
Bhubnshvar
Calcutta
Chennai
Coimbtore
Cuttack
Dadri
Dehradun
Dispur
Faridabad
Gandhinagar
Gangtok
Gurgaon
Gwalior
Habra
Hubli
Hyderabad
Indore
Itarsi
Jaipur
Jammu
Jalandhar
Kanpur
Kochi
Lucknow
Ludhiana
Mumbai
Mysore
Nagpur
Nasik
New Delhi
Noida
Pathankot
Patna
Pondicherry
Pune
Sagar
Solapur
Surat
Thiruvanantpuram
Ujjain
and many more……
Branch - Head Office
A-206, Phoenix House, 2nd Floor, Senapati Bapat Marg, Lower Parel, Mumbai- 400 013.
Telephone No: 022-24989000
PROMOTION
Sharekhan limited is one of the leading broketage house in India , as it is 85 years old it is
has become a brand name in India. For attracting more customers sharekhan also do some
other activities for its promotion. The punchline of the company in it advertisement is “
YOUR GUIDE TO FINANCIAL JUNGLE” as the logo of the company is the lion so
this combination itself suggest Sharekhan to be a king of the financial jungle or share
market.Company advertise itself through print media most, The ECONOMIC TIMES,
BUSNESS WORLD etc. mostly contain its advertisement. Apart from the print
Advertisements, company also advertise itself by way of Canopies etc. Other than way of
advertisement the company promote itself by presales and post sale services as by
arranging first step seminars and by customer care department. The company provide free
of cost research repots 3 times a day to the clients for which mostly other companies
charges heavy fees. The company provide yearly magazines to the clients named Value
line. Thus by the adoption of the different means the company is promoting itself.
MARKETING STRATEGIES
PRODUCT STRATEGY
The product strategy of the company is very good to compete with competitors the
company operate in different field of the financial services which provide the different
kind of variety in the services the major services it provide are
Equity Trading
Futures and options in Equity
Commodity
Futures in Commodity
Mutual Funds
IPO
Depository
PRICE STRATEGY
The price strategy of the Sharekhan is also better. The following are its features:
It have a different range of prices for different type of customersIt have very less
maintenance charges.It have different range of brokerage charges that depends upon the
order of the client.It do not charge any hidden charges like many other competitors which
is helpful in building its clean image.
It do not charge on research reports provided to the clients for which many other
competitors charge very high amount.
PLACE STRATEGY
Share khan have 588 share shops in 213 cities across the India. That provide a large
network of its services. That is easily accessible to the customer.
PROMOTIONAL STRATEGY
Sharekhan adopt different promotional activities for the growth of the organization. The
following are some ways adopted by sharekhan to promote itself.
Advertising:
Print advertising as magazines, newspapers, Hoardings etc.
Awareness:
Sharekhan also promote itself by awaring people about itself and share market, its
product and services. These modes are
Seminars
Canopies.
Surveys.
TIE UP
As its strategies Sharekhan have also tie up with the following Banks. These banks are
Union Bank Of India
HDFC Bank.
Oriental Bank of Commerce
ING Vysya Bank.
INDIA BULLS
Introduction of the company.
Indiabulls Financial Services Ltd (Indiabulls) is one of the leading integrated retail
financial services company in India which was incorporated in the year 2000. It offers a
full range of financial services and products ranging from Equities to Insurance.
Indiabulls provides full access to all the products and services through multi-channels. As
on 31st March 2005,the company operates through its head office in New Delhi and 148
offices in 76 cities. The company completed its Initial Public Offering in September 2004
by issuing 27,187,519 equity shares of par value of Rs.2 per share at a premium of Rs.17
per share. The company has raised Rs.51.6 crores and has used the proceeds to
consoildate its market leadership and enter new businesses. Indiabulls Securities Ltd,
Indiabulls Insurance Advisors Pvt Ltd, Indiabulls Commodities Pvt Ltd, Indiabulls Credit
Services Ltd and Indiabulls Finance Company Pvt Ltd are the subisidiaries of the
company. Indiabulls Professional Network is a flagship product which offers real-time
prices, detailed data and news, intelligent analytics and electronic trading capabilities.
Management of the Company
Whole-time Director Sameer Gehlaut
President & CFO Rajiv Rattan
Director Saurabh K Mittal
Director Aishwarya Katoch
Director Shamsher Singh
Director
Director Kartar Singh Gulia
Whole time director Gagan Banga
Company Secretary Amit Jain.
Product Details of the company.
Product Name Un Installed Production Sales Sales
it Capacity Quantity Quantity Value
Commission Rs. 0 0 0 0
Interest Rs. 0 0 0 51.89
License Fees Rs. 0 0 0 0.06
Profit on sale of
InvestmentRs. 0 0 0 0
Financial Summary
FigureHeads DataType Months Months
Date - 200603 200512
For Months - 3 9
Net Sales/Income Rs 66.52 140.57
Other Income Rs 4.85 0
OPBDIT Rs 58.72 115.12
DEP Rs 0.2 0.1
Tax Rs 9.99 24.52
PAT Rs 22.12 52.58
Equity Rs 32.05 32.41
EPS Rs 1.77 1.77
BV Rs 26.3 26.3
P/E % 55.3587
The Top 5 mutual Funds that own the company
Scheme Name No.Of Shares % Portfolio
JM FMP - Yrly SA2 (G) 0 26.33
JM FMP - Yrly SA2 (D) 0 26.33
JM FMP - Yrly SA2 (G) 0 26.32
JM FMP - Yrly SA2 (D) 0 26.32
JM FMP - Yrly SA2 (G) 0 26.31
Top 3 Share holding of the company
Shareholding Top 3
Total Foreign 58.22
Total Promoters 30.34
Total Public & Others 4.71
Quarterly Results of the Company
Name Mar-2006 Dec-2005 Sep-2005 Jun-2005
Sales Turnover 66.52 42.16 54.42 43.12
Other Income 4.85 0.28 0.11 0.47
Total Income 71.37 42.44 54.53 43.59
Total Expenditure 12.65 8.65 9.29 7.51
Operating Profit 58.72 33.79 45.24 36.08
Interest 26.41 13.32 16.49 8.1
Gross profit 32.31 20.47 28.75 27.98
Depreciation 0.2 0.07 0.03 0.01
Tax 9.99 5.47 9.48 9.56
Reported Profit After
Tax22.12 14.93 19.24 18.41
Extra-ordinary Items 0 0 0 0
Adjusted Profit After
Extra-ordinary item22.12 14.93 19.24 18.41
EPS (Unit Curr.) 1.38 0.92 1.35 1.37
Book Value (Unit Curr.) 0 0 0 0
Dividend (%) 0 25 0 0
Equity 32.05 32.41 32.41 26.89
PBIDTM(%) 88.2742 80.1471 83.1312 83.6735
PBDTM(%) 48.5719 48.5531 52.8298 64.8887
PATM(%) 33.2532 35.4127 35.3546 42.6948
Price Status of the company as on the july 3,2006
BSE 265.55 +10.90
NSE 265.80 +11.80
52 Week High 368.70
52 Week Low 22.05
AdditionalCompanyProfileDetails
Chairman/BoardMembers:
Whole-timeDirector : SameerGehlaut
FaceValue:2
MarketLot:1
Industry:Finance & Invest
Fortis financial Services ltd.
Introduction of the company
Promoted by Ranbaxy Laboratories, Fortis Financial Services (FFSL) was incorporated
in Mar.'94. FFSL, together with its associates, have acquired 16.52 lac fully paid-up
equity shares of Rs 10 each of the Empire Finance Company (EFCL) at a price of Rs
48.75 per share, representing 43.95% of the voting capital of EFCL, later in Jul' 95 it was
amalgamated with the company. The company is engaged in the business of leasing, hire
purchase and other related financial services. FFSL went public in Feb.'95 to augment
resources to meet the needs of its planned growth. The company obtained its category-I
merchant banking registration from SEBI in Apr.'95. The company has been trying hard
to recover money from various clients through legal procedures and various other
measures and accordingly written off Rs 37.14 cr during the year 1999-2000 in cases
where there are no changes of any recovery.
BIO DATA OF THE COMPANY
Industry Name: Finance & Invest
House Name: BQ
Year Of Incorporation: 1994
Regd.Office
Address: 55 Hanuman Road,Connaught Place
District: New Delhi
State: New Delhi
Pin Code: 110001
Telephone No.: 91-011-51512000
Fax: 91-011-23354944
Email Id: WebSite:
Auditors
Auditors: R V Shah & Co Company Status: A
Registrars
Name: Intime Spectrum Registry Ltd Address: 3rd Flr Phase I A-31,Naraina Indl
Area,Near Payal Cinema,New Delhi - 110 028
Tel No.: 91-11-51410592-94 Fax: 91-11-25896530
Management of the Company
Name
Chairman Harpal Singh
Managing Director Sunil Godhwani
Director Malvinder Mohan Singh
Director Shivinder Mohan Singh
Director Vinay Kumar Kaul
Director V M Bhutani
Director Umesh Kumar Khaitan
Company Secretary Sunil Kumar Garg
Product of the Company
Product Unit Installed Production Sales Sales
Name Capacity Quantity Quantity Value
Commission Rs. 0 0 0 0.21
Consultancy Rs. 0 0 0 0.69
Dividend Rs. 0 0 0 0.05
Interest Rs. 0 0 0 0.63
Lease Rentals Rs. 0 0 0 0.12
Sale Of Shares Rs. 0 0 0 4.56
Services
ChargesRs. 0 0 0 0.22
Financial Summary of the Company
FigureHeads DataType Months Months
Date - 200603 200512
For Months - 3 9
Net Sales/Income Rs 0.06 26.03
Other Income Rs 12.22 0
OPBDIT Rs -9.35 25.27
DEP Rs 0.24 0.74
Tax Rs 0.07 0.01
PAT Rs -9.7 24.07
Equity Rs 25.86 25.86
EPS Rs 0.16 0.16
BV Rs -4.17 -4.17
P/E % 6.52948
Major share holding of the Company
Shareholding Top 3
Total Promoters 78.71
Total Public & Others 13.49
Total Foreign 4.16
Price status of the company as on july 3, 2006
BSE 55.80 +2.65
NSE 0 0
52 Week High 66.00
52 Week Low 12.90
AdditionalCompanyProfileDetails
Chairman/BoardMembers:
Chairman : HarpalSingh
FaceValue:10
MarketLot:1
Industry:Finance&investments
History of the company
The KVB had commenced its banking business in 1916 at Karur Town (Tamilnadu).
Over the years the bank has graduated to become one of the top banks in the private
sector with strong and healthy fundamentals. The Bank has been changing its style of
functioning to fit into the constantly changing and dynamic technology age. Apart from
regular banking business, the bank also focuses on merchant banking, leasing and other
fee-based business. Merchant banking activities have been beefed up by creating separate
cells in Mumbai, New Delhi, Chennai and Secunderbad to cash in on the growing
potential in the capital market following the growth impulses in the economy. In 1995 the
bank issued 20,00,000 bonus shares in the ratio of 1:1 which was followed by rights issue
in the ratio of 1:2 at a premium of Rs 25 per share in 1996. Thus equity share capital
stood increased to Rs.6 crore. KVB and I-Flex Solutions have announced that I-Flex's
flagship product Flexcube was deployed across 183 branches of the bank. The
deployment of the flexcube Universal Banking system will enhance bank's corporate and
Retail banking business, and offer its customers anytime, anywhere access to its services.
This software would help to offer customers services through multiple delivery channels
including internet, phone, ATM etc. With a view to strengthening its capital base and
offering attractive returns to shareholders,the bank has awarded its shareholders bonus
shares in the ratio of 1:1. Subsequent to this bonus issue the Equity capital stood
increased at Rs.16.40 crores from Rs.6 crores. The bank has also made rights issue on
December,2002 to January,2003 and the issue was oversubscribed. The bank has tied up
for bancassurance with Bajaj Allianz General Insurance to hawk their non-life insurance
products through their branches. The total branches as at March,2005 were 231 and the
ATMs at 156. During 2004-05 the company introduced 6 new loan products i.e KVB
Special Home Loan, IPO Funding Scheme, KVB Kisan Mithra Scheme, Easy Trade Fin
Scheme, KVB Happy Kisan Scheme and Gold Card Scheme for Export Constituents of
the Bank. Further the company has launched a new product 'Cash Passport' which is
similar to ATM/Debit card and this product is offered in pursuance of the agreement
entered into with 'Travelex' which is engaged in travel related services all over the world.
Travellers going abroad can use this card preloaded with the required foreign exchange.
In 2004-05 the bank has entered into an agreement with MITR consortium because of
which ATMs of Punjab National Bank, Oriental Bank of Commerce, Indian Bank and
UTI Bank are at the services of KVB customers. Further the bank has implemented
RTGS facility for instant funds transfer across the country in 26 centres. In 2005-06 the
bank has launched Mobile Top-up facility to re-charge the cell phone of all service
providers through our ATM.
Financial Summary of Company
FigureHeads DataType Months Months
Date - 200603 200512
For Months - 3 9
Net Sales/Income Rs 172.14 566.4
Other Income Rs 56.46 0
OPBDIT Rs 151.46 398.25
DEP Rs 0 0
Tax Rs 8.9 37.5
PAT Rs 44.37 90.98
Equity Rs 17.98 17.98
EPS Rs 73.59 73.59
BV Rs 484.78 484.78
P/E % 7.69341
Balance Sheet of Company for last 5 financial years
Name Mar-2006 Mar-2005 Mar-2004 Mar-2003 Mar-2002
CAPITAL AND
LIABILITIES
Capital + 17.98 17.98 17.98 16.41 6
Reserves and Surplus
+853.65 742.9 694.05 542.27 424.11
Deposits + 7576.84 6672.19 5911.48 5121.92 4180.06
Borrowings + 195.62 92.31 103.19 267.76 306.09
Other Liabilities &
Provisions +363.8 359.47 380.74 225.35 193.86
TOTAL 9007.89 7884.85 7107.44 6173.71 5110.12
ASSETS
Cash & Balances
with RBI470.62 381.49 327.05 230.47 209.96
Balances with Banks
& money at Call &
311.73 273.68 270.75 458.58 649.09
Short Notice
Investments + 2298.13 2219.03 2173.01 1845.08 1538.91
Advances + 5555.45 4619.8 4023.24 3344.4 2460.03
Fixed Assets + 98.43 102.16 92.18 85.53 73.7
Other Assets + 273.53 288.69 221.21 209.65 178.43
TOTAL 9007.89 7884.85 7107.44 6173.71 5110.12
Contingent
Liabilities +2552.92 2003.48 1958.66 1999.1 1907.03
Bills for collection 520.23 381.67 432.87 234.44 213.49
The Product of Company
Product Name Un Installed Production Sales Sales
it Capacity Quantity Quantity Value
Income on investments Rs. 0 0 0 186.94
Interest on balance with
RBIRs. 0 0 0 10.68
Interest/disc on
advance/billsRs. 0 0 0 453.16
Others Rs. 0 0 0 0.07
10.32
6.8
3.52
2.24
1.9
3.86
2.32
28.6
Top 3 Shareholding of the company.
Total Public & Others 55.10
Total Foreign 20.93
Total Institutions 10.56
Top 5 Mutual funds That own the company
Scheme NameNo.Of Shares % Portfolio
DSP ML FTP - Series 3 (G) 5000.00 30.00
DSP ML FTP - Series 3 (D) 5000.00 30.00
Kotak FMP - Series 13 (G) 0 19.00
Kotak FMP - Series 13 (D) 0 19.00
DWS FTF - Series 4 (G) 0 14.99
The Price Status of the Company as on july 3, 2006
BSE 573.95 +8.10
NSE 573.65 +7.35
52 Week High 698.00
52 Week Low 271.15
Additional Company Profile Details
Chairman/BoarMembers:
Chairman&CEO : PTKuppuswamy
FaceValue:10
MarketLot:50
Industry:Banks - Private Sector
ICICI direct
History of the Company
ICICI Bank (ICICIBK) is a commercial bank promoted by ICICI Ltd, an Indian Financial
Institution. It was incorporated in Jan.'94 and received its banking licence from Reserve
Bank of India in May.'94. It is the 2nd largest bank in India. The bank has 562 branches
& extension counters across India and 1910 ATMs. The Bank offers a wide spectrum of
domestic and international banking services to facilitate trade, investment
banking ,Insurance, Venture Captial, asset management, cross border business & treasury
and foreign exchange services besides providing a full range of deposit and ancillary
services for both individuals and corporates through various delivery Channels and
specialized subsidiaries. All the branches are fully computerised with the state-of-the-art
technology and systems, networked through VSAT technology. The bank is connected to
the SWIFT International network. The bank has 14 subsidiaries across India and other
countries like Uk, Canada and Russia. To maintain the leadership status bank foray into
internet banking by web- enable its existing products and services. It has gained
favourable acceptance from its customer for its initiatives in business to business and
business to customer solution. To efficiently distribute its products and services, the bank
has developed multiple access channels comprising lean brick and mortar branches,
ATMs, call centers and Internet banking. The Bank has introduced the concept of mobile
ATMs in the remote/rural areas. It has also extended its mobile banking services to all
cellular service providers across India and NRI customers in USA,UK,Middle-East and
Singapore. In 2000-01 the Bank of Madura (BOM) got merged with ICICIBK. With this
merger ICICIBK has become one of the largest private sector banks in India. The Board
of Directors , approved the merger of ICICI(Financial Institution) with ICICI Bank in
2001. The two subsidiaries of ICICI Ltd viz ICICI Personal Financial Serivces and ICICI
Capital Services was also merged with the ICICI Bank with effective from 28th March
2002. During May,2003 the bank has acquired Transamerica Appple Distribution Finance
Private Ltd which is primarily engaged in financing in the two-wheeler segment. After
acquisition the name of the company was changed to ICICI Distribution Finance Private
Limited. The Banks subordinated long-term foreign currency debt was upgraded to Baa3
to Ba1 by Moody's Investor Service. In the Wholesale Banking segment,the bank has
achieved a significant milestone in the market making activity by expanding the product
suite to include foreign exchange options against Indian Rupee as RBI allowed them to
be traded w.e.f.07.07.2003.The bank has emerged as one of the largest market-makers in
merchant as well as inter-bank markets for this product.
FigureHeads DataType Months Months
Date - 200603 200512
For Months - 3 9
Net Sales/Income Rs 3989.79 13175.9
Other Income Rs 1601.92 0
OPBDIT Rs 3658.22 9035.83
DEP Rs 0 0
Tax Rs 239.87 462.45
PAT Rs 789.93 1750.14
Equity Rs 889.83 740.92
EPS Rs 25.99 25.99
BV Rs 170.34 170.34
P/E % 17.2226
Financial Summary of The Company
Management of the company
Chairman N Vaghul
Managing Director & CEO K V Kamath
Director Somesh R Sathe
Director Lakshmi N Mittal
Director Anupam Puri
Director Marti G Subrahmanyam
Deputy Managing Director Nachiket Mor
Joint Managing Director Kalpana Morparia
Deputy Managing Director Chanda D Kochhar
Director P M Sinha
Nominee (Govt) Vinod Rai
Director M K Sharma
General Manager & CS Jyotin Mehta
Joint Managing Director Lalita D Gupte
Addtnl Non-Executive Director V Prem Watsa
Director Sridhar Iyengar
Director T S Vijayan
Additional Director R K Joshi
Additional Director Narendra Murkumbi
Product of the Company
Product Name UnitInstalled
Capacity
Production
Quantity
Sales
Quantity
Sales
Value
Income on investments Rs. 0 0 0 2229.44
Income on investments Rs. 0 0 0 3692.76
Interest on balance with
RBIRs. 0 0 0 232.01
Interest on balance with
RBIRs. 0 0 0 335.46
Interest/disc on
advance/billsRs. 0 0 0 6752.83
Interest/disc on
advance/billsRs. 0 0 0 9684.96
Others Rs. 0 0 0 71.32
Others Rs. 0 0 0 195.61
Finantial analysis by Ratios
Name Mar-2005 Mar-2004 Mar-2003 Mar-2002 Mar-2001
Key Ratios
Credit-Deposit(%) 91.74 99.7 125 111.56 40.73
Investment / Deposit (%) 55.52 67.26 88.91 90.95 48.02
Cash / Deposit (%) 7 8.85 8.3 6.2 7.44
Interest Expended /
Interest Earned (%)69.83 77.93 84.8 72.44 67.44
Other Income / Total
Income (%)27.33 25.41 25.26 22.1 15.45
Operating Expenses /
Total Income (%)25.49 21.32 16.1 23.07 22.07
Interest Income / Total
Funds (%)6.39 7.7 8.84 3.47 7.81
Interest Expended / Total
Funds (%)4.46 6 7.5 2.52 5.27
Net Interest Income /
Total Funds (%)1.93 1.7 1.34 0.96 2.54
Non Interest Income /
Total Funds (%)2.4 2.62 2.99 0.98 1.43
Operating Expenses /
Total Funds (%)2.24 2.2 1.9 1.03 2.04
Profit before Provisions / 2.09 2.12 2.43 0.91 1.93
Total Funds (%)
Net Profit / Total funds
(%)1.36 1.4 1.14 0.42 1.01
RONW (%) 19.51 21.91 18.87 7.23 13.21
Bio Data of the Company
Industry Name: Finance & Invest
House Name: BQ
Year Of Incorporation: 1994
Regd.Office
Address: 55 Hanuman Road,Connaught Place
District: New Delhi
State: New Delhi
Pin Code: 110001
Telephone No.: 91-011-51512000
Fax: 91-011-23354944
Email Id: WebSite:
Auditors
Auditors: R V Shah & Co Company Status: A
Registrars
Name: Intime Spectrum Registry Ltd Address: 3rd Flr Phase I A-31,Naraina Indl
Area,Near Payal Cinema,New Delhi - 110 028
Tel No.: 91-11-51410592-94 Fax: 91-11-25896530
Balance Sheet of the Company
NameMar-
2005
Mar-
2004
Mar-
2003
Mar-
2002Mar-2001
CAPITAL AND
LIABILITIES
Capital + 1086.76 966.4 962.66 962.55 196.82
Reserves and 11813.2 7394.16 6320.65 5632.41 1092.26
Surplus +
Deposits + 99818.8 68108.6 48169.3 32085.1 16378.2
Borrowings + 33544.5 30740.2 33178.5 48681.2 1032.79
Other Liabilities
& Provisions +22172.1 18940.2 19129.1 17598.2 1036.51
TOTAL 168435 126150 107760 104959 19736.6
ASSETS
Cash & Balances
with RBI6344.9 5408 4886.14 1774.47 1231.66
Balances with
Banks & money
at Call & Short
Notice
6585.08 3062.64 1602.86 11011.9 2362.03
Investments + 50487.3 42742.9 35462.3 35891.1 8186.86
Advances + 91405.1 62647.6 53279.4 47034.9 7031.46
Fixed Assets + 4038.04 4056.41 4060.73 4239.34 384.75
Other Assets + 9574.82 8232.02 8468.83 5007.82 539.83
TOTAL 168435 126150 107760 104959 19736.6
Contingent
Liabilities +268154 202942 89438.5 39446.6 13848
Bills for collection 2392.09 1510.93 1336.78 1323.42 1229.8
Total Shareholding of the Company
Total Foreign 73.52
Total Institutions 15.07
Total Public & Others 6.76
Top 5 Mutual Funds that own the company
Scheme Name No.Of Shares % Portfolio
Birla FMP - Annual Series 2 (D) 0 426.28
Birla FMP - Annual Series 2 (G) 0 426.28
Grindlays FMP - 11 - A (G) 0 99.91
Grindlays FMP - 11 - A (D) 0 99.91
Grindlays FMP - 11 - A (G) 0 99.90
Additional Company Profile Details
Chairman/BoardMembers:
Chairman : NVaghul
FaceValue:10
MarketLot:1
Industry:Banks-PvtSector
Karvy
Karur Vysya Bank Ltd. The KVB had commenced its banking business in 1916 at
Karur Town (Tamilnadu). Over the years the bank has graduated to become one of the
top banks in the private sector with strong and healthy fund.
Financial Summary
FigureHeads DataType Months Months
Date - 200603 200512
For Months - 3 9
Net Sales/Income Rs 172.14 566.4
Other Income Rs 56.46 0
OPBDIT Rs 151.46 398.25
DEP Rs 0 0
Tax Rs 8.9 37.5
PAT Rs 44.37 90.98
Equity Rs 17.98 17.98
EPS Rs 73.59 73.59
BV Rs 484.78 484.78
P/E % 7.69341
Balance sheet
Name Mar-2006 Mar-2005 Mar-2004 Mar-2003 Mar-2002
CAPITAL AND LIABILITIES
Capital + 17.98 17.98 17.98 16.41 6
Reserves and Surplus +
853.65 742.9 694.05 542.27 424.11
Deposits + 7576.84 6672.19 5911.48 5121.92 4180.06
Borrowings + 195.62 92.31 103.19 267.76 306.09
Other Liabilities & Provisions +
363.8 359.47 380.74 225.35 193.86
TOTAL 9007.89 7884.85 7107.44 6173.71 5110.12
ASSETS
Cash & Balances with RBI
470.62 381.49 327.05 230.47 209.96
Balances with Banks & money at Call & Short Notice
311.73 273.68 270.75 458.58 649.09
Investments + 2298.13 2219.03 2173.01 1845.08 1538.91
Advances + 5555.45 4619.8 4023.24 3344.4 2460.03
Fixed Assets + 98.43 102.16 92.18 85.53 73.7
Other Assets + 273.53 288.69 221.21 209.65 178.43
TOTAL 9007.89 7884.85 7107.44 6173.71 5110.12
Contingent Liabilities +
2552.92 2003.48 1958.66 1999.1 1907.03
Bills for collection 520.23 381.67 432.87 234.44 213.49
Financial Analysis of the company
Name Mar-2006 Mar-2005 Mar-2004 Mar-2003 Mar-2002
Key Ratios
Credit-Deposit(%) 71.41 68.68 66.78 62.4 60.47
Investment / Deposit (%) 31.7 34.9 36.42 36.38 35.58
Cash / Deposit (%) 5.98 5.63 5.05 4.73 5.76
Interest Expended / Interest Earned (%)
56.54 56.55 54.11 67.21 65.88
Other Income / Total Income (%)
18.03 16.07 11.04 20.66 17.84
Operating Expenses / Total Income (%)
24.87 24.24 21.56 16.03 15.12
Interest Income / Total Funds (%)
7.71 7.88 9.75 9.14 10.32
Interest Expended / Total Funds (%)
4.36 4.46 5.28 6.14 6.8
Net Interest Income / Total Funds (%)
3.35 3.42 4.48 3 3.52
Non Interest Income / Total Funds (%)
1.7 1.51 1.21 2.38 2.24
Operating Expenses / Total Funds (%)
2.34 2.28 2.36 1.85 1.9
Profit before Provisions / Total Funds (%)
2.71 2.66 3.32 3.53 3.86
Net Profit / Total funds (%)
1.6 1.41 2.43 2.22 2.32
RONW (%) 16.58 14.3 25.35 25.28 28.6
Top 5 mutual funds that own the company
Scheme Name No.Of Shares % Portfolio
DSP ML FTP - Series 3 (G) 5000.00 30.00
DSP ML FTP - Series 3 (D) 5000.00 30.00
Kotak FMP - Series 13 (G) 0 19.00
Kotak FMP - Series 13 (D) 0 19.00
DWS FTF - Series 4 (G) 0 14.99
Top 3 shareholder of the company
Shareholding Top 3
Total Public & Others 55.10
Total Foreign 20.93
Total Institutions 10.56
MANAGEMENT THEORY AND SHAREKHAN
In the management theory we study the principles of managemen that should be present
in any organization for the management of that organization. These principles are about:
Division of work
Work Specialization
Equity
Organizational hierarchy
Unity of command
Unity of direction
Scalar chain
Spain of control
Communication
Equity
Motivation
Team Work
Employees’ participation.
In Sharekhan I observed most of these principles are outdated or are present in the
organization in some different or modified form these principles are discussed with
context to Sharekhan as follows
Division of work ;
Work division is done in Sharekhan securities ltd. but the employees working there are
not restricted to do other jobs which are not assigned to them in case of necessity the
employees cooperate each other and handle the other jobs also.
Work specialization
This principle is followed in Sharekhan, the employees are placed according to their
area of specialization, like marketing people for marketing of product of company and
people from finance for research analyst jobs and technicians are people from
Technology background.
Organizational hierarchy
The organizational hierarchies shows the relationships of subordinates and superiors in
the organization , this is not exactly same in the Sharekhan however the relationship of
the people in Sharekhan have been pre defined but the employees work together as team
work. There are no formalities an executive can directly meet to the branch manager
without informing his superior manager, territory manager .
The organizational chart in the Sharekhan is as follows mainly concerned for a single branch
Chief executive officer
Regional Manager
Regional Manager
Regional Manager
Regional Manager
Branch Manager
Territory Manager
Thus this is the hierarchy of the management in the Sharekhan ltd, but the relations for accountability, communication etc. are not so much formal as provided in management theories.
GOVERNMENT POLICY
Sharekhan limited is a brokerage house. The company follows all the rules and
regulations as prescribed by companies law, and other guidelines issued by the SEBI for
a company. Morever SEBI also regulates the brokers of the share market. As being a
brokerage company Sharekhan always follow the prescribed guidelines issued by the
SEBI and other laws. That is the reason that ShareKhan was not listed in the black
list companies issued by SEBI like others India bulls etc.
Share khan always comply with legal formalities. It functions very transperantly, It
always secure the investors interest.
AssistantManager
Relation manager
Executives
Sharekhan mainly comply with all SEBI guidelines issued as:
Capital adequacy norms
Registration norms
Payment of prescribed fees
Prevention from insider trading
Prevention of market manipulation
Investors interest protection
Code of conduct etc.
Thus Sharekhan is a clean and transperant organization which follows all the rules and
regulations and its government policy is cooperating.
NATIONAL AND INTERNATIONAL POSTION
Sharekhan ltd. is one of India’s leading and largest sharebroker company. It is 85
years old in this field. The original name of co. is S.S. Kantilal Ishwar lal Securties pvt.
Ltd. but the company advertise it with the name of Sharekhan ltd.
Customers have to do is walk into any of Sharekhan 588 share shops across 213 cities in
India to get a host of trading related services – the friendly customer service staff will
also help them with any account related queries they have.Share Khan ltd. has many
branches all over the India through where the company operates in different areas and
The company has a long chain of frenchisee also so the network of sharekhan ltd has
spread across all over the country. Share khan shops are available many places in India
and customer can take assistance from any of these for the product and service availing.
Regarding international image company deals in share market in India, so the other
countries people mostly through their own country stock exchanges and Sharekhan is
registered with Indian stock exchanges, however as it functions through online system so
the NRIs who invest in Indian share market mainly invest through Sharekhan so the
companies international image among NRIs is very good.
TAX ASPECTS OF THE COMPANY
About the taxation of the company Sharekhan ltd always pay its taxes timely and
honestly.
The main type of taxes paid by the company are
1. Income Tax
2. Sales Tax
3. Service Tax
However the company don’t reveal the amount of the taxes paid by it as it is not
necessary for it to do so because the company operate its business under the name of
“S.S. Kantilal ishwar lal securities pvt. Ltd, which is not a public company and hence
It need not to publish its accounts reports publicly.
Morever Sharekhan ltd. as being public company have not issued its capital to the
general public so it also need not to show its accounts reports like P&l and Balance
Sheets publicly.
SHARE MARKET POSTION OF SHAREKHAN LTD.
A company’ s share market postion refers its prices position of shares in the securities
maeket. Those companies which are listed on the stock exchanges have already issued its
shares in the market.
But Sharekhan limited has not issued its capital in the general public till now,
consequently it is not listed on the stock exchanges and hence the question for the share
market position of the company do not arise.
MAJOR PROBLEMS FACED BY THE COMPANY
The main problems faced by the company are as follows:
1. lack of motivation of investment in securities market in Indian public.
2. Dependence of Indian stock market on the foreign stock markets.
3. Tight legal frame work of SEBI.
4. Throat Cut Competition by other companies in the broking business.
5. Very much legal formalities for investors.
6. lack of updation of technology in India.
ACHIEVEMENTS OF THE COMPANY
The company has highest no. of share shops in India.
The company’s network has been spread in 215 cities of India.
The company is the winner of Best Research publisher awards for 2005 for its
magazine Value line.
FUTURE PROSPECTS OF THE COMPANY
The future prospects of the company are bright of course, because of its fair and
transparent operations in the business, it has never been black listed by any legal body
for any breach of law.
The company in coming years will left its all competitors a far back and will be the only
name in the field of broking house .
Being stable for more than 85 years in the same field the company has created a brand
image in the financial market services. In coming month company is also issuing its First
IPO for the general public and hence its shares will also be traded in the securities
market.
Thus the company’s future prospects are good.
CHAPTER.2
RISK IN SECURITIES MARKET
The goal of the return management process is to maximize earnings in the context of an
acceptable level of risk. With risk defined as uncertainty, a key component of this process
is delineating the possible rangeof outcomes and their root causes. As new circumstances
arise, a securities lender is prompted to study existing data in order to achieve a greater
comprehension of potential situations a client may face.
The risks inherent in any securities lending program — including market, credit,
liquidity, operational,legal, and regulatory — are all relevant to the principal common
objectives shared by participants, namelystability of income and preservation of
principal. Of these, market and credit risk most readily lendthemselves to quantification
and modeling due to the greater frequency and depth of the data available.The following
narrative focuses on the cash-collateralized securities lending transaction. The risks to
participants who accept non-cash collateral will be addressed in a later article that will
deal, at length,with credit risk modeling efforts.
MARKET RISK
Market risk potentially impacts both the future market value of a portfolio of assets
and/or liabilities andspread income associated with the portfolio. The market risk
associated with securities lending containstwo components: interest rate risk and spread
rate risk.
INTEREST RATE RISK
Interest rate risk is the risk of interest rate fluctuations impacting spread income and/or
the value of theintegrated portfolio (i.e., both the collateral reinvestment and the funding
portfolios). This type of riskarises from maturity/reset timing mismatches between the
asset and liability positions. As noted in theDigest’s first article, the liability is the cash
collateral received from a borrower that has a certain costwhile the asset is the investment
purchased using that cash collateral that generates a certain yield.
RISK AND RETURNIN SECURITIES LENDING
To illustrate, consider an agent lender who has the opportunity to invest the cash received
from anovernight loan for which the agent lender will have to pay a 3.00% rebate into a
security that matures inthree months and yields 3.50%. There is a risk that over the
course of the three months the loan supporting this investment will become more
expensive (the rebate will rise), and thus spread incomewill decrease or potentially turn
negative. Had overnight rates actually increased more than anticipatedduring this period,
the three-month investment may not have been the optimal alternative.In this case,an
investment with a shorter maturity, albeit possibly initially at a lower spread, might have
been thebetter alternative since it would have matured more quickly and been reinvested
into an instrument reflecting current interest rates. In fact, as part of the interest rate risk
management process, a numberof alternative interest rate “paths” are modeled, in which
the timing and magnitude of potential ratechanges are examined.In addition to having an
effect on spread income, changes in interest rates also have an impact on themarket value
of the portfolio. The relationship of market value to purchase price is captured by its net
asset value (NAV). The first Digest article noted that a security with a yield exceeding
the current market rate of interest for an investment with a similar maturity structure and
credit quality will be valued inexcess of par, while a security with a yield lower than the
current market rate of interest for a similarinvestment will be valued below par. In the
latter scenario, an investment might have seemed attractiveat the inception of the
transaction, but with an increase in interest rates, the original investment is now trading at
a value below the purchase price.
In effect, if the security is not sold prior tomaturity, it willonly incur an opportunity cost
or a foregone opportunity to earn the current yield. As such, NAV is anindicator of how
the portfolio will perform relative to the market going forward, and increased volatilityof
this measure is suggestive of a build-up of risk.
SPREAD RATE RISK
Spread rate risk can be viewed as either market- or credit-related, but is best summarized
as the marketrisk associated with the macro-economic credit outlook. This risk affects
floating rate securities, whose return is impacted by the following elements: the index
rate and the spread over this rate, and an elementnof risk associated with each. The index
rate, or reference rate, is a designated interest rate to which the coupon of a floating rate
security changes (e.g., Prime, LIBOR). For example, consider a security that pays LIBOR
+ 5 basis points and resets on the 15th of every month. The interest rate risk component is
a function of the time to reset for the index rate, which in this case would be between the
16th of the current month and the 14th of the following month. Spread rate risk would be
the risk that the 5-basis-point spread to the index is no longer at a level appropriate to the
security.
The effect of such a change in market spreads has significant market value implications
for floating rate securities, which generally have longer expected maturities than the fixed
rate securitiestypically purchased in a securities lending program. However, the widening
and tightening of such spreads, which generally occurs in response to changes in
perceived credit quality for the class of
securities of which this issue is a part (e.g., AA Finance), will typically vary within a
narrow band. Over the longer term, it is the potential for changes in interest rate levels —
and not spreads — that poses the greater risk to earnings. Clearly, this is what makes
floating rate securities an attractive investment.
CREDIT RISK
The second primary risk factor is credit risk. This risk takes two forms — reinvestment
credit risk and borrower credit risk.
REINVESTMENT CREDIT RISK
Reinvestment credit risk is the risk that a change in the creditworthiness of an issuer will
result in achange in the market value of the issue. Whereas spread rate risk measures the
risk of a change in valuedue to changes in broad market credit concerns, credit risk, in
this context, measures the risk to value for a specific issue/issuer. In the extreme case, it
is the risk that default, or the inability of the issuer tomeet payment obligations, will
result in a substantial erosion in value. Changes in credit quality that do not result in
default will not have a realized monetary consequence unless the issue is sold prior to
maturity. However, there is an opportunity cost to the extent that the yield available to
current purchasers of the security is higher. A defaulted issue can impair the value of the
reinvestment portfolio to the extent most or all of its value is not recoverable.
All securities purchased for the collateral reinvestment vehicles must meet strict credit
quality standards. State Street devotes significant resources to analyzing the ongoing
creditworthiness of both its approved issuers and any prospective issuers under
consideration for inclusion in the program.
BORROWER CREDIT RISK
Borrower credit risk arises from the potential inability of a borrower to return the loaned
securities. Losses can arise when the collateral on hand is insufficient to purchase
replacement securities at the time at which a borrower defaults. Borrower credit risk is
alow probability, but potentially high impact, event.This risk can be mitigated by entering
into lending agreements with highly rated counterparties and by ensuring that the loans
are properly collateralized on a daily basis. State Street conducts a rigorous analysis of
both prospective and current borrowers to ensure clients’ financial protection. This
analysis is performed on a regular basis — monthly, quarterly and annually —and
includes a review of corporate and regulatory financial statements.
To measure the combination of both market and credit risks, a statistical model was
developed that combines the potential residual (or unsecured) risk in a portfolio of loans
and collateral to an individual borrower with an assessment of the likelihood that the
borrower will default over a defined time horizon. Residual risk is defined as the level of
price risk that may be unsupported by the collateral margin held.The beneficial effect of
an indemnification against borrower default, where provided, becomes part of
the analysis.
SL PERFORMANCEANALYZER®
The monitoring and management of risk within the portfolio is a key part of the return
management process at State Street. SL PerformanceAnalyzer®, a proprietary risk-
adjusted performance measurement toolset for securities lending, enables participants to
track portfolio market and credit risk through time, view program earnings within this
context, and view an internally derived combined estimate of this risk going forward.
THE RISKRETURN TREND ANALYZER
The RiskReturn Trend Analyzer enables a securities lending participant to view spread
return relative to aggregate portfolio risk (market and credit) over time. These risks will
vary in response to both changes in the risk profile of the portfolio and the volatility of
underlying market rates and spreads. The light bluebars represent spread, dark blue bars
represent risk and the grey line represents risk-return ratio. By having
access to such a metric, a participant can easily identify changes to the risk-return
dynamic.
THE NAV TREND ANALYZER
The NAV Trend Analyzer provides a view of the current NAV of a given collateral
reinvestment portfolio in the context of an estimate of how much it can be expected to
fluctuate over the coming month. This estimate combines the risk profile of the collateral
reinvestment portfolio with State Street’s expectationfor the volatility of underlying rates
and spreads to create a probability-based interval within which the NAV
is expected to fall. The preceding table details the current NAV projected one month
forward and the level at which the NAV actually settled. Widened bands signal periods of
increased risk while narrowed bands signal periods of reduced risk.
Risk comes in a variety of forms and is something to be managed by traders and investors
alike. The Optionetics approach to the markets prioritizes this topic and provides
individuals with a foundation for success. This three-part series will focus on investment
risk, trading risk and individual security risk.
InflationRisk
The existence of inflation risk is the main reason individuals accept market risk. If
current dollars had the same purchasing power any number of years going forward, a
compounded rate of return from liquid US Treasury Bills would certainly satisfy a
conservative investor and may even compete for the aggressive investor’s attention.
Convinced? If not, consider the following:
Data Period Jan 1, 1934 – Feb 1, 2006
Initial Value $1
Number of Months 869 (72.4 years)
Final Value (with inflation*) $1.0012
72.4 year return $15.55
Final Value (no inflation) 0.12%
72.4 year return 1,555%
* Based upon actual CPI data for the period (St. Louis Federal Reserve Bank)
So if the next 72.4 years exactly duplicates the period evaluated, you have less than a
penny to spare if your expenses remain the same and you choose to invest solely in T-
Bills. However, if you use a more realistic three month holding period for T-bill assets
with a 100% return of principal, you are left with an amount that is slightly under $1.00
(0.9924). Without inflation the picture is much rosier; an initial investment in T-Bills that
is left to compound monthly (rather than a quarterly compounding) will provide the
investor with an account valued at more than 15 times the initial amount.
Bond Risk
Perhaps a conservative investor decides that by simply replacing the T-Bill investment
with a longer maturity Treasury Note, all will be fine regardless of inflation. Although T-
Note yields are usually higher than T-Bills, periods where the yield curve becomes flat or
inverted highlight a different issue. The timing of the purchase for the longer term
security will impact investment returns. This brings us the next investor risk—interest
rate risk.
Interest rate risk is the risk of having your money locked into a lower rate of return while
interest rates rise. Assuming the asset is held to maturity to benefit from the 100% return
of principal, the longer the term to maturity for the fixed income investment, the more
opportunity there is to receive sub-standard interest rates on the money. As a result, the
10-year note is deemed to have more interest rate risk than the 90-day bill, while the 30-
year bond is deemed to have more interest rate risk than both
the note and the bill.
Using 10-year note returns starting in April 1953 and assuming 10-year holding periods
with inflation, a $1.00 initial investment will be valued at $2.65 after 639 months (53
years). A conservative investor may feel a 265% return is acceptable; however, we have
yet to account for taxes. Incorporating a 20% tax applied semi-annually results in a final
value of $2.21.
An investor who seeks even better fixed income returns by entering the municipal or
corporate bond world encounters yet another type of investment risk: credit risk.
Although defaults are less likely in the municipal bond market (reflected by lower
yields), they can occur. This type of investment does benefit from favorable tax treatment
which will boost returns.
In terms of corporate bonds, credit risk increases (reflected by higher yields) as the credit
rating decreases. AAA corporate bonds are deemed less risky from a return of principal
standpoint and, therefore, offer lower yields. Interest is taxable and there’s no guarantee
that the credit rating and safety of the bond downgraded during the holding
period.
At this time, historical municipal and corporate bond returns will not be analyzed. Those
interested may want to pursue this analysis using a bond fund or exchange traded fund
(ETF) proxy such as a Nuveen Municipal Fund or the iSharesâ Corporate Bond Fund. The
biggest challenges include obtaining sufficient return histories and incorporating
appropriate tax impacts to the result.
Stock Market Risk
Investors generally turn to the stock market to realize returns that will outpace inflation
after accounting for taxes. Modern Portfolio Theory (MPT) serves as one basis for
portfolio construction aimed at maximizing reward while minimizing risk. This approach
makes use of the following:
1) Capital Market LineRepresents optimal portfolios for a given level of
risk.
2)Capital Asset Pricing Model
(CAPM)
Assesses the risk-return impact of adding a security
into a well diversified portfolio.
3) Security Market LineA linear asset price model based on risk versus
returns.
4) Security Characteristic Line
A linear model of asset return versus market
return, using an asset risk measure to identify
undervalued and overvalued securities.
Beta, alpha and Sharpe Ratios are all measurements associated with portfolio and security
risk.
Inflation risk represents the risk associated with insufficient returns (not keeping pace
with increasing costs), while credit risk and stock market risk highlight represent the risk
associated with seeking higher returns (losses). Investors generally seek maximize returns
for a given level of risk by diversifying their portfolios (see previous articles on
diversification and correlations). However, market risk cannot be diversified away—
some risk of losses will always remain even in optimal
portfolios.
A two-part personal finance article series from May & June this year examined the
impact of inflation and investment returns on retirement savings when income was
removed from the account on a monthly basis. A thirty year period was examined starting
with January 1962 and investment returns were based upon actual S&P 500
returns, including dividends.
Using the same monthly inflation, return and dividend data, a $1 initial investment and a
20% tax rate on gains applied quarterly, the ending balance after 30 years was
approximately $1.26, or 126%. Remove the dividend returns and the account is valued at
$1.11 after thirty years. This represent a nice difference from compounded fixed income
returns over a 70 year period, particularly since the return of 0.12% excluded taxes.
Although the 10-year note return of 221% over 53 years seems to be an argument for T-
Note investments, when using thirty year results from initiation of the strategy, $0.96
remained in the account. Again, due to interest rate risk, the month of initiation will
impact the results with different thirty year periods yielding different returns.
Investment Risk
Investment risk includes credit risk and market risk which are both taken to combat
another risk—that due to inflation. Investors and traders need to assess the actual historic
results of the strategies they employ for their investments. In this manner, the individual
can better understand the true financial risks in which they are exposing themselves. Two
such risks may include under-investing in bonds & equities or over-investing in assets
with limited returns (i.e. an SPY Leap only strategy that misses
dividends).
In order to manage necessary market risk, the investor must take steps to understand
those risks and minimize them through diversification. Numerous articles on the topic of
asset allocation and diversification are available in the Optionetics.com article archives
by completing an author keyword search for 2004 and 2005. Traders, who often self-
direct their investment accounts, must proactively examine the performance of
investments versus the performance of trading given the goals for each, and re-evaluate
those allocations.
RETURNS IN SECURITIES MARKET
HISTORY
How to Calculate ReturnsThe Relationship between Inflation and Returns The Historical
Record: Year-to-year total returns on
common stocks - Average annual returns What’s the difference between Returns and
Risk PremiumsHow to calculate measures of the VariabilityReturns: standard deviations
and frequency distributions Risk and Return - 2
1. IF WE GET A 15% RETURN IS THAT GOOD?
- relative to a benchmark
- inflation
- discount rate
- risk: dispersion
2. HOW TO QUANTIFY AND ADJUST FOR RISK?
• DISTINCTION: EXPECTED RETURN VS. REALIZED
RETURN
DISPERSION ( WANT TO SAVE JOB)
OIL PROJECT: EXPECTED RETURN 15%
STD 30%
Prof. Gordon M. Phillips
Risk and Return - 3
3. WHERE DO DISCOUNT RATES COME FROM?
• WHY IS THE DISCOUNT RATE THE OPPORTUNITY
COST FOR THE FIRM?
• LOOK TO HISTORY AS GUIDE FOR PRESENT
4. WHY DO WE CARE ABOUT RISK?
• POTENTIAL OF BAD OUTCOMES
• IF TRULY INDEPENDENT CAN GET MULTIPLE BAD
DRAWS
==> HISTORICAL PRICE DOES NOT MATTER
• Risk Preferences are different across individuals!
==> It is not enough to say "I DON'T LIKE RISK" - We want to
measure how much risk individuals want to avoid.
2.0 CALCULATING RETURNS
• income component - direct cash payments such as dividends orinterest
• price change - loosely, capital gain or lossThe return calculation is unaffected by the
decision to cash out or holdsecurities.
Percentage Return: Refers to the rate per dollar invested.
Realized Percentage Return =
Dividend Yield + Capital Gains Yield
Where: Dividend Yield = Dt/Pt-1
Capital Gains Yield = (Pt - Pt-1) / Pt-1
Prof. Gordon M. Phillips
Risk and Return - 5
3.0 INFLATION AND RETURNS
A. Real versus Nominal Returns
• Nominal Returns - returns not adjusted for inflation; percentagechange in nominal
dollars.
• Real Returns - returns that have been adjusted for inflation;percentage change in
purchasing power.
B. The Fisher Effect
1. A expected relationship between nominal returns, real returns, and
the expected inflation rate. Let r be the nominal rate, R be the real
rate, and if be the expected inflation rate,
(1 + r) = (1+ R) x (1+ if)
hence r = R + if + (R x if)).
2. A definition whereby the real rate can be found by deflating the
nominal rate by the inflation rate:
R = (1 + r) / (1 + if) - 1.
Risk and Return - 6
4.0 AVERAGE RETURNS
A. Calculating Average Returns
ARITHMETIC AVERAGE: add them up, divide by T
• AAR =
• used in calculation of single period expectation.
GEOMETRIC AVERAGE RETURNS (HOLDING PERIODRETURNS):
• GAR =
• used in calculation of holding period returns.
GAR/Holding-Period Returns
The holding period return is the return that aninvestor would get when holding an
investmentover a period of n years, when the return during year i is given as ri :1 ) 1 ( ) 1
( ) 1 (
return period holding
2 1 - + ??+ ?+ =
=
n r r r m
Note the GAR is annualized version of theholding period return.Risk and Return - 8
Holding Period Return: ExampleSuppose your investment provides the following returns
over a four-year period:
Year Return
1 10%
2 -5%
3 20%
4 15% % 21 . 44 4421 .
1 ) 15 . 1 ( ) 20 . 1 ( ) 95 (. ) 10 . 1 (
1 ) 1 ( ) 1 ( ) 1 ( ) 1 (
return period holding Your
4 3 2 1
= =
- ???=
- + ?+ ?+ ?+ =
=
r r r r
Prof. Gordon M. Phillips
Risk and Return - 9
Holding Period Return (GAR): ExampleAn investor who held this investment would
have actually realized an annual return of 9.58%:
Year Return
1 10%
2 -5%
3 20%
4 15% % 58 . 9 095844 .
1 ) 15 . 1 ( ) 20 . 1 ( ) 95 (. ) 10 . 1 (
) 1 ( ) 1 ( ) 1 ( ) 1 ( ) 1 (
return average Geometric
4
4 3 2 1
4
= =
- ???=
+ ?+ ?+ ?+ = +
=
g
g
r
r r r r r
• So, our investor made 9.58% on his money for fouryears, realizing a holding period
return of 44.21%4 ) 095844 . 1 ( 4421 . 1 =
Risk and Return - 10
Arithmetic Average Return: Example
Note that the arithmetic average is not the same thing
as the holding period or geometric average:
Year Return
1 10%
2 -5%
3 20%
4 15%
% 10
4
% 15 % 20 % 5 % 10
4
return average Arithmetic 4 3 2 1
= + + - =
+ + + = r r r r
Prof. Gordon M. Phillips
Risk and Return - 11
Holding Period Returns
A famous set of studies dealing with the rates of returns oncommon stocks, bonds, and
Treasury bills was conductedby Roger Ibbotson and Rex Sinquefield.They present year-
by-year historical rates of return starting in
1926 for the following five important types of financial instruments in the United States:
• Large-Company Common Stocks
• Small-company Common Stocks
• Long-Term Corporate Bonds
• Long-Term U.S. Government Bonds
• U.S. Treasury Bills
Risk and Return - 12
The Future Value of an Investment of $1 in 1926
0.1
10
1000
1930 1940 1950 1960 1970 1980 1990 2000
Common Stocks
Long T-Bonds
T-Bills
$40.22
$15.64
63 . 845 , 2 $ ) 1 ( ) 1 ( ) 1 ( 1 $ 1999 1927 1926 = + ??+ ?+ ?r r r l
•
•the frequency distribution of the returns (see next slide).
•the standard deviation of those returns
Risk and Return - 14
U.S. Historical Returns, 1926-1999
Source: © Stocks, Bonds, Bills, and Inflation 2000 Yearbook™, Ibbotson Associates,
Inc., Chicago (annually updates work by
Roger G. Ibbotson and Rex A. Sinquefield).
– 90% + 90% 0%
Average Standard
Series Annual Return Deviation Distribution
Large Company Stocks 13.0% 20.1%
Small Company Stocks 17.7 33.9
Long-Term Corporate Bonds 6.1 8.7
Long-Term Government Bonds 5.6 9.2
U.S. Treasury Bills 3.8 3.2
Inflation 3.2 4.5
Prof. Gordon M. Phillips
Risk and Return - 15
5.0 RISK
A. ALONG WITH RETURN COMES RISK:
==>Security returns are examples of random variables -
•
B. The Principle of Diversification
• Principle of diversification: variability of multiple assets heldtogether less than the
variability of typical stock.
• The portion of variability present in a typical single security that isnot present in a large
group of assets held together (portfolio of assets) is termed diversifiable risk or unique
risk.
• Why does risk go down for a portfolio? Unique risks tend to
cancel each other out.
• The level of variance that is present in collections of assets is termed undiversifiable
risk or systematic risk.
• A typical single stock on NYSE annual
DRAWS
==> HISTORICAL PRICE DOES NOT MATTER
• Risk Preferences are different across individuals!
==> It is not enough to say "I DON'T LIKE RISK" - We want to
measure how much risk individuals want to avoid.
2.0 CALCULATING RETURNS
• income component - direct cash payments such as dividends orinterest
• price change - loosely, capital gain or lossThe return calculation is unaffected by the
decision to cash out or holdsecurities.
Percentage Return: Refers to the rate per dollar invested.
Realized Percentage Return =
Dividend Yield + Capital Gains Yield
Where: Dividend Yield = Dt/Pt-1
Capital Gains Yield = (Pt - Pt-1) / Pt-1
Prof. Gordon M. Phillips
Risk and Return - 5
3.0 INFLATION AND RETURNS
A. Real versus Nominal Returns
• Nominal Returns - returns not adjusted for inflation; percentagechange in nominal
dollars.
• Real Returns - returns that have been adjusted for inflation;percentage change in
purchasing power.
B. The Fisher Effect
1. A expected relationship between nominal returns, real returns, and
the expected inflation rate. Let r be the nominal rate, R be the real
rate, and if be the expected inflation rate,
(1 + r) = (1+ R) x (1+ if)
hence r = R + if + (R x if)).
2. A definition whereby the real rate can be found by deflating the
nominal rate by the inflation rate:
R = (1 + r) / (1 + if) - 1.
Risk and Return - 6
4.0 AVERAGE RETURNS
A. Calculating Average Returns
ARITHMETIC AVERAGE: add them up, divide by T
• AAR =
• used in calculation of single period expectation.
GEOMETRIC AVERAGE RETURNS (HOLDING PERIODRETURNS):
• GAR =
• used in calculation of holding period returns.
GAR/Holding-Period Returns
The holding period return is the return that aninvestor would get when holding an
investmentover a period of n years, when the return during year i is given as ri :1 ) 1 ( ) 1
( ) 1 (
return period holding
2 1 - + ??+ ?+ =
=
n r r r m
Note the GAR is annualized version of theholding period return.Risk and Return - 8
Holding Period Return: ExampleSuppose your investment provides the following returns
over a four-year period:
Year Return
1 10%
2 -5%
3 20%
4 15% % 21 . 44 4421 .
1 ) 15 . 1 ( ) 20 . 1 ( ) 95 (. ) 10 . 1 (
1 ) 1 ( ) 1 ( ) 1 ( ) 1 (
return period holding Your
4 3 2 1
= =
- ???=
- + ?+ ?+ ?+ =
=
r r r r
Prof. Gordon M. Phillips
Risk and Return - 9
Holding Period Return (GAR): ExampleAn investor who held this investment would
have actually realized an annual return of 9.58%:
Year Return
1 10%
2 -5%
3 20%
4 15% % 58 . 9 095844 .
1 ) 15 . 1 ( ) 20 . 1 ( ) 95 (. ) 10 . 1 (
) 1 ( ) 1 ( ) 1 ( ) 1 ( ) 1 (
return average Geometric
4
4 3 2 1
4
= =
- ???=
+ ?+ ?+ ?+ = +
=
g
g
r
r r r r r
• So, our investor made 9.58% on his money for fouryears, realizing a holding period
return of 44.21%4 ) 095844 . 1 ( 4421 . 1 =
Risk and Return - 10
Arithmetic Average Return: Example
Note that the arithmetic average is not the same thing
as the holding period or geometric average:
Year Return
1 10%
2 -5%
3 20%
4 15%
% 10
4
% 15 % 20 % 5 % 10
4
return average Arithmetic 4 3 2 1
= + + - =
+ + + = r r r r
Prof. Gordon M. Phillips
Risk and Return - 11
Holding Period Returns
A famous set of studies dealing with the rates of returns oncommon stocks, bonds, and
Treasury bills was conductedby Roger Ibbotson and Rex Sinquefield.They present year-
by-year historical rates of return starting in
1926 for the following five important types of financial instruments in the United States:
• Large-Company Common Stocks
• Small-company Common Stocks
• Long-Term Corporate Bonds
• Long-Term U.S. Government Bonds
• U.S. Treasury Bills
Risk and Return - 12
The Future Value of an Investment of $1 in 1926
0.1
10
1000
1930 1940 1950 1960 1970 1980 1990 2000
Common Stocks
Long T-Bonds
T-Bills
$40.22
$15.64
63 . 845 , 2 $ ) 1 ( ) 1 ( ) 1 ( 1 $ 1999 1927 1926 = + ??+ ?+ ?r r r l
•
•the frequency distribution of the returns (see next slide).
•the standard deviation of those returns
Risk and Return - 14
U.S. Historical Returns, 1926-1999
Source: © Stocks, Bonds, Bills, and Inflation 2000 Yearbook™, Ibbotson Associates,
Inc., Chicago (annually updates work by
Roger G. Ibbotson and Rex A. Sinquefield).
– 90% + 90% 0%
Average Standard
Series Annual Return Deviation Distribution
Large Company Stocks 13.0% 20.1%
Small Company Stocks 17.7 33.9
Long-Term Corporate Bonds 6.1 8.7
Long-Term Government Bonds 5.6 9.2
U.S. Treasury Bills 3.8 3.2
Inflation 3.2 4.5
Prof. Gordon M. Phillips
Risk and Return - 15
5.0 RISK
A. ALONG WITH RETURN COMES RISK:
==>Security returns are examples of random variables -
•
B. The Principle of Diversification
• Principle of diversification: variability of multiple assets heldtogether less than the
variability of typical stock.
• The portion of variability present in a typical single security that isnot present in a large
group of assets held together (portfolio of assets) is termed diversifiable risk or unique
risk.
• Why does risk go down for a portfolio? Unique risks tend to
cancel each other out.
• The level of variance that is present in collections of assets is termed undiversifiable
risk or systematic risk.
• A typical single stock on NYSE annual
To derive implications for systematic risk compensation and selectivity biases, we model
individual market returns as a two-regime process. We interpret regime 1 as the regime
when the market is open to international investors and investabledollar-denominated
returns are available for a given market. We discuss the details of this in the data section
below. Regime 2 is the regime where investable dollar returns
are not available and the market is inaccessible to international investors. We view the
transition from regime 1 to regime 2 as being associated with expropriationof
international investors. This expropriation can take various forms, includingcapital
controls, foreign exchange restrictions, and taxes on repatriations of foreign investments.
Information regarding the payoffs to international investors duringthe transition from
regime 1 to regime 2 and the ex-ante probability with which thistransition can happen are
not observable to an econometrician. In essence, we viewemerging markets returns akin
to payoffs of a defaultable bond which has not defaulted.As with the defaultable bond,
the likelihood of transition from regime 1to regime 2 affects measured mean returns
obtained solely from data sampled fromregime 1. Hence, one would expect that observed
mean returns, particularly for anemerging market, to be higher than the ex-ante risk
premium. This bias measures the compensation for expropriation and helps us understand
the risk-return relationacross markets. In equation (3), we present a time-series
representation of returns that will allowus to derive separate systematic risk
compensation from sample selectivity biasesin expected returns. Let yit+1 represent an
indicator for the regime in market i att + 1 being 1 or 2. The indicator yit+1 is equal to
one if the regime at t + 1 is 1 (open to international investors), and zero otherwise. The
return process, expressed indollars, is specified as
Rit+1 = E (Rit+1|It) + yit+1 (bi1et+1 + hi1t+1) + (1 - yit+1) (bi2et+1 + hi2t+1) , (3)
where E(Rit+1|It) is the ex-ante conditional mean of the gross return, et+1 is the
innovation in the systematic risk component, and hi1t+1 and hi2t+1 are diversifiable risk
components specific to market i. The exposure of the return to systematic risk is
determined by bi1 and bi2.Let rit+1 denote the excess return on market i, that is, rit+1 =
Rit+1 - Rf t.
Assuming that yit = 1, the valuation condition (1) then implies thatE (rit+1|It) = ls2et
[pitbi1 + (1 - pit) bi2] , (4)
where pit is the probability of the regime where market i is accessible to international
investors at time t. In other words, pit is the conditional probability that yit+1 = 1,and (1 -
pit) is the probability of a switch to regime 2. The risk premium is determinedby the
aggregate market price of risk, ls2et, and an overall beta which is aprobability-weighted
average of the betas in the two regimes. Next, we describe thedetermination of regimes 1
and 2.
The Sample Selectivity Process
Let y_it be a latent process that determines the opening and closing for market i. Thatis, it
determines if the regime is 1 or 2. In particular, if y_it >0 then the regime is 1and if y_it _
0, then the regime is classified as 2. Given this classification, it follows 6
thatyit = ( 1 if y_it > 0,0 otherwise.(5)
Following Heckman (1976, 1979), we assume that the conditional mean of the
latentprocess is determined by a vector of pre-determined variables xit. Hence, we
assume
That y_it+1 = d0ixit + #it+1, #it+1|xit _ N(0, 1) , (6)
where #it+1, by assumption, is a standard normal error. Brown, Goetzmann, andRoss
(1995) argue that the survival of a market (the analogue of our regime 1) is
determinedsolely by the price process itself. However, this seems restrictive, as many
emerging markets, such as Thailand, Indonesia, and Malaysia, have had comparable
drops in the market capitalization, but only Malaysia, directly expropriatedinternational
investors. This suggests that other economic considerations may be important in
determining whether investable dollar-denominated returns are availableto an
international investor. These other influences are captured by xit and #it+1. Further, the
latent variable model of selectivity provides connections betweendefault risk in sovereign
dollar-denominated bonds and the likelihood of capitalcontrols. This allows us to provide
a link between the cross-section of equity risk
premia and country risk ratings.Let f(.) denote the standard normal probability density
function, and let F (.)
denote the standard normal cumulative distribution function. It is straightforward to show
that the conditional probability that yit+1 = 1 is characterized by pit = E (yit+1|yit = 1,
xit) = Z ¥ -d0i xitf(#it+1)d#it+1 = Z d0i xit¥f(#it+1)d#it+1 = F _d0ixit_,(7)
where the third equality follows from the symmetry of the normal distribution. As#it+1
and the innovation in the return of asset i may be correlated, consider the
followingconditional projections for the different regimes
bi1et+1 + hi1t+1 = gi1#it+1 + vi1t+1, (8)
bi2et+1 + hi2t+1 = gi2#it+1 + vi2t+1, (9)
where gi1 and gi1 are the projection coefficients between bijet+1 +hit+1 and #it+1
andbijet+1 + hit+1 and #it+1, respectively, and vi1t+1 and vi2t+1 are projection errors.
The above equations then imply that the excess return process can be written as
rit+1 = E (rit+1|It) + yit+1(gi1#it+1 + vi1t+1) + (1 - yit+1)(gi2#it+1 + vi2t+1). (10)
2.4. The Sample Selectivity Criteria
We consider the case where data are missing as an outcome of an attrition process.That
is, we consider the sample selectivity effects of only observing the regimewhere the
markets are accessible to international investors. In this case, the restrictionon the
empirical conditional mean of the returns is
E (rit+1|It, yit = 1, yit+1 = 1) = E (rit+1|It) + gi1E _#it+1|#it+1 > -d0ixit_, (11)
where we have conditioned on the fact that the market is in regime 1 today (yit = 1)and
tomorrow (yit+1 = 1). This captures our view that an econometrician only
observesinvestable dollar return sample from regime 1. Note that E_#it+1|#it+1 > -
d0ixit_ is the same as E(#it+1|yit+1 = 1). Moreover, this quantity satisfies the relation
E (#it+1|yit+1 = 1) =1pit Z ¥-d0i xit#it+1f(#it+1)d#it+1, (12)
which can be further simplified as follows
1pit Z ¥-d0i xit #it+1 f(#it+1) d#it+1 = f(d0ixit)pit= f(d0ixit)F(d0ixit). (13)
This is typically referred to as a hazard rate, or the inverse Mill’s ratio. We denotethis by
hit, that is, hit = f(d0ixit)/F(d0ixit). Based on the above results, it follows thatthe
conditional mean of the excess return is given by
E (rit+1|It, yit = 1, yit+1 = 1) = ls2
et [pitbi1 + (1 - pit) bi2] + gi1hit. (14)
This restriction shows that there are two biases in measuring the ex-ante risk
premium.The first bias stems from the fact that the econometrician does not
observeregime 2 (the regime when investable dollar returns are not available). This is re-
flected in the first term of (14). bi1 can obviously be identified in the time series
fromobservations when the market is open. bi2 is the beta at transition from regime 1 to2.
Identification of bi2 and the transition probability of going to regime 2 (that is,81 - pit)
can not be measured without additional restrictions. Note that the resultingbias is on the
ex-ante mean of the return and we refer to it as a peso problem.
The second bias is due to sample selectivity, and the effects of this can be seenin the
second term of (14). This is an adjustment to the ex-post mean to correctly estimate the
ex-ante risk premium. Conditional on the availability of dollar returnstoday and
tomorrow, the risk premium is biased upwards. Put differently, investors require, on
average, a higher return when the market offers dollar returns, muchlike a defaultable
bond.
Brown, Goetzmann, and Ross (1995) focus on the second effect. It seems that the
measured risk premium will also be affected by the beta associated with the market shut-
down regime. If this beta is higher that in the regime for which datais available, then the
ex-ante mean asset will be higher, and in standard time-series regression this will show
up as an abnormal return, or an alpha. However, purgingthe empirical means of these two
effects implies that the ex-ante means lie on thesecurity market line.
In the special case of the world CAPM, equation (14) can be stated asE (rit+1|It, yit = 1,
yit+1 = 1) = E (rMt+1|It) [pitbi1 + (1 - pit) bi2] + gi1hit, (15)
where E (rMt+1|It) is the conditional risk premium on the world market portfolio,and the
betas are the world CAPM betas for the two regimes. As discussed above,taking account
of the peso problem requires measurement of bi2 and pit. In practise, estimating bi2 from
returns during a regime-switch is infeasible as there are very few, if any, in available
return data. Hence, in the empirical work, we will assumethat
bi1 = bi2 = biM. This gives us the following cross-sectional implications
E (rit+1|It, yit = 1, yit+1 = 1) = lMtbiM + gi1hit, (16)
where lMt = E (rMt+1|It). In the empirical work we also consider time-variation inbetas.
Allowing for this time-variation is straightforward and does not affect any of the
derivations above.
Finally, note that for high survival probabilities, the hazard rate in equation (13)is almost
linear in the probabilities. Under the assumption that the probabilitiesabout expropriation
in equity markets and sovereign debt markets are highly re-lated, it is straightforward to
show that pit can essentially be backed out from observed sovereign bond spreads (see
Appendix A). The premise that probabilitiesof bond default and expropriation in equity
markets are related is supported bythe events in Malaysia in 1998 and the more recent
events in Argentina. Note thatfor small default probabilities, the hazard rate is almost
linear in sovereign bondspreads. As discussed and documented later, at least for the few
sovereign spreads that we observe, the spreads are highly correlated with observed
measures of country ratings. Hence, we can use the more extensively available data on
country ratingsto measure the hazard rates themselves.
3. Data
We collect monthly return data on 46 developed and emerging markets from
Datastream.According to International Finance Corporation (IFC) of the World Bank, 21
of these markets are classified as developed and 25 as emerging markets. The
underlyingsources of the data are Morgan Stanley Capital International (MSCI)
fordeveloped markets and IFC for emerging markets. The returns from IFC are the
investable returns that incorporate foreign investment restrictions (including special
classes of shares, sector restrictions, single foreign shareholder limits,
restrictionsallowing only authorized investors, company statues, and national limits). We
also consider the return on the MSCI world market portfolio. All returns are in U.S.
dollars,
and excess returns are calculated by subtracting the one-month Eurodollar rate =pp for
each month.The sample period is January 1984 to November 2000. It is, however, well
known that many emerging markets only were accessible for international investors
beginning in the late 1980s and the early 1990s. This is reflected in our data base. Data
for emerging markets are included as and when they open up. We let the opening date of
an emerging market be the date when IFC begins to record investable returns. The
inclusion date for each market is shown in Table 1. The inclusion dates are similar to
what other studies have considered to be the financial market liberalization dates (see, for
instance, Kim and Singal, 2000, Bekaert and Harvey, 2000, and Henry,2000). Our
empirical results are not sensitive to using alternative choices of liberalizationdates. The
total number of observations for developed markets is 203 and for emerging markets the
number of observations varies between 90 and 144.In Table 1 we report summary
statistics of the monthly dollar returns. The averagereturns across developed and
emerging markets are about the same, 1.32%and 1.34% per month, respectively.
However, the average standard deviation ofemerging markets is about twice as high as
for developed markets. It also seemsto be greater dispersion in returns and return
volatilities of emerging economies.The correlation with the world market return is much
higher for developed markets
than for emerging markets.Table 2 presents information regarding various attributes of
the countries. Theseattributes are used in our cross-sectional analysis of risk premia. The
Real GDP perCapita attribute is the real GDP per capita in constant dollars in 1990
(expressed ininternational prices, base 1985). The Trading Activity attribute is the sum of
exportsand imports divided by GDP in 1990. The real GDP per capita and trading
activityattributes are collected from the World Penn Tables. The Economic Rating and
theFinancial Rating attributes refer to the average country ratings from inclusion dateto
November 2000, and is provided by the International Country Risk Guide (ICRG).
The economic risk rating is meant to measure an economy’s current strengths
andweaknesses, whereas the financial risk rating is meant to measure an
economy’sability to finance its official, commercial, and trade obligations. More
specifically,the variables determining the economic rating include a weighted average of
inflation,debt service as a percent of exports, international liquidity ratios, foreign
tradecollection experience, current account balance, and foreign exchange market
indicators.In the empirical work our measure of reputation is the financial rating, whichis
a weighted average of loan default, delayed payment of suppliers’ credit, repudiationof
contracts by government, losses from exchange controls, and expropriation of private
investment. The country ratings are published on a scale from 0 to 50 where a higher
number indicates lower risks. We have re-scaled the ratings to be
between 0 (low) and 100 (high). A rating of 0 to 49 then indicates a very high risk;50 to
59 high risk; 60 to 69 moderate risk; 70 to 79 low risk; and 80 or more verylow risk. The
country rating are used by Erb, Harvey, and Viskanta (1996) in theirstudy of the time-
series predictability of future returns. La Porta, Lopez-de-Silanes,Shleifer, and Vishny
(1998) use these ratings to study investor protection and ownership structure across
countries. In this paperwe use the ratings to measure sampleselectivity.
Finally, we report betas versus the MSCI world market portfolio. The betas are,on
average, about the same for developed and emerging markets. However, thedispersion in
betas is much larger across emerging markets ranging from 0.07 to1.80, whereas they are
all about one in the developed markets.
It is evident from Table 2 that the emerging economies are economies with relativelylow
GDP per capita. Further, emerging economies have a much lower countryratings than
developed economies. In fact, the correlation between the real GDP percapita and the
ratings are 70% (economic rating) and 80% (financial rating). Thetrading activity
attribute has a lower correlation with the real GDP per capita (about20%). The
correlations between trading activity and the ratings are about 20% and40%. There are a
few outliers (notably Hong Kong and Singapore), but excludingthem does not affect the
correlation between trading activity and credit rating significantly.We also collect
sovereign spreads for nine emerging economies from J.P. Morgan.2 These are economies
with Brady bonds (restructured dollar-denominateddebt). We argue that the country
ratings contain much of the cross-sectional information in the spreads. For each month,
we computed the correlation between the sovereign spreads and the composite country
ratings. The correlations variedfrom -95% to -46% with an average of -72%. That is,
sovereign nations with a highspread on their dollar-denominated debt tend to have a low
country rating. This isalso highlighted in Figure 1, which shows the spreads versus
country ratings afterthe averages of the variables for each month have been subtracted.
That is, the variablesare measured as deviation from month averages to sweep out time
effects. Thecorrelation is about -58% and is highly significant (a p-value close to zero).
Similar results are obtained with either financial or economic country ratings.Our sample
begins in 1984 for developed markets, and in the late 1980s andearly 1990s for emerging
markets. Consequently, only brief data histories are available,particularly for emerging
economies. This makes it difficulty to solely rely ontime-series methods for measurement
and statistical inference. For this reason, weextensively use pooled cross-sectional
methods in the estimation. Importantly, therelative rankings of the attributes do not vary a
lot over time, indicating that mostof the information is in the cross-section. We typically
rely on the time series to
2The nine economies are Argentina, Brazil, Colombia, Korea, Mexico, Peru, Poland,
Turkey, andVenezuela.estimate exposures to risk sources, but evaluates the asset pricing
implications inthe cross-section. Increasing the sample for developed markets (going
back to 1976)does not change our results qualitatively and are therefore not reported.In
some specifications we allow the beta of a market versus the world market portfolio to
vary according a conditional information variable, namely the world excess dividend
yield (i.e., the dividend yield on the world market portfolio in excess of the one-month
Eurodollar deposit rate). These series are collected from Datastream.
4. Estimation and Methodology
In this section we present the estimation approach and discuss testable implications in the
time series as well as in the cross-section. We employ the generalized methodof moments
(GMM) of Hansen (1982) to estimate all parameters simultaneously asin Cochrane
(2001), and similar to Bansal and Dahlquist (2000) and Jagannathan andWang (2002). In
this framework, specific distributional assumptions of the asset returnsare not required,
and wedo not need to work in a normally independently and identically distributed
setting. We can handle both conditional heteroskedasticityand serial correlation in pricing
errors. The approach is different from traditional approaches as we avoid the problem of
generated regressors, and it is not necessary
to develop further methods and corrections as in two-step procedures.We have to deal
with missing data as the dollar return series for emerging markets
handle the missing data as in Bansal and Dahlquist (2000). The idea is to balance the data
set, and then apply the asymptotic results in the standard GMM framework. This is
further discussed below. We are interested in estimating the risk exposures and risk
premia simultaneously.Consider N markets (i = 1, 2, . . . , N), each with T observations (t
= 1, 2, . . . , T).Recall that the emerging markets have different lengths of histories. We
describe the
estimation approach for the world CAPM with time-varying betas.As in Jagannathan and
Wang (1996), Cochrane (1996), amongst others, we evaluate
the implications of the model in the cross-section as their is considerable cross- sectional
variation in the mean returns. Consider the cross-sectional risk premium implications in
equation (16). In addition, allow for the market beta of an asset to be time-varying
according to biM + biMzzt, where zt is a variable known at time t capturing time
variation in the market beta. The implications for the cross-section
of unconditional mean excess returns can then be written as
E (rit+1) = lMbiM + lMzbiMz + gihi, (17)
where lM = E (lMt) and lMz = E (lMtzt). The biMs and biMzs are the standard
time series projection coefficients. Hence, our first sets of moment conditions, for
each market i, are
E [(rit+1 - aiM - biMrMt+1 - biMzrMt+1zt) yityit+1] = 0, (18)
E [(rit+1 - aiM - biMrMt+1 - biMzrMt+1zt) rMt+1yityit+1] = 0, (19)
E [(rit+1 - aiM - biMrMt+1 - biMzrMt+1zt) rMt+1ztyityit+1] = 0. (20)
These moment conditions are exactly identified. We have 3N moment conditionsand the
same number of parameters. The point estimates from these moment conditions
correspond to the usual least squares estimates. We follow the literature and add
constants, or alphas. In the world CAPM, the aiMs should be equal to zero.Indeed, we
will evaluate the CAPM by checking whether the alphas are all equal tozero in the time
series. Our focus, however, is on the ability of the various modelsh and without sample
selectivity) to explain the cross-section of risk premia.
Note that we use the regime indicator variable to make our unbalanced panel abalanced
panel as in Bansal and Dahlquist (2000). That is, the moment conditions are multiplied
with the product of the regime indicators at time t and t + 1, yityit+1. The product
yityit+1 selects returns when markets are open both at time t and t + 1. In essence, this
procedure treats missing observations as zeros. This has a practical advantage since the
usual moment conditions which contain missing data can be filled with zeros, and then
standard GMM routines can be utilized.3
The sample selectivity part in equation (17) is gihi. As noted in the discussion of equation
(14), under simplifying assumptions, the probability of default can be recovered from the
sovereign bond spread. Further, this spread can be used to 3Hayashi (2000) considers,
also in an analysis of panel data, a similar approach. Stambaugh (1997) presents an
alternative approach to address this econometric issue.
14
completely characterize the hazard rate at time t. However, the data on sovereign interest
rate spreads are not available for many economies in our sample period. As shown
earlier, there is a high negative correlation between the country ratings and the spreads in
the cross-section (for economies where sovereign spread data are available). That is, a
country with a low rating tends to have a high spread (a high probability of default).
Consequently, to characterize the cross-section of hazardrates, we model the hazard rate
for market i as follows
gihi = (g00 + g01Ci) Ai, (21)
where Ai proxies for hi in the cross-section. For example, we let Ai equal the countryis
economic rating which then captures the cross-sectional variation in the hazardrate.
Further, to allow for controlled cross-sectional heterogeneity in gi, we model it as
gi = g00 + g01Ci, where Ci denotes a country-specific attribute such as its return
volatility, financial rating, or its trading activity.
The cross-sectional parameters (i.e., the risk premium parameters and the g00and g01
parameters) are then identified in the last set of moment conditions for each asset i
E [(rit+1 - lMbiM - lMzbiMz - g00Ai - g01AiCi) yityit+1] = 0. (22)
We also consider a specification with a constant term
E [(rit+1 - l0 - lMbiM - lMzbiMz - g00Ai - g01AiCi) yityit+1] = 0. (23)
The constant term l0 should be zero according to theory, and a non-zero constant
indicates that a model cannot price the assets on average. Alternatively, a non-zero
constant can be interpreted as a zero-beta rate different from the riskfree rate thatis
imposed. Note that all parameters including the betas and the cross-sectional parameters
l0, lM, lMz, g00, and g01 are jointly estimated using GMM. Details of the estimation are
given in Appendix B.
Results
This section presents the empirical results. Recall, that we earlier reported that the cross-
sectional dispersion in the average returns is fairly large for emerging markets and small
for developed markets. This cross-sectional dispersion poses a serious challenge to asset
pricing models. Variables that characterize the selectivity bias, such as country ratings,
have very little time-series variation, but considerablecross-sectional variation. Hence,
the effects of selectivity are primarily identifiable in the cross-section. Given the large
cross-sectional dispersion in the data along with the short data histories for many
emerging markets, we, as in Black, Jensen, and Scholes (1972), Fama and MacBeth
(1973), and Jagannathan and Wang (1996), focusprimarily on the explaining the cross-
sectional differences in risk-premia.
We first discuss the ability of the various models to capture the cross-section of average
returns through only systematic risk. We then include sample selectivity in the cross-
section. Finally, we discuss the results and provide further interpretations
of the results.
Evidence in the Absence of Selectivity
In Table 3, we provide evidence from the cross-section of asset returns. The estimated
risk premium for the market portfolio is negative, as can be seen in row 2 of Panel A.
This is a standard finding (see, for instance, Jagannathan and Wang, 1996). The ability of
the CAPM with constant betas to explain the cross-section of average returns is basically
zero as indicated by the adjusted R-square. In Panel B, we consider the CAPM where the
market betas are allowed to be time-varying. The model fails to capture the cross-
sectional dispersion in average returns in this specification as well The adjusted R-square
is only about 8%. The theoretical restriction that l0 = 0 can be rejected at the 5%
significance level. However, the constant term, as discussed below, is not particularly
relevant when sample selectivity is included in
the model. For completeness, we have conducted the time-series tests for both the
constant beta and timevaryingbeta versions of the CAPM (not reported in a table). We
find that the joint test of zero alphasis rejected in both cases. The rejections seem to be
primarily due to abnormal returns in emergingmarkets—this is consistent with Harvey
(1995) who also shows that CAPM implications are rejectedin emerging markets data.
The failure of the CAPM can also be seen in Figure 2 where we plot the averagereturns
against the predicted expected returns from the model. A true model would,ignoring
estimation errors, produce observations along the 45-degree line. The figure reveals that
there is almost no dispersion in predicted expected returns. Hence, the model does not
capture the large cross-sectional variation in average returns.
Evidence with Selectivity Included
The model specifications with sample selectivity are also reported in Table 3. Inthese
specifications the selectivity is modelled as (g00 + g01si)Ai, where si is the annual return
volatility for market i, and Ai is defined as the economic rating for country i less the
economic rating for the U.S. The expression (g00 + g01si) is the gi for country i. The
proxy for the hazard rate for country i is Ai, based on thereasoning provided earlier. Note
that Ai is negative for emerging economies and close to zero for developed economies.
This specification captures the intuition that as economies improve their economic rating
they become akin to developed marketsand the sample selectivity term would fall.When
sample selectivity is incorporated in the standard CAPM (in Panel A), thecross-sectional
R-square rises to 41% and the parameters associated with the selectivityterm are
significant (a p-value of 3%). The time-varying beta based CAPMwith sample selectivity
included is reported in Panel B. This specification does quitewell in capturing the cross-
sectionalvariation in risk premia, and has an adjustedR-square of 61%. The magnitudes of
the parameters that govern the selectivity bias(that is, g00 andg01) are similar across
different specifications. They are in all casesjointly significant at usual significance levels
(see the column labelled “Test ofJointSignificance”). The last two rows of Panel B
highlights the relevance, or the lackthereof, of the constant term l0. The empirical results
across thetwo cases (includingl0, or not) are comparable. Hence, as suggested by theory,
the constant term isnot particularly important.Table 4 provides themagnitudes of the
overall risk premia explained by systematicrisk and by sample selectivity. Economies
with poor economic rating have alarger andpositive selectivity bias. For developed
economies the variable Ai is essentiallyzero and hence the effect of selectivity on their
mean returns is absent.
the constant beta CAPM, the systematic risk contribution is about 0.45% per month for
both emerging and developed economies. However, the selectivity premium is0.50% per
month for emerging markets and close to zero for developed markets. Inthe model with
time-varying betas (see PanelB),the fraction of the emerging marketreturn attributed to
the selectivity bias is somewhat higher, and now stands at 0.58%per month. For
emergingmarkets more than 1/2 of the ex-post risk premium can beattributed to
selectivity. That is, sample selectivity seems to be the dominant influenceon the measured
risk premiums in emerging economies. Sample selectivity isnot an important dimension
for understanding measured risk premiain developedmarkets.The higher explanatory
ability of the world CAPM with time-varying betas andsample selectivity can be seen in
Figure 3which displays the average returns againstpredicted expected returns. The
improvement in fit is visible and the model is ableto produce the highdispersion in
average returns.We also considered alternative specifications for the parameter gi. In
particular,we replaced si with a reputationalvariable—the financial rating of an economy
i lessthe comparable rating for the U.S. The ability of this specification in terms of
capturingthecross-sectional variation in risk premia (i.e., adjusted R-square) is about30%.
This R-square is quite high relative the specifications withouttheselectivityeffects. As
shown in Table 4 the average emerging markets risk premium is still predominantlydue
to selectivity bias. Yet anotherchoice for the specification of gi, thetrading activity
variable, produces again similar results.Finally, we consider a specification where we use
thespreads (short samplesavailable for nine economies) on the Brady bonds to measure
the hazard rates directly.The approach is as follows. We projecttheaverage spreads on the
averagecountry ratings and use these projection coefficients to infer spreads and
defaulprobabilities (under theassumption of zero recovery as in Appendix A) for
allemerging markets. Recall that we are making the assumption that the probabilityof
default indebt markets coincides with the probability of expropriation. Fromthe
probabilities we can then compute the hazard rates (i.e., the his). For developedmarkets
we assume a zero default probability (and hence a zero hazard rate). Withthis measure of
the hazard rate, and the same specificationforthe gi as before, we estimate the cross-
sectional regression in equation (22). This specification capturesabout 36% of the cross-
sectional variation in risk premia. We find thatthe selectivity18term, that is gihi, is about
the 0.63% per month. The average probability of the defaultis about half a percent per
month. Hence, the bias in the mean return of about 0.63% per month can be supported by
rather small probability of default (risk of expropriation).Note that the empirical evidence
for this specification is quite similarto that discussed in the time-varying beta case in
Panel B of Table 3.
5.3. What Drives the Selectivity Bias?
In Panel A of Table 5 we inquire what economic variables can explain the
crosssectionaldispersion in the selectivity bias for emerging markets. In particular, weare
interested in whether the measured selectivity premium is related to trading activityand/or
measures of reputation. To do so, we considerthe measures of theselectivity premium
based on the specification where gi = (g00 + g01si), and therelative economic rating is
the proxy for thehazard rate. This specification was reportedin Table 3. The reputational
variable (the financial rating of country i lessthe comparable rating for theU.S.)is able to
explain about 32% of the dispersion inthe selectivity premium. This regression also
shows that the selectivity premiumrises as thecountry’s financial rating falls. Similarly,
when we use the tradingactivity variable, this explains about 19% of the dispersion in the
selectivitypremium.Economies with larger trading activity have a smaller selectivity
premium.In essence our evidence suggests that both trading activityand reputational
considerationsare important for explaining the selectivity premium.Allowing the
selectivity premium to depend on the volatilityinthe cross-sectionis motivated by
arguments presented in Brown, Goetzmann, and Ross (1995). Ourevidence indicates that
this attribute is notuniquely important to capture the crosssectionaldifferences in risk
premia. Indeed, the trade activity and financial reputationvariables do, at least
ineconomic terms, a comparable job of explaining the cross-sectional differences in the
risk premia. Thus, it seems to us that this is dueto the fact that the volatility of returns are
related to these variables.Thisis shownin Panel B of Table 5: return volatility is
decreasing in both trading activity andfinancial reputation. These variables, based on the
work of Eaton and Gersovitz(1981), and Bulow and Rogoff (1989a, 1989b) should matter
to the compensationthat emerging markets have toadditionallypay, due to risks of
expropriation. Wefind that this indeed is the case.
In this paper we show that the cross-sectional differences in the equity returns
acrosssovereign economies is determined by two features—systematic risk and a
selectivitypremium. We show that the selectivity premium captures more than 1/2 of
theaverage risk premium in emerging markets. The equity riskpremia in
developedmarkets seems to be driven solely by systematic risk. The main economic
implicationof this result is that after taking account ofselectivity premium all
internationalequity returns reflect systematic risk, as predicted by theory.Our empirical
work also shows that sovereigns thathave better financial marketreputations and trade
more actively have to pay a smaller selectivity premium. Thisempirical evidence lends
support to theview that both reputations and fear of tradesanctions are important in
determining the cost of equity borrowing for a sovereignnation.
Measuring Hazard Rates From Sovereign Spreads
This Appendix shows how the hazard rate can be measured from sovereign bondspreads.
Consider a dollar denominated pure discount bond issued by a country.The payoff is
equal to one if there is no default, and µb + bbet+1 +hbt+1 if the countrydefaults. The
payoff process can thus be written as qbt+1= ybt+1 + (1 - ybt+1) (µb + bbet+1 + hbt+1) .
(24)
For simplicity, we assume that bb = 0. That is, we assume that the recovery valueof the
bond is not related to the systematic risk in the world economy.Further, theexpected
payoff on this bond in default is less than one (i.e., µb < 1). Valuing thispayoff using the
stochastic discount factor implies that
1/Rbt = [pbt + (1 - pbt) µb] /Rf t. (25)
Solving for the probability of no default, we obtain
pbt =Rf t - µbRbtRbt (1 - µb). (26)
Assume that the probabilities of default for the bond correspond to the probabilityof a
market shut-down, that is, pbt = pit. Under the further assumptionthat therecovery rate is
zero, we can directly recover the probability of default. Further,given the normal
cumulative distribution function wecan completely characterizethe hazard rate.The above
expression can also be used to compute the ex-ante beta on market i.We denote this with
bit =pitbi1 + (1 - pit) bi2. If we assume that the ratio of thebetas across the two regimes is
equal to a constant c and µb = 0, it follows that
bit = bi1 _Rf tRbt + c _Rbt - Rf tRbt __. (27)
Note that Rf t and Rbt can be observed directly from U.S. Treasuries and
Sovereignbonds, or as we demonstrate, approximated with a country’s relativecountry
rating.Hence, conditional on c, one can estimate the model with both a peso problem
andsample selectivity. In the special case with c = 1,there is no peso problem and wehave
that bit = bi1 = bi2.21B. Estimation DetailsThis Appendix shows the estimation in more
detail. Let q0 denote thetrue parametervector that we want to estimate. The typical
elements in q0 are aiM, biM, and biMz that are specific to each market, and the common
parameters l0, lM, lMz, g00and g01. By stacking the sample counterparts of the moment
conditions in (18) to (20), and (23), we have a vector of moment conditions
gT (q) =1T Tåt=1f (Xt, q) , (28)
where Xt summarizes the data used to form the moments conditions. The vectorgT (q)
has the dimension 4N. The moment conditions, given by (18) to (20), exactlyidentify the
aiM, biM, and biMz parameters. However, the moment conditions, givenby (23), is
overidentified. We have N moment conditions, but only 5 parameters (l0,lM, lMz, g00
and g01).We estimate the parameters by setting linear combinations of gT equal to zero.
That is, the moment conditions can be written as
ATgT = 0, (29)
where AT is a (3N + 5) × 4N matrix. In particular, our choice of AT is designed toensure
that the point estimates are the ones given by ordinary least squares.
Let AT
be the product of two matrices denoted by A1T and A2T (that is, AT = A1TA2T).
Thefollowing matrices result in least square point estimates
A1T =26666666664I3N 03N · · · 03N003N 1 · · · 1003Nˆ b1M · · · ˆ bNM003Nˆ b1Mz ·
· · ˆ bNMz003N A1 · · · AN003N C1A1 · · · CNAN, (30)
where I3N is the identity matrix with dimension 3N, 03N is a 3N vector of zeros,0N is an
N vector of zeros, and A2T is a diagonal matrix with typical element equalto 1/ åTt=1
yit+1. The ˆ biMs and ˆ biMzs are estimates of biMs and biMzs, and they are
given in the estimation. The ˆ biMs and ˆ biMzs are exactly the least square
estimatesobtained in a regression of the assets’ excess returns on the market excess
returnand scaled market excess returns as in (18) to (20). Further, the estimates of l0,
lM,lMz, g00, and g01 coincide with the least square estimates obtained in a regressionof
average returns on the betas and the proxies for sample selectivity. Our choice ofAT
ensures that ATgT (qT) = 0.
Based on Hansen (1982) we know that when linear combinations of gT are setequal to
zero as in (29), the asymptotic distribution of the point estimator qT is given
by
pT (qT - q0) d
! N _0, (A0D0)-1 _A0S0A00_(A0D0)-10_, (31)
where D0 is the gradient of the moment conditions in (28), and where S0 is the
variance-covariance matrix of the moment conditions and given by
S0 =¥åj=-¥E hf (Xt, q0) f _Xt-j, q0_0i. (32) The sample counterpart ST is estimated
using the procedure in Newey and West(1987) with four lags. D0 and A0 can be
estimated by their sample counterparts DTand AT. Note that the standard errors based on
(31) are robust to heteroskedasticity
and serial correlation in the moment conditions.
Table 1: Summary Statistics of Global Equity Returns
Mean Standard Deviaiion coreallation With world nclusion
Panel A. Developed Markets
Australia 1.07 6.84 0.52 84-01 203
Austria 1.16 7.33 0.34 84-01 203
Belgium 1.60 5.55 0.64 84-01 203
Canada 0.99 5.13 0.70 84-01 203
Denmark 1.18 5.66 0.53 84-01 203
Finland 1.91 8.62 0.54 88-01 156
France 1.59 6.07 0.70 84-01 203
Germany 1.38 6.27 0.60 84-01 203
Hong Kong 1.84 8.76 0.53 84-01 203
Ireland 1.03 5.73 0.65 88-01 156
Italy 1.46 7.51 0.51 84-01 203
Japan 1.02 7.36 0.76 84-01 203
Netherlands 1.56 4.73 0.75 84-01 203
New Zealand 0.30 7.02 0.47 88-01 156
Norway 1.09 7.26 0.58 84-01 203
Singapore 0.85 8.05 0.54 84-01 203
Spain 1.81 7.03 0.66 84-01 203
Sweden 1.65 6.92 0.63 84-01 203
Switzerland 1.51 5.38 0.66 84-01 203
U.K. 1.33 5.41 0.76 84-01 203
U.S. 1.38 4.37 0.79 84-01 203
Average 1.32 6.52 0.61
Panel B. Emerging Markets
Argentina 3.89 23.46 0.06 89-01 144
Brazil 3.16 19.61 0.31 89-01 144
Chile 1.83 7.73 0.24 89-01 144
China 0.26 13.41 0.27 93-01 96
Colombia 1.43 10.91 0.10 91-03 118
Greece 2.18 12.22 0.19 89-01 144
Hungary 1.61 13.15 0.47 93-01 96
India 0.36 8.68 0.13 92-12 97
Indonesia -0.08 15.13 0.41 90-10 123
Jordan 0.69 4.85 0.22 89-01 144
Korea 0.37 14.15 0.39 92-02 107
Malaysia 0.23 10.12 0.39 89-01 116
Mexico 2.04 10.18 0.42 89-01 144
Pakistan 0.99 12.56 0.08 91-04 117
Peru 0.75 9.14 0.34 93-01 96
Philippines 0.42 11.42 0.40 89-01 144
Poland 3.24 17.91 0.37 93-01 96
Portugal 0.96 6.91 0.51 89-01 144
South Africa 0.98 8.34 0.53 93-01 96
Sri Lanka -0.22 10.07 0.31 93-01 96
Taiwan 0.72 10.50 0.37 91-02 119
Thailand 0.38 12.55 0.43 89-01 144
Turkey 2.48 19.39 0.16 89-09 136
Venezuela 3.00 17.43 0.02 90-02 131
Zimbabwe 1.80 12.81 0.21 93-07 90
Average 1.34 12.50 0.29
Panel C. World
World 1.21 4.24 1.00 84-01 203
This table presents summary statistics of monthly dollar returns in global equity markets
from nclusion date to November 2000. Panels A, B and C show statistics for developed
markets, emrging markets and theWorld, respectively. The labels Average in Panels A
and B refer to theaverage (equally-weighted) across developed and emerging markets,
respectively. The meansand standard deviations are expressed in % per month.
Correlation with World refers to thecorrelation coefficient with the world market
portfolio. The inclusion date (year-month) is thefirst month with observations of
investable returns. The last observation of Malaysia is August1998. T refers to the
number of observations for each market.
CHAPTER.3
RISK AND RETURN: PORTFOLIO THEORY AND ASSET
PRICING MODELS
a. A portfolio is made up of a group of individual assets held incombination. An asset that
would be relatively risky if held inisolation may have little, or even no risk if held in a
welldiversifiedportfolio.
b. The feasible, or attainable, set represents all portfolios that canbe constructed from a
given set of stocks. This set is onlyefficient for part of its combinations.
c. An efficient portfolio is that portfolio which provides the highestexpected return for
any degree of risk. Alternatively, the efficient portfolio is that which provides the lowest
degree of risk for any expected return.
d. The efficient frontier is the set of efficient portfolios out of thefull set of potential
portfolios. On a graph, the efficientfrontier constitutes the boundary line of the set of
potentialportfolios.
e. An indifference curve is the risk/return trade-off function for aparticular investor and
reflects that investor's attitude towardrisk. The indifference curve specifies an investor's
required rate of return for a given level of risk. The greater the slope of theindifference
curve, the greater is the investor's risk aversion.
f. The optimal portfolio for an investor is the point at which theefficient set of portfolios--
the efficient frontier--is just tangentto the investor's indifference curve. This point marks
the highest level of satisfaction an investor can attain given the set ofpotential portfolios.
g. The Capital Asset Pricing Model (CAPM) is a general equilibriumnmarket model
developed to analyze the relationship between risk andrequired rates of return on assets
when they are held in welldiversifiedportfolios. The SML is part of the CAPM.
h. The Capital Market Line (CML) specifies the efficient set of portfolios an investor can
attain by combining a risk-free asset and the risky market portfolio M. The CML states
that the expected return on any efficient portfolio is equal to the riskless rate plusa risk
premium, and thus describes a linear relationship between expected return and risk.
i. The characteristic line for a particular stock is obtained byregressing the historical
returns on that stock against thehistorical returns on the general stock market. The slope
of thecharacteristic line is the stock's beta, which measures the amountby which the
stock's expected return increases for a given increasein the expected return on the market.
j. The beta coefficient (b) is a measure of a stock's market risk. Itmeasures the stock's
volatility relative to an average stock, whichhas a beta of 1.0.
k. Arbitrage Pricing Theory (APT) is an approach to measuring theequilibrium risk/return
relationship for a given stock as a functionof multiple factors, rather than the single factor
(the market return) used by the CAPM. The APT is based on complex mathematicaland
statistical theory, but can account for several factors (such asGNP and the level of
inflation) in determining the required return
for a particular stock.
l. The Fama-French 3-factor model has one factor for the excess marketreturn (the market
return minus the risk free rate), a second factor
for size (defined as the return on a portfolio of small firms minusthe return on a portfolio
of big firms), and a third factor for thebook-to-market effect (defined as the return on a
portfolio of firmswith a high book-to-market ratio minus the return on a portfolio offirms
with a low book-to-market ratio).
m. Most people don’t behave rationally in all aspects of their personallives, and
behavioral finance assume that investors have the same
types of psychological behaviors in their financial lives as intheir personal lives.
Security A is less risky if held in a diversified portfolio because ofits lower beta and
negative correlation with other stocks. In a
single-asset portfolio, Security A would be more risky because sA > sBand CVA >
CVB.The intercept, a, seems to be about 3.5. Using a calculator with atleast squares
regression routine, we find the exact equation to be
X k = 3.7 + 0.56 M k , with r = 0.96.
b. The arithmetic average return for Stock X is calculated as follows:
%. 6 . 10
7
2 . 18 ... 0 . 23 0 . 14 (
kAvg =
+ + + -
=
The arithmetic average rate of return on the market portfolio,determined similarly, is
12.1%.
.6
Several points should be noted: (1) sM over this particular periodis higher than the
historic average sM of about 15 percent,indicating that the stock market was relatively
volatile during thisperiod; (2) Stock X, with sX = 13.1%, has much less total risk thanan
average stock, with sAvg = 22.6%; and (3) this example demonstrates that it is possible
for a very low-risk single stock to have lessrisk than a portfolio of average stocks, since
sX < sM.
c. Since Stock X is in equilibrium and plots on the Security MarketLine (SML), and
given the further assumption that X X k kˆ = and
M M k kˆ = --and this assumption often does not hold--then this equationmust hold:
Since sp is only 62 percent of sM, the probability distribution for Condition 2 is clearly
more peaked than that for Condition 3; thus,we can be reasonably confident of the
relevant locations of the distributions for Conditions 2 and 3.With regard to Condition 1,
the single-asset portfolio, we can besure that its probability distribution is less peaked
than that forthe 100-stock portfolio. Analytically, since b = 0.62 both for thesingle stock
portfolio and for the 100-stock portfolio,
. 0 ) 62 . 0 ( ) 62 . 0 ( 2
p
2
M
2
e
2
M
2
Y s » + s > s + s = s
We can also say on the basis of the available information that sY is smaller than sM;
Stock Y's market risk is only 62 percent of the"market," but it does have company-
specific risk, while the market portfolio does not. However, we know from the given data
thatsY = 13.8%, while sM = 19.6%. Thus, we have drawn the distribution or the single
stock portfolio more peaked than that of the market.The relative rates of return are not
reasonable. The return for any
stock should be
ki = kRF + (kM - kRF)bi.
Stock Y has b = 0.62, while the average stock (M) has b = 1.0;
therefore,
kY = kRF + (kM - kRF)0.62 < kM = kRF + (kM - kRF)1.0.
A disequilibrium exists--Stock Y should be bid up to drive its yielddown. More likely,
however, the data simply reflect the fact thatpast returns are not an exact basis for
expectations of future returns.
Portfolio Theory
Suppose Asset A has an expected return of 10 percent and a standard
deviation of 20 percent. Asset B has an expected return of 16 percent and a
standard deviation of 40 percent. If the correlation between A and B is 0.6, what
are the expected return and standard deviation for a portfolio comprised of 30
percent Asset A and 70 percent Asset B?
Portfolio Expected Return
Portfolio Standard Deviation
%.2.14142.0
)16.0(7.0)1.0(3.0
r̂)w1(r̂wr̂ BAAAP
309.0
)4.0)(2.0)(4.0)(7.0)(3.0(2)4.0(7.0)2.0(3.0
)W1(W2)W1(W
2222
BAABAA2B
2A
2A
2Ap
Attainable Portfolios AB = 0.4
AB = +0.4: Attainable Set of
Risk/Return Combinations
0%
5%
10%
15%
20%
0% 10% 20% 30% 40%
Risk, p
Ex
pe
cte
d r
etu
rn
Attainable Portfolios: rAB = +1
AB = +1.0: Attainable Set of Risk/Return
Combinations
0%
5%
10%
15%
20%
0% 10% 20% 30% 40%
Risk, p
Ex
pe
cte
d r
etu
rn
Attainable Portfolios: rAB = -1
AB = -1.0: Attainable Set of Risk/Return
Combinations
0%
5%
10%
15%
20%
0% 10% 20% 30% 40%
Risk, p
Ex
pe
cte
d r
etu
rn
Attainable Portfolios with Risk-Free Asset (Expected risk-free return = 5%)
asible and Efficient Portfolios
Attainable Set of Risk/Return Combinations with Risk-Free Asset
0%
5%
10%
15%
0% 5% 10% 15% 20%
Risk, p
Exp
ecte
d r
etu
rn
Feasible and Efficient portfolios
The feasible set of portfolios represents all portfolios that can be constructed from a
given set of stocks.
An efficient portfolio is one that offers:
the most return for a given amount of risk, or
the least risk for a give amount of return.
The collection of efficient portfolios is called the efficient set or efficient frontier.
5 - 11
IB2 IB1
IA2IA1
Optimal PortfolioInvestor A
Optimal PortfolioInvestor B
Risk p
ExpectedReturn, rp
Optimal Portfolios
An investor’s optimal portfolio is defined by the tangency point between the
efficient set and the investor’s indifference curve.
Indifference curves reflect an investor’s attitude toward risk as reflected in
his or her risk/return tradeoff function. They differ among investors because of
differences in risk aversion.
What are the assumptions of the CAPM?
Investors all think in terms of single holding period.
All investors have identical expectations.
Investors can borrow or lend unlimited amounts at the risk-free rate.
All assets are perfectly divisible.
There are no taxes and no transactions costs.
All investors are price takers, that is, investors’ buying and selling won’t
influence stock prices.
Quantities of all assets are given and fixed.
What impact does RF have onthe efficient frontier?
When a risk-free asset is added to the feasible set, investors can create
portfolios that combine this asset with a portfolio of risky assets.
The straight line connecting rRF with M, the tangency point between the line
and the old efficient set, becomes the new efficient frontier.
What is the Capital Market Line?
The Capital Market Line (CML) is all linear combinations of the risk-free
asset and Portfolio M.
Portfolios below the CML are inferior.
The CML defines the new efficient set.
All investors will choose a portfolio on the CML.
5 - 24
Run a regression line of past returns on Stock i versus returns on the market.
The regression line is called the characteristic line.
The slope coefficient of the characteristic line is defined as the beta coefficient.
How are betas calculated?
5 - 25
Illustration of beta calculation
Year rM ri
1 15% 18%2 -5 -103 12 16
ri
_
rM
_-5 0 5 10 15 20
20
15
10
5
-5
-10
.
.
.
ri = -2.59 + 1.44 kM^ ^
5 - 28
Interpreting Regression Results
The R2 measures the percent of a stock’s variance that is explained by the market. The typical R2 is:0.3 for an individual stock
over 0.9 for a well diversified portfolio
5 - 30
2 = b2 2 + e2.
2 = variance= stand-alone risk of Stock j.
b2 2 = market risk of Stock j.
e2 = variance of error term
= diversifiable risk of Stock j.
The relationship between stand-alone, market, and diversifiable risk.
j j M j
j
j
j M
5 - 31
Beta stability tests
Tests based on the slope of the SML
Two potential tests that can be conducted to verify the CAPM?
5 - 33
Betas of individual securities are not good estimators of future risk.
Betas of portfolios of 10 or more randomly selected stocks are reasonably stable.
Past portfolio betas are good estimates of future portfolio volatility.
5 - 37
The CAPM is a single factor model.
The APT proposes that the relationship between risk and return is more complex and may be due to multiple factors such as GDP growth, expected inflation, tax rate changes, and dividend yield.
The difference between the CAPM and the Arbitrage
Pricing Theory (APT)?
5 - 38
ri = rRF + (r1 - rRF)b1 + (r2 - rRF)b2
+ ... + (rj - rRF)bj.
bj = sensitivity of Stock i to economicFactor j.
rj = required rate of return on a portfoliosensitive only to economic Factor j.
Required Return for Stock i under the APT
5 - 39
The APT is being used for some real world applications.
Its acceptance has been slow because the model does not specify what factors influence stock returns.
More research on risk and return models is needed to find a model that is theoretically sound, empirically verified, and easy to use.
What is the status of the APT?
5 - 41
Fama-French 3-Factor Model (Continued)
the return on, H, a portfolio of firms with high book-to-market ratios (using market equity and book equity) minus the return on L, a portfolio of firms with low book-to-market ratios. This return is called rHML, for H minus L.
5 - 42
ri = rRF + (rM - rRF)bi + (rSMB)ci + (rHMB)di
bi = sensitivity of Stock i to the market return.
cj = sensitivity of Stock i to the size factor.
dj = sensitivity of Stock i to the book-to-market factor.
Required Return for Stock i under the Fama-French 3-Factor Model
Markowitz Portfolio Theory
Combining stocks into portfolios can reduce standard deviation below the level
obtained from a simple weighted average calculation.
Less than perfect correlation coefficients make this possible.
The various weighted combinations of stocks that create this standard deviations
constitute the set of efficient portfolios.
FIXED INCOME – AN EVOLUTION IN RISK AND
RETURN
Declining bond yields and a flat yield curve, driven by fallinginflation, have dominated
returns from traditional fixed income over the last few years. At the same time, the
growth of thecorporate bond market has meant a greater focus on credit. Nolonger are the
returns from taking on credit risk seen as secondary.Instead credit has become a
legitimate asset class in its own right.These developments have opened up opportunities
for clientsto receive alternative sources of return by accepting different types of risk from
some of the newer fixed income classes.
Rewards from new sources of risk
In the past, investors’ fixed income return was mainly from government
bonds.Consequently, their exposure was primarily to interest rate risk, receiving a
premiumas compensation for this exposure. In recent years, new fixed income securities
rewardinvestors with returns from other types of risks. For example, corporate bonds
generatereturns by taking on credit risk; convertible bonds produce returns from credit
and equityrisk; and hybrid securities can source returns from credit risk, correlation,
liquidity risk,and equity risk. The chart below shows the different risks associated with
the newer fixed
income securities and how interest rate risk has diminished as a prime contributor to
riskand return. Some securities comprise an aggregate of risks, whereas others have only
oneor two risk types like credit default swaps (CDS).
Risks associated with differing fixed income securities
Customised credit risk via CDOs
The evolution of fixed income securities such as collateralised debt obligations
(CDOs)allows investors exposure to customised credit risk. CDOs are interest-bearing
securitiescomprising a portfolio of credit risk. The portfolio is made up of physical
corporate bonds or synthetic credit positions using credit default swaps or a combination
of both. This poolof credit assets is then securitised, similar to a mortgage pool and made
into tranches.
Risk
Each tranche is given an S&P rating ranging from AAA to sub-investment grade
(BBB-).Investors can buy into one or more of these tranches, thereby customising their
creditrisk. The riskiest or sub-investment grade tranche is exposed to the first losses if
thereis a default, while the AAA tranche is subject to the last default. Investors in the
riskier tranches are compensated by a higher income than investors in the highly rated
tranches.Historical data has shown that an investor is paid an average of around 1%
above cash forholding investment grade credit, 2% for sub-investment grade, for
example, anything lessthan BBB- and 3% above cash for emerging market debt.
CDSs help unbundle risk
While the advent of new investment risks has generated return opportunities forinvestors,
the development of derivative markets has permitted the unbundling of theserisks to
allow more effective investment strategies. A credit default swap (CDS) is a newtype of
credit derivative contract between two parties that separates and transfers thecredit risk of
an asset such as a corporate bond from one party to another. For example,
investors may previously have purchased a corporate bond and consequently
beenexposed to interest rate risk and credit risk. Now instruments such CDSs allow
investment managers to extract and hedge the unwanted risks such as the interest rate risk
givinginvestors exposure to only credit risk. CDSs are commonly used to leverage a
portfolio’s exposure to credit with the aim of adding value.
The benefits of correlation risk
Analysis of credit risks and returns (excluding traditional bond interest rate risk)
fromemerging market debt, corporate bonds and high yield securities reveals that there
arebenefits in diversifying between different types of credit securities due to low
correlations.These correlations are unlike those from traditional bond risks, in that the
investor is notbenefiting from different interest rates but from varying credit risks.
However, portfoliosof credit risk may reduce the diversification benefit between bonds
and equities as creditrisks are modestly correlated to equity performance, posing new
challenges in strategicasset allocation.
An active approach to risk management
As risks within fixed income markets evolve, the skills and methods managers use
togenerate outperformance from these risks have also developed. Investment
managersnot only aim to add value from credit through relative value sector and stock
selection,but by taking an active approach to risk management and avoiding defaults.
Because default costs are real, issuers are forced by the market to a pay a higher spread to
compensate for the chance of a default over and above the risk free rate. There is
anopportunity to capture this excess return.With an active approach to risk management,
default risk can be minimised, potentiallydoubling an investor’s potential excess return.
The lighter section on the chart below,
called the default cost, is the percentage of the amount that can erode an investor’s excess
return. As the quality of credit securities fall, the value of active stock selection and
diversification, incorporating credit analysis, rises.
Premiums earned in different credit categories
Fixed income – an evolution
A greater focus on credit has seen an evolution in the way investors source their return
fromfixed income as well as the type of securities they invest in.Risks have also changed
alongwith the development of techniques to manage these risks. With complex new
securitiessuch as hybrids, floating ratenotes and more recently CDOs and CDSs, the need
for activemanagement and experienced resources is key, with the aim of adding value for
investors.
Portfolio Theory
Capital Asset Pricing Model (CAPM)
Efficient frontier
Capital Market Line (CML)
Security Market Line (SML)
Beta calculation
Arbitrage pricing theory
Fama-French 3-factor model
Chapter 4
Expropriation Risk and Return in Global Equity Markets
So if the next 72.4 years exactly duplicates the period evaluated, you have less than a
penny to spare if your expenses remain the same and you choose to invest solely in T-
Bills. However, if you use a more realistic three month holding period for T-bill assets
with a 100% return of principal, you are left with an amount that is slightly under $1.00
(0.9924). Without inflation the picture is much rosier; an initial investment in T-Bills that
is left to compound monthly (rather than a quarterly compounding) will provide the
investor with an account valued at more than 15 times the initial amount
BondRisk
Perhaps a conservative investor decides that by simply replacing the T-Bill investment
with a longer maturity Treasury Note, all will be fine regardless of inflation. Although T-
Note yields are usually higher than T-Bills, periods where the yield curve becomes flat or
inverted highlight a different issue. The timing of the purchase for the longer term
security will impact investment returns. This brings us the next investor risk—interest
raterisk.
Interest rate risk is the risk of having your money locked into a lower rate of return while
interest rates rise. Assuming the asset is held to maturity to benefit from the 100% return
of principal, the longer the term to maturity for the fixed income investment, the more
opportunity there is to receive sub-standard interest rates on the money. As a result, the
10-year note is deemed to have more interest rate risk than the 90-day bill, while the 30-
year bond is deemedto have more interest rate risk than both the note
andthebill.
Using 10-year note returns starting in April 1953 and assuming 10-year holding periods
with inflation, a $1.00 initial investment will be valued at $2.65 after 639 months (53
years). A conservative investor may feel a 265% return is acceptable; however, we have
yet to account for taxes. Incorporating a 20% tax applied semi-annually results in a final
valueof$2.21.
An investor who seeks even better fixed income returns by entering the municipal or
corporate bond world encounters yet another type of investment risk: credit risk.
Although defaults are less likely in the municipal bond market (reflected by lower
yields), they can occur. This type of investment does benefit from favorable tax treatment
whichwillboostreturns.
In terms of corporate bonds, credit risk increases (reflected by higher yields) as the credit
rating decreases. AAA corporate bonds are deemed less risky from a return of principal
standpoint and, therefore, offer lower yields. Interest is taxable and there’s no guarantee
that the credit rating and safety of the bond downgraded during the holding
period.
At this time, historical municipal and corporate bond returns will not be analyzed. Those
interested may want to pursue this analysis using a bond fund or exchange traded fund
(ETF) proxy such as a Nuveen Municipal Fund or the iSharesâ Corporate Bond Fund. The
biggest challenges include obtaining sufficient return histories and incorporating
appropriate tax impacts to the result.
Stock Market Risk
Investors generally turn to the stock market to realize returns that will outpace inflation
after accounting for taxes. Modern Portfolio Theory (MPT) serves as one basis for
portfolio construction aimed at maximizing reward while minimizing risk. This approach
makes use of the following:
1) Capital Market LineRepresents optimal portfolios for a given level of
risk.
2)Capital Asset Pricing Model
(CAPM)
Assesses the risk-return impact of adding a security
into a well diversified portfolio.
3) Security Market LineA linear asset price model based on risk versus
returns.
4) Security Characteristic Line
A linear model of asset return versus market
return, using an asset risk measure to identify
undervalued and overvalued securities.
Beta, alpha and Sharpe Ratios are all measurements associated with portfolio and security
risk.
Inflation risk represents the risk associated with insufficient returns (not keeping pace
with increasing costs), while credit risk and stock market risk highlight represent the risk
associated with seeking higher returns (losses). Investors generally seek maximize returns
for a given level of risk by diversifying their portfolios (see previous articles on
diversification and correlations). However, market risk cannot be diversified away—
some risk of losseswill alwayremain even in optimal portfolios.
A two-part personal finance article series from May & June this year examined the
impact of inflation and investment returns on retirement savings when income was
removed from the account on a monthly basis. A thirty year period was examined starting
with January 1962 and investment returns were based upon actual S&P 500
returns, including dividends.
Using the same monthly inflation, return and dividend data, a $1 initial investment and a
20% tax rate on gains applied quarterly, the ending balance after 30 years was
approximately $1.26, or 126%. Remove the dividend returns and the account is valued at
$1.11 after thirty years. This represent a nice difference from compounded fixed income
returns over a 70 year period, particularly since the return of 0.12% excluded taxes.
Although the 10-year note return of 221% over 53 years seems to be an argument for T-
Note investments, when using thirty year results from initiation of the strategy, $0.96
remained in the account. Again, due to interest rate risk, the month of initiation will
impact the results with different thirty year periods yielding different returns.
InvestmentRisk
Investment risk includes credit risk and market risk which are both taken to combat
another risk—that due to inflation. Investors and traders need to assess the actual historic
results of the strategies they employ for their investments. In this manner, the individual
can better understand the true financial risks in which they are exposing themselves.
In order to manage necessary market risk, the investor must take steps to understand
those risks and minimize them through diversification. Numerous articles on the topic of
asset allocation and diversification are available in the Optionetics.com article archives
by completing an author keyword search for 2004 and 2005. Traders, who often self-
direct their investment accounts, must proactively examine the performance of
investments versus the performance of trading given the goals for each, and re-evaluate
those allocations.
Country Return Dynamics
To derive implications for systematic risk compensation and selectivity biases, we model
individual market returns as a two-regime process. We interpret regime 1 as the regime
when the market is open to international investors and investabledollar-denominated
returns are available for a given market. We discuss the details of this in the data section
below. Regime 2 is the regime where investable dollar returns
are not available and the market is inaccessible to international investors. We view the
transition from regime 1 to regime 2 as being associated with expropriationof
international investors. This expropriation can take various forms, includingcapital
controls, foreign exchange restrictions, and taxes on repatriations of foreign investments.
Information regarding the payoffs to international investors duringthe transition from
regime 1 to regime 2 and the ex-ante probability with which thistransition can happen are
not observable to an econometrician. In essence, we viewemerging markets returns akin
to payoffs of a defaultable bond which has not defaulted.As with the defaultable bond,
the likelihood of transition from regime 1
to regime 2 affects measured mean returns obtained solely from data sampled from
regime 1. Hence, one would expect that observed mean returns, particularly for an
emerging market, to be higher than the ex-ante risk premium. This bias measures the
compensation for expropriation and helps us understand the risk-return relationacross
markets. In equation (3), we present a time-series representation of returns that will
allowus to derive separate systematic risk compensation from sample selectivity biasesin
expected returns. Let yit+1 represent an indicator for the regime in market i att + 1 being
1 or 2. The indicator yit+1 is equal to one if the regime at t + 1 is 1 (open to international
investors), and zero otherwise. The return process, expressed indollars, is specified as
Rit+1 = E (Rit+1|It) + yit+1 (bi1et+1 + hi1t+1) + (1 - yit+1) (bi2et+1 + hi2t+1) , (3)
where E(Rit+1|It) is the ex-ante conditional mean of the gross return, et+1 is the
innovation in the systematic risk component, and hi1t+1 and hi2t+1 are diversifiable risk
components specific to market i. The exposure of the return to systematic risk is
determined by bi1 and bi2.Let rit+1 denote the excess return on market i, that is, rit+1 =
Rit+1 - Rf t.
Assuming that yit = 1, the valuation condition (1) then implies thatE (rit+1|It) = ls2et
[pitbi1 + (1 - pit) bi2] , (4)
where pit is the probability of the regime where market i is accessible to international
investors at time t. In other words, pit is the conditional probability that yit+1 = 1,and (1 -
pit) is the probability of a switch to regime 2. The risk premium is determinedby the
aggregate market price of risk, ls2et, and an overall beta which is aprobability-weighted
average of the betas in the two regimes. Next, we describe thedetermination of regimes 1
and 2.
The Sample Selectivity Process
Let y_it be a latent process that determines the opening and closing for market i. Thatis, it
determines if the regime is 1 or 2. In particular, if y_it >0 then the regime is 1and if y_it _
0, then the regime is classified as 2. Given this classification, it follows 6
thatyit = ( 1 if y_it > 0,0 otherwise.(5)
Following Heckman (1976, 1979), we assume that the conditional mean of the
latentprocess is determined by a vector of pre-determined variables xit. Hence, we
assume
That y_it+1 = d0ixit + #it+1, #it+1|xit _ N(0, 1) , (6)
where #it+1, by assumption, is a standard normal error. Brown, Goetzmann, andRoss
(1995) argue that the survival of a market (the analogue of our regime 1) is
determinedsolely by the price process itself. However, this seems restrictive, as many
emerging markets, such as Thailand, Indonesia, and Malaysia, have had comparable
drops in the market capitalization, but only Malaysia, directly expropriatedinternational
investors. This suggests that other economic considerations may be important in
determining whether investable dollar-denominated returns are availableto an
international investor. These other influences are captured by xit and #it+1. Further, the
latent variable model of selectivity provides connections betweendefault risk in sovereign
dollar-denominated bonds and the likelihood of capitalcontrols. This allows us to provide
a link between the cross-section of equity risk
premia and country risk ratings.Let f(.) denote the standard normal probability density
function, and let F (.)
denote the standard normal cumulative distribution function. It is straightforward to show
that the conditional probability that yit+1 = 1 is characterized by pit = E (yit+1|yit = 1,
xit) = Z ¥ -d0i xitf(#it+1)d#it+1 = Z d0i xit¥f(#it+1)d#it+1 = F _d0ixit_,(7)
where the third equality follows from the symmetry of the normal distribution. As#it+1
and the innovation in the return of asset i may be correlated, consider the
followingconditional projections for the different regimes
bi1et+1 + hi1t+1 = gi1#it+1 + vi1t+1, (8)
bi2et+1 + hi2t+1 = gi2#it+1 + vi2t+1, (9)
where gi1 and gi1 are the projection coefficients between bijet+1 +hit+1 and #it+1
andbijet+1 + hit+1 and #it+1, respectively, and vi1t+1 and vi2t+1 are projection errors.
The above equations then imply that the excess return process can be written as
rit+1 = E (rit+1|It) + yit+1(gi1#it+1 + vi1t+1) + (1 - yit+1)(gi2#it+1 + vi2t+1). (10)
2.4. The Sample Selectivity Criteria
We consider the case where data are missing as an outcome of an attrition process.That
is, we consider the sample selectivity effects of only observing the regimewhere the
markets are accessible to international investors. In this case, the restrictionon the
empirical conditional mean of the returns is
E (rit+1|It, yit = 1, yit+1 = 1) = E (rit+1|It) + gi1E _#it+1|#it+1 > -d0ixit_, (11)
where we have conditioned on the fact that the market is in regime 1 today (yit = 1)and
tomorrow (yit+1 = 1). This captures our view that an econometrician only
observesinvestable dollar return sample from regime 1. Note that E_#it+1|#it+1 > -
d0ixit_ is the same as E(#it+1|yit+1 = 1). Moreover, this quantity satisfies the relation
E (#it+1|yit+1 = 1) =1pit Z ¥-d0i xit#it+1f(#it+1)d#it+1, (12)
which can be further simplified as follows
1pit Z ¥-d0i xit #it+1 f(#it+1) d#it+1 = f(d0ixit)pit= f(d0ixit)F(d0ixit). (13)
This is typically referred to as a hazard rate, or the inverse Mill’s ratio. We denotethis by
hit, that is, hit = f(d0ixit)/F(d0ixit). Based on the above results, it follows thatthe
conditional mean of the excess return is given by
E (rit+1|It, yit = 1, yit+1 = 1) = ls2
et [pitbi1 + (1 - pit) bi2] + gi1hit. (14)
This restriction shows that there are two biases in measuring the ex-ante risk
premium.The first bias stems from the fact that the econometrician does not
observeregime 2 (the regime when investable dollar returns are not available). This is re-
flected in the first term of (14). bi1 can obviously be identified in the time series
fromobservations when the market is open. bi2 is the beta at transition from regime 1 to2.
Identification of bi2 and the transition probability of going to regime 2 (that is,81 - pit)
can not be measured without additional restrictions. Note that the resultingbias is on the
ex-ante mean of the return and we refer to it as a peso problem.
The second bias is due to sample selectivity, and the effects of this can be seenin the
second term of (14). This is an adjustment to the ex-post mean to correctly estimate the
ex-ante risk premium. Conditional on the availability of dollar returnstoday and
tomorrow, the risk premium is biased upwards. Put differently, investors require, on
average, a higher return when the market offers dollar returns, muchlike a defaultable
bond.
Brown, Goetzmann, and Ross (1995) focus on the second effect. It seems that the
measured risk premium will also be affected by the beta associated with the market shut-
down regime. If this beta is higher that in the regime for which datais available, then the
ex-ante mean asset will be higher, and in standard time-series regression this will show
up as an abnormal return, or an alpha. However, purgingthe empirical means of these two
effects implies that the ex-ante means lie on thesecurity market line.
In the special case of the world CAPM, equation (14) can be stated asE (rit+1|It, yit = 1,
yit+1 = 1) = E (rMt+1|It) [pitbi1 + (1 - pit) bi2] + gi1hit, (15)
where E (rMt+1|It) is the conditional risk premium on the world market portfolio,and the
betas are the world CAPM betas for the two regimes. As discussed above,taking account
of the peso problem requires measurement of bi2 and pit. In practise, estimating bi2 from
returns during a regime-switch is infeasible as there are very few, if any, in available
return data. Hence, in the empirical work, we will assumethat
bi1 = bi2 = biM. This gives us the following cross-sectional implications
E (rit+1|It, yit = 1, yit+1 = 1) = lMtbiM + gi1hit, (16)
where lMt = E (rMt+1|It). In the empirical work we also consider time-variation inbetas.
Allowing for this time-variation is straightforward and does not affect any of the
derivations above.
Finally, note that for high survival probabilities, the hazard rate in equation (13)is almost
linear in the probabilities. Under the assumption that the probabilitiesabout expropriation
in equity markets and sovereign debt markets are highly re-lated, it is straightforward to
show that pit can essentially be backed out from observed sovereign bond spreads (see
Appendix A). The premise that probabilitiesof bond default and expropriation in equity
markets are related is supported bythe events in Malaysia in 1998 and the more recent
events in Argentina. Note thatfor small default probabilities, the hazard rate is almost
linear in sovereign bondspreads. As discussed and documented later, at least for the few
sovereign spreads that we observe, the spreads are highly correlated with observed
measures of country ratings. Hence, we can use the more extensively available data on
country ratingsto measure the hazard rates themselves.
3. Data
We collect monthly return data on 46 developed and emerging markets from
Datastream.According to International Finance Corporation (IFC) of the World Bank, 21
of these markets are classified as developed and 25 as emerging markets. The
underlyingsources of the data are Morgan Stanley Capital International (MSCI)
fordeveloped markets and IFC for emerging markets. The returns from IFC are the
investable returns that incorporate foreign investment restrictions (including special
classes of shares, sector restrictions, single foreign shareholder limits,
restrictionsallowing only authorized investors, company statues, and national limits). We
also consider the return on the MSCI world market portfolio. All returns are in U.S.
dollars,
and excess returns are calculated by subtracting the one-month Eurodollar rate =pp for
each month.The sample period is January 1984 to November 2000. It is, however, well
known that many emerging markets only were accessible for international investors
beginning in the late 1980s and the early 1990s. This is reflected in our data base. Data
for emerging markets are included as and when they open up. We let the opening date of
an emerging market be the date when IFC begins to record investable returns. The
inclusion date for each market is shown in Table 1. The inclusion dates are similar to
what other studies have considered to be the financial market liberalization dates (see, for
instance, Kim and Singal, 2000, Bekaert and Harvey, 2000, and Henry,2000). Our
empirical results are not sensitive to using alternative choices of liberalizationdates. The
total number of observations for developed markets is 203 and for emerging markets the
number of observations varies between 90 and 144.In Table 1 we report summary
statistics of the monthly dollar returns. The averagereturns across developed and
emerging markets are about the same, 1.32%and 1.34% per month, respectively.
However, the average standard deviation ofemerging markets is about twice as high as
for developed markets. It also seemsto be greater dispersion in returns and return
volatilities of emerging economies.The correlation with the world market return is much
higher for developed markets
than for emerging markets.Table 2 presents information regarding various attributes of
the countries. Theseattributes are used in our cross-sectional analysis of risk premia. The
Real GDP perCapita attribute is the real GDP per capita in constant dollars in 1990
(expressed ininternational prices, base 1985). The Trading Activity attribute is the sum of
exportsand imports divided by GDP in 1990. The real GDP per capita and trading
activityattributes are collected from the World Penn Tables. The Economic Rating and
theFinancial Rating attributes refer to the average country ratings from inclusion dateto
November 2000, and is provided by the International Country Risk Guide (ICRG).
The economic risk rating is meant to measure an economy’s current strengths
andweaknesses, whereas the financial risk rating is meant to measure an
economy’sability to finance its official, commercial, and trade obligations. More
specifically,the variables determining the economic rating include a weighted average of
inflation,debt service as a percent of exports, international liquidity ratios, foreign
tradecollection experience, current account balance, and foreign exchange market
indicators.In the empirical work our measure of reputation is the financial rating, whichis
a weighted average of loan default, delayed payment of suppliers’ credit, repudiationof
contracts by government, losses from exchange controls, and expropriation of private
investment. The country ratings are published on a scale from 0 to 50 where a higher
number indicates lower risks. We have re-scaled the ratings to
between 0 (low) and 100 (high). A rating of 0 to 49 then indicates a very high risk;50 to
59 high risk; 60 to 69 moderate risk; 70 to 79 low risk; and 80 or more verylow risk. The
country rating are used by Erb, Harvey, and Viskanta (1996) in theirstudy of the time-
series predictability of future returns. La Porta, Lopez-de-Silanes,Shleifer, and Vishny
(1998) use these ratings to study investor protection and ownership structure across
countries. In this paperwe use the ratings to measure sampleselectivity.
Finally, we report betas versus the MSCI world market portfolio. The betas are,on
average, about the same for developed and emerging markets. However, thedispersion in
betas is much larger across emerging markets ranging from 0.07 to1.80, whereas they are
all about one in the developed markets.
It is evident from Table 2 that the emerging economies are economies with relativelylow
GDP per capita. Further, emerging economies have a much lower countryratings than
developed economies. In fact, the correlation between the real GDP percapita and the
ratings are 70% (economic rating) and 80% (financial rating). Thetrading activity
attribute has a lower correlation with the real GDP per capita (about20%). The
correlations between trading activity and the ratings are about 20% and40%. There are a
few outliers (notably Hong Kong and Singapore), but excludingthem does not affect the
correlation between trading activity and credit rating significantly.We also collect
sovereign spreads for nine emerging economies from J.P. Morgan.2 These are economies
with Brady bonds (restructured dollar-denominateddebt). We argue that the country
ratings contain much of the cross-sectional information in the spreads. For each month,
we computed the correlation between the sovereign spreads and the composite country
ratings. The correlations variedfrom -95% to -46% with an average of -72%. That is,
sovereign nations with a highspread on their dollar-denominated debt tend to have a low
country rating. This isalso highlighted in Figure 1, which shows the spreads versus
country ratings afterthe averages of the variables for each month have been subtracted.
That is, the variablesare measured as deviation from month averages to sweep out time
effects. Thecorrelation is about -58% and is highly significant (a p-value close to zero).
Similar results are obtained with either financial or economic country ratings.Our sample
begins in 1984 for developed markets, and in the late 1980s andearly 1990s for emerging
markets. Consequently, only brief data histories are available,particularly for emerging
economies. This makes it difficulty to solely rely ontime-series methods for measurement
and statistical inference. For this reason, weextensively use pooled cross-sectional
methods in the estimation. Importantly, therelative rankings of the attributes do not vary a
lot over time, indicating that mostof the information is in the cross-section. We typically
rely on the time series to
2The nine economies are Argentina, Brazil, Colombia, Korea, Mexico, Peru, Poland,
Turkey, andVenezuela.estimate exposures to risk sources, but evaluates the asset pricing
implications inthe cross-section. Increasing the sample for developed markets (going
back to 1976)does not change our results qualitatively and are therefore not reported.In
some specifications we allow the beta of a market versus the world market portfolio to
vary according a conditional information variable, namely the world excess dividend
yield (i.e., the dividend yield on the world market portfolio in excess of the one-month
Eurodollar deposit rate). These series are collected from Datastream.
4. Estimation and Methodology
In this section we present the estimation approach and discuss testable implications in the
time series as well as in the cross-section. We employ the generalized methodof moments
(GMM) of Hansen (1982) to estimate all parameters simultaneously asin Cochrane
(2001), and similar to Bansal and Dahlquist (2000) and Jagannathan andWang (2002). In
this framework, specific distributional assumptions of the asset returnsare not required,
and wedo not need to work in a normally independently and identically distributed
setting. We can handle both conditional heteroskedasticityand serial correlation in pricing
errors. The approach is different from traditional approaches as we avoid the problem of
generated regressors, and it is not necessary
to develop further methods and corrections as in two-step procedures.We have to deal
with missing data as the dollar return series for emerging markets
handle the missing data as in Bansal and Dahlquist (2000). The idea is to balance the data
set, and then apply the asymptotic results in the standard GMM framework. This is
further discussed below. We are interested in estimating the risk exposures and risk
premia simultaneously.Consider N markets (i = 1, 2, . . . , N), each with T observations (t
= 1, 2, . . . , T).Recall that the emerging markets have different lengths of histories. We
describe the
estimation approach for the world CAPM with time-varying betas.As in Jagannathan and
Wang (1996), Cochrane (1996), amongst others, we evaluate
the implications of the model in the cross-section as their is considerable cross- sectional
variation in the mean returns. Consider the cross-sectional risk premium implications in
equation (16). In addition, allow for the market beta of an asset to be time-varying
according to biM + biMzzt, where zt is a variable known at time t capturing time
variation in the market beta. The implications for the cross-section
of unconditional mean excess returns can then be written as
E (rit+1) = lMbiM + lMzbiMz + gihi, (17)
where lM = E (lMt) and lMz = E (lMtzt). The biMs and biMzs are the standard
time series projection coefficients. Hence, our first sets of moment conditions, for
each market i, are
E [(rit+1 - aiM - biMrMt+1 - biMzrMt+1zt) yityit+1] = 0, (18)
E [(rit+1 - aiM - biMrMt+1 - biMzrMt+1zt) rMt+1yityit+1] = 0, (19)
E [(rit+1 - aiM - biMrMt+1 - biMzrMt+1zt) rMt+1ztyityit+1] = 0. (20)
These moment conditions are exactly identified. We have 3N moment conditionsand the
same number of parameters. The point estimates from these moment conditions
correspond to the usual least squares estimates. We follow the literature and add
constants, or alphas. In the world CAPM, the aiMs should be equal to zero.Indeed, we
will evaluate the CAPM by checking whether the alphas are all equal tozero in the time
series. Our focus, however, is on the ability of the various modelsh and without sample
selectivity) to explain the cross-section of risk premia.
Note that we use the regime indicator variable to make our unbalanced panel abalanced
panel as in Bansal and Dahlquist (2000). That is, the moment conditions are multiplied
with the product of the regime indicators at time t and t + 1, yityit+1. The product
yityit+1 selects returns when markets are open both at time t and t + 1. In essence, this
procedure treats missing observations as zeros. This has a practical advantage since the
usual moment conditions which contain missing data can be filled with zeros, and then
standard GMM routines can be utilized.3
The sample selectivity part in equation (17) is gihi. As noted in the discussion of equation
(14), under simplifying assumptions, the probability of default can be recovered from the
sovereign bond spread. Further, this spread can be used to 3Hayashi (2000) considers,
also in an analysis of panel data, a similar approach. Stambaugh (1997) presents an
alternative approach to address this econometric issue.
14
completely characterize the hazard rate at time t. However, the data on sovereign interest
rate spreads are not available for many economies in our sample period. As shown
earlier, there is a high negative correlation between the country ratings and the spreads in
the cross-section (for economies where sovereign spread data are available). That is, a
country with a low rating tends to have a high spread (a high probability of default).
Consequently, to characterize the cross-section of hazardrates, we model the hazard rate
for market i as follows
gihi = (g00 + g01Ci) Ai, (21)
where Ai proxies for hi in the cross-section. For example, we let Ai equal the countryis
economic rating which then captures the cross-sectional variation in the hazardrate.
Further, to allow for controlled cross-sectional heterogeneity in gi, we model it as
gi = g00 + g01Ci, where Ci denotes a country-specific attribute such as its return
volatility, financial rating, or its trading activity.
The cross-sectional parameters (i.e., the risk premium parameters and the g00and g01
parameters) are then identified in the last set of moment conditions for each asset i
E [(rit+1 - lMbiM - lMzbiMz - g00Ai - g01AiCi) yityit+1] = 0. (22)
We also consider a specification with a constant term
E [(rit+1 - l0 - lMbiM - lMzbiMz - g00Ai - g01AiCi) yityit+1] = 0. (23)
The constant term l0 should be zero according to theory, and a non-zero constant
indicates that a model cannot price the assets on average. Alternatively, a non-zero
constant can be interpreted as a zero-beta rate different from the riskfree rate thatis
imposed. Note that all parameters including the betas and the cross-sectional parameters
l0, lM, lMz, g00, and g01 are jointly estimated using GMM. Details of the estimation are
given in Appendix B.
Results
This section presents the empirical results. Recall, that we earlier reported that the cross-
sectional dispersion in the average returns is fairly large for emerging markets and small
for developed markets. This cross-sectional dispersion poses a serious challenge to asset
pricing models. Variables that characterize the selectivity bias, such as country ratings,
have very little time-series variation, but considerablecross-sectional variation. Hence,
the effects of selectivity are primarily identifiable in the cross-section. Given the large
cross-sectional dispersion in the data along with the short data histories for many
emerging markets, we, as in Black, Jensen, and Scholes (1972), Fama and MacBeth
(1973), and Jagannathan and Wang (1996), focusprimarily on the explaining the cross-
sectional differences in risk-premia.
We first discuss the ability of the various models to capture the cross-section of average
returns through only systematic risk. We then include sample selectivity in the cross-
section. Finally, we discuss the results and provide further interpretations
of the results.
Evidence in the Absence of Selectivity
In Table 3, we provide evidence from the cross-section of asset returns. The estimated
risk premium for the market portfolio is negative, as can be seen in row 2 of Panel A.
This is a standard finding (see, for instance, Jagannathan and Wang, 1996). The ability of
the CAPM with constant betas to explain the cross-section of average returns is basically
zero as indicated by the adjusted R-square. In Panel B, we consider the CAPM where the
market betas are allowed to be time-varying. The model fails to capture the cross-
sectional dispersion in average returns in this specification as well The adjusted R-square
is only about 8%. The theoretical restriction that l0 = 0 can be rejected at the 5%
significance level. However, the constant term, as discussed below, is not particularly
relevant when sample selectivity is included in
the model. For completeness, we have conducted the time-series tests for both the
constant beta and timevaryingbeta versions of the CAPM (not reported in a table). We
find that the joint test of zero alphasis rejected in both cases. The rejections seem to be
primarily due to abnormal returns in emergingmarkets—this is consistent with Harvey
(1995) who also shows that CAPM implications are rejectedin emerging markets data.
The failure of the CAPM can also be seen in Figure 2 where we plot the averagereturns
against the predicted expected returns from the model. A true model would,ignoring
estimation errors, produce observations along the 45-degree line. The figure reveals that
there is almost no dispersion in predicted expected returns. Hence, the model does not
capture the large cross-sectional variation in average returns.
Evidence with Selectivity Included
The model specifications with sample selectivity are also reported in Table 3. Inthese
specifications the selectivity is modelled as (g00 + g01si)Ai, where si is the annual return
volatility for market i, and Ai is defined as the economic rating for country i less the
economic rating for the U.S. The expression (g00 + g01si) is the gi for country i. The
proxy for the hazard rate for country i is Ai, based on thereasoning provided earlier. Note
that Ai is negative for emerging economies and close to zero for developed economies.
This specification captures the intuition that as economies improve their economic rating
they become akin to developed marketsand the sample selectivity term would fall.When
sample selectivity is incorporated in the standard CAPM (in Panel A), thecross-sectional
R-square rises to 41% and the parameters associated with the selectivityterm are
significant (a p-value of 3%). The time-varying beta based CAPMwith sample selectivity
included is reported in Panel B. This specification does quitewell in capturing the cross-
sectionalvariation in risk premia, and has an adjustedR-square of 61%. The magnitudes of
the parameters that govern the selectivity bias(that is, g00 andg01) are similar across
different specifications. They are in all casesjointly significant at usual significance levels
(see the column labelled “Test ofJointSignificance”). The last two rows of Panel B
highlights the relevance, or the lackthereof, of the constant term l0. The empirical results
across thetwo cases (includingl0, or not) are comparable. Hence, as suggested by theory,
the constant term isnot particularly important.Table 4 provides themagnitudes of the
overall risk premia explained by systematicrisk and by sample selectivity. Economies
with poor economic rating have alarger andpositive selectivity bias. For developed
economies the variable Ai is essentiallyzero and hence the effect of selectivity on their
mean returns is absent.
the constant beta CAPM, the systematic risk contribution is about 0.45% per month for
both emerging and developed economies. However, the selectivity premium is0.50% per
month for emerging markets and close to zero for developed markets. Inthe model with
time-varying betas (see PanelB),the fraction of the emerging marketreturn attributed to
the selectivity bias is somewhat higher, and now stands at 0.58%per month. For
emergingmarkets more than 1/2 of the ex-post risk premium can beattributed to
selectivity. That is, sample selectivity seems to be the dominant influenceon the measured
risk premiums in emerging economies. Sample selectivity isnot an important dimension
for understanding measured risk premiain developedmarkets.The higher explanatory
ability of the world CAPM with time-varying betas andsample selectivity can be seen in
Figure 3which displays the average returns againstpredicted expected returns. The
improvement in fit is visible and the model is ableto produce the highdispersion in
average returns.We also considered alternative specifications for the parameter gi. In
particular,we replaced si with a reputationalvariable—the financial rating of an economy
i lessthe comparable rating for the U.S. The ability of this specification in terms of
capturingthecross-sectional variation in risk premia (i.e., adjusted R-square) is about30%.
This R-square is quite high relative the specifications withouttheselectivityeffects. As
shown in Table 4 the average emerging markets risk premium is still predominantlydue
to selectivity bias. Yet anotherchoice for the specification of gi, thetrading activity
variable, produces again similar results.Finally, we consider a specification where we use
thespreads (short samplesavailable for nine economies) on the Brady bonds to measure
the hazard rates directly.The approach is as follows. We projecttheaverage spreads on the
averagecountry ratings and use these projection coefficients to infer spreads and
defaulprobabilities (under theassumption of zero recovery as in Appendix A) for
allemerging markets. Recall that we are making the assumption that the probabilityof
default indebt markets coincides with the probability of expropriation. Fromthe
probabilities we can then compute the hazard rates (i.e., the his). For developedmarkets
we assume a zero default probability (and hence a zero hazard rate). Withthis measure of
the hazard rate, and the same specificationforthe gi as before, we estimate the cross-
sectional regression in equation (22). This specification capturesabout 36% of the cross-
sectional variation in risk premia. We find thatthe selectivity18term, that is gihi, is about
the 0.63% per month. The average probability of the defaultis about half a percent per
month. Hence, the bias in the mean return of about 0.63% per month can be supported by
rather small probability of default (risk of expropriation).Note that the empirical evidence
for this specification is quite similarto that discussed in the time-varying beta case in
Panel B of Table 3.
5.3. What Drives the Selectivity Bias?
In Panel A of Table 5 we inquire what economic variables can explain the
crosssectionaldispersion in the selectivity bias for emerging markets. In particular, weare
interested in whether the measured selectivity premium is related to trading activityand/or
measures of reputation. To do so, we considerthe measures of theselectivity premium
based on the specification where gi = (g00 + g01si), and therelative economic rating is
the proxy for thehazard rate. This specification was reportedin Table 3. The reputational
variable (the financial rating of country i lessthe comparable rating for theU.S.)is able to
explain about 32% of the dispersion inthe selectivity premium. This regression also
shows that the selectivity premiumrises as thecountry’s financial rating falls. Similarly,
when we use the tradingactivity variable, this explains about 19% of the dispersion in the
selectivitypremium.Economies with larger trading activity have a smaller selectivity
premium.In essence our evidence suggests that both trading activityand reputational
considerationsare important for explaining the selectivity premium.Allowing the
selectivity premium to depend on the volatilityinthe cross-sectionis motivated by
arguments presented in Brown, Goetzmann, and Ross (1995). Ourevidence indicates that
this attribute is notuniquely important to capture the crosssectionaldifferences in risk
premia. Indeed, the trade activity and financial reputationvariables do, at least
ineconomic terms, a comparable job of explaining the cross-sectional differences in the
risk premia. Thus, it seems to us that this is dueto the fact that the volatility of returns are
related to these variables.Thisis shownin Panel B of Table 5: return volatility is
decreasing in both trading activity andfinancial reputation. These variables, based on the
work of Eaton and Gersovitz(1981), and Bulow and Rogoff (1989a, 1989b) should matter
to the compensationthat emerging markets have toadditionallypay, due to risks of
expropriation. Wefind that this indeed is the case.
Conclusion
In this paper we show that the cross-sectional differences in the equity returns
acrosssovereign economies is determined by two features—systematic risk and a
selectivitypremium. We show that the selectivity premium captures more than 1/2 of
theaverage risk premium in emerging markets. The equity riskpremia in
developedmarkets seems to be driven solely by systematic risk. The main economic
implicationof this result is that after taking account ofselectivity premium all
internationalequity returns reflect systematic risk, as predicted by theory.Our empirical
work also shows that sovereigns thathave better financial marketreputations and trade
more actively have to pay a smaller selectivity premium. Thisempirical evidence lends
support to theview that both reputations and fear of tradesanctions are important in
determining the cost of equity borrowing for a sovereignnation.
Measuring Hazard Rates From Sovereign Spreads
This Appendix shows how the hazard rate can be measured from sovereign bondspreads.
Consider a dollar denominated pure discount bond issued by a country.The payoff is
equal to one if there is no default, and µb + bbet+1 +hbt+1 if the countrydefaults. The
payoff process can thus be written as qbt+1= ybt+1 + (1 - ybt+1) (µb + bbet+1 + hbt+1) .
(24)
For simplicity, we assume that bb = 0. That is, we assume that the recovery valueof the
bond is not related to the systematic risk in the world economy.Further, theexpected
payoff on this bond in default is less than one (i.e., µb < 1). Valuing thispayoff using the
stochastic discount factor implies that
1/Rbt = [pbt + (1 - pbt) µb] /Rf t. (25)
Solving for the probability of no default, we obtain
pbt =Rf t - µbRbtRbt (1 - µb). (26)
Assume that the probabilities of default for the bond correspond to the probabilityof a
market shut-down, that is, pbt = pit. Under the further assumptionthat therecovery rate is
zero, we can directly recover the probability of default. Further,given the normal
cumulative distribution function wecan completely characterizethe hazard rate.The above
expression can also be used to compute the ex-ante beta on market i.We denote this with
bit =pitbi1 + (1 - pit) bi2. If we assume that the ratio of thebetas across the two regimes is
equal to a constant c and µb = 0, it follows that
bit = bi1 _Rf tRbt + c _Rbt - Rf tRbt __. (27)
Note that Rf t and Rbt can be observed directly from U.S. Treasuries and
Sovereignbonds, or as we demonstrate, approximated with a country’s relativecountry
rating.Hence, conditional on c, one can estimate the model with both a peso problem
andsample selectivity. In the special case with c = 1,there is no peso problem and wehave
that bit = bi1 = bi2.21B. Estimation DetailsThis Appendix shows the estimation in more
detail. Let q0 denote thetrue parametervector that we want to estimate. The typical
elements in q0 are aiM, biM, and biMz that are specific to each market, and the common
parameters l0, lM, lMz, g00and g01. By stacking the sample counterparts of the moment
conditions in (18) to (20), and (23), we have a vector of moment conditions
gT (q) =1T Tåt=1f (Xt, q) , (28)
where Xt summarizes the data used to form the moments conditions. The vectorgT (q)
has the dimension 4N. The moment conditions, given by (18) to (20), exactlyidentify the
aiM, biM, and biMz parameters. However, the moment conditions, givenby (23), is
overidentified. We have N moment conditions, but only 5 parameters (l0,lM, lMz, g00
and g01).We estimate the parameters by setting linear combinations of gT equal to zero.
That is, the moment conditions can be written as
ATgT = 0, (29)
where AT is a (3N + 5) × 4N matrix. In particular, our choice of AT is designed toensure
that the point estimates are the ones given by ordinary least squares.
Let AT
be the product of two matrices denoted by A1T and A2T (that is, AT = A1TA2T).
Thefollowing matrices result in least square point estimates
A1T =26666666664I3N 03N · · · 03N003N 1 · · · 1003Nˆ b1M · · · ˆ bNM003Nˆ b1Mz ·
· · ˆ bNMz003N A1 · · · AN003N C1A1 · · · CNAN, (30)
where I3N is the identity matrix with dimension 3N, 03N is a 3N vector of zeros,0N is an
N vector of zeros, and A2T is a diagonal matrix with typical element equalto 1/ åTt=1
yit+1. The ˆ biMs and ˆ biMzs are estimates of biMs and biMzs, and they are
given in the estimation. The ˆ biMs and ˆ biMzs are exactly the least square
estimatesobtained in a regression of the assets’ excess returns on the market excess
returnand scaled market excess returns as in (18) to (20). Further, the estimates of l0,
lM,lMz, g00, and g01 coincide with the least square estimates obtained in a regressionof
average returns on the betas and the proxies for sample selectivity. Our choice ofAT
ensures that ATgT (qT) = 0.
Based on Hansen (1982) we know that when linear combinations of gT are setequal to
zero as in (29), the asymptotic distribution of the point estimator qT is given
by
pT (qT - q0) d
! N _0, (A0D0)-1 _A0S0A00_(A0D0)-10_, (31)
where D0 is the gradient of the moment conditions in (28), and where S0 is the
variance-covariance matrix of the moment conditions and given by
S0 =¥åj=-¥E hf (Xt, q0) f _Xt-j, q0_0i. (32) The sample counterpart ST is estimated
using the procedure in Newey and West(1987) with four lags. D0 and A0 can be
estimated by their sample counterparts DTand AT. Note that the standard errors based on
(31) are robust to heteroskedasticity
and serial correlation in the moment conditions.
Investment Risk
Investment risk includes credit risk and market risk which are both taken to combat
another risk—that due to inflation. Investors and traders need to assess the actual historic
results of the strategies they employ for their investments. In this manner, the individual
can better understand the true financial risks in which they are exposing themselves. Two
such risks may include under-investing in bonds & equities or over-investing in assets).
In order to manage necessary market risk, the investor must take steps to understand
those risks and minimize them through diversification. Numerous articles on the topic of
asset allocation and diversification are available in the Optionetics.com article archives
by completing an author keyword search for 2004 and 2005. Traders, who often self-
direct their investment accounts, must proactively examine the performance of
investments versus the performance of trading given the goals for each, and re-evaluate
those allocations.
Country Return Dynamics
To derive implications for systematic risk compensation and selectivity biases, we model
individual market returns as a two-regime process. We interpret regime 1 as the regime
when the market is open to international investors and investabledollar-denominated
returns are available for a given market. We discuss the details of this in the data section
below. Regime 2 is the regime where investable dollar returns
are not available and the market is inaccessible to international investors. We view the
transition from regime 1 to regime 2 as being associated with expropriationof
international investors. This expropriation can take various forms, includingcapital
controls, foreign exchange restrictions, and taxes on repatriations of foreign investments.
Information regarding the payoffs to international investors duringthe transition from
regime 1 to regime 2 and the ex-ante probability with which thistransition can happen are
not observable to an econometrician. In essence, we viewemerging markets returns akin
to payoffs of a defaultable bond which has not defaulted.As with the defaultable bond,
the likelihood of transition from regime 1
to regime 2 affects measured mean returns obtained solely from data sampled from
regime 1. Hence, one would expect that observed mean returns, particularly for an
emerging market, to be higher than the ex-ante risk premium. This bias measures the
compensation for expropriation and helps us understand the risk-return relationacross
markets. In equation (3), we present a time-series representation of returns that will
allowus to derive separate systematic risk compensation from sample selectivity biasesin
expected returns. Let yit+1 represent an indicator for the regime in market i att + 1 being
1 or 2. The indicator yit+1 is equal to one if the regime at t + 1 is 1 (open to international
investors), and zero otherwise. The return process, expressed indollars, is specified as
Rit+1 = E (Rit+1|It) + yit+1 (bi1et+1 + hi1t+1) + (1 - yit+1) (bi2et+1 + hi2t+1) , (3)
where E(Rit+1|It) is the ex-ante conditional mean of the gross return, et+1 is the
innovation in the systematic risk component, and hi1t+1 and hi2t+1 are diversifiable risk
components specific to market i. The exposure of the return to systematic risk is
determined by bi1 and bi2.Let rit+1 denote the excess return on market i, that is, rit+1 =
Rit+1 - Rf t.
Assuming that yit = 1, the valuation condition (1) then implies thatE (rit+1|It) = ls2et
[pitbi1 + (1 - pit) bi2] , (4)
where pit is the probability of the regime where market i is accessible to international
investors at time t. In other words, pit is the conditional probability that yit+1 = 1,and (1 -
pit) is the probability of a switch to regime 2. The risk premium is determinedby the
aggregate market price of risk, ls2et, and an overall beta which is aprobability-weighted
average of the betas in the two regimes. Next, we describe thedetermination of regimes 1
and 2.
The Sample Selectivity Process
Let y_it be a latent process that determines the opening and closing for market i. Thatis, it
determines if the regime is 1 or 2. In particular, if y_it >0 then the regime is 1and if y_it _
0, then the regime is classified as 2. Given this classification, it follows 6
thatyit = ( 1 if y_it > 0,0 otherwise.(5)
Following Heckman (1976, 1979), we assume that the conditional mean of the
latentprocess is determined by a vector of pre-determined variables xit. Hence, we
assume
That y_it+1 = d0ixit + #it+1, #it+1|xit _ N(0, 1) , (6)
where #it+1, by assumption, is a standard normal error. Brown, Goetzmann, andRoss
(1995) argue that the survival of a market (the analogue of our regime 1) is
determinedsolely by the price process itself. However, this seems restrictive, as many
emerging markets, such as Thailand, Indonesia, and Malaysia, have had comparable
drops in the market capitalization, but only Malaysia, directly expropriatedinternational
investors. This suggests that other economic considerations may be important in
determining whether investable dollar-denominated returns are availableto an
international investor. These other influences are captured by xit and #it+1. Further, the
latent variable model of selectivity provides connections betweendefault risk in sovereign
dollar-denominated bonds and the likelihood of capitalcontrols. This allows us to provide
a link between the cross-section of equity risk
premia and country risk ratings.Let f(.) denote the standard normal probability density
function, and let F (.)
denote the standard normal cumulative distribution function. It is straightforward to show
that the conditional probability that yit+1 = 1 is characterized by pit = E (yit+1|yit = 1,
xit) = Z ¥ -d0i xitf(#it+1)d#it+1 = Z d0i xit¥f(#it+1)d#it+1 = F _d0ixit_,(7)
where the third equality follows from the symmetry of the normal distribution. As#it+1
and the innovation in the return of asset i may be correlated, consider the
followingconditional projections for the different regimes
bi1et+1 + hi1t+1 = gi1#it+1 + vi1t+1, (8)
bi2et+1 + hi2t+1 = gi2#it+1 + vi2t+1, (9)
where gi1 and gi1 are the projection coefficients between bijet+1 +hit+1 and #it+1
andbijet+1 + hit+1 and #it+1, respectively, and vi1t+1 and vi2t+1 are projection errors.
The above equations then imply that the excess return process can be written as
rit+1 = E (rit+1|It) + yit+1(gi1#it+1 + vi1t+1) + (1 - yit+1)(gi2#it+1 + vi2t+1). (10)
2.4. The Sample Selectivity Criteria
We consider the case where data are missing as an outcome of an attrition process.That
is, we consider the sample selectivity effects of only observing the regimewhere the
markets are accessible to international investors. In this case, the restrictionon the
empirical conditional mean of the returns is
E (rit+1|It, yit = 1, yit+1 = 1) = E (rit+1|It) + gi1E _#it+1|#it+1 > -d0ixit_, (11)
where we have conditioned on the fact that the market is in regime 1 today (yit = 1)and
tomorrow (yit+1 = 1). This captures our view that an econometrician only
observesinvestable dollar return sample from regime 1. Note that E_#it+1|#it+1 > -
d0ixit_ is the same as E(#it+1|yit+1 = 1). Moreover, this quantity satisfies the relation
E (#it+1|yit+1 = 1) =1pit Z ¥-d0i xit#it+1f(#it+1)d#it+1, (12)
which can be further simplified as follows
1pit Z ¥-d0i xit #it+1 f(#it+1) d#it+1 = f(d0ixit)pit= f(d0ixit)F(d0ixit). (13)
This is typically referred to as a hazard rate, or the inverse Mill’s ratio. We denotethis by
hit, that is, hit = f(d0ixit)/F(d0ixit). Based on the above results, it follows thatthe
conditional mean of the excess return is given by
E (rit+1|It, yit = 1, yit+1 = 1) = ls2
et [pitbi1 + (1 - pit) bi2] + gi1hit. (14)
This restriction shows that there are two biases in measuring the ex-ante risk
premium.The first bias stems from the fact that the econometrician does not
observeregime 2 (the regime when investable dollar returns are not available). This is re-
flected in the first term of (14). bi1 can obviously be identified in the time series
fromobservations when the market is open. bi2 is the beta at transition from regime 1 to2.
Identification of bi2 and the transition probability of going to regime 2 (that is,81 - pit)
can not be measured without additional restrictions. Note that the resultingbias is on the
ex-ante mean of the return and we refer to it as a peso problem.
The second bias is due to sample selectivity, and the effects of this can be seenin the
second term of (14). This is an adjustment to the ex-post mean to correctly estimate the
ex-ante risk premium. Conditional on the availability of dollar returnstoday and
tomorrow, the risk premium is biased upwards. Put differently, investors require, on
average, a higher return when the market offers dollar returns, muchlike a defaultable
bond.
Brown, Goetzmann, and Ross (1995) focus on the second effect. It seems that the
measured risk premium will also be affected by the beta associated with the market shut-
down regime. If this beta is higher that in the regime for which datais available, then the
ex-ante mean asset will be higher, and in standard time-series regression this will show
up as an abnormal return, or an alpha. However, purgingthe empirical means of these two
effects implies that the ex-ante means lie on thesecurity market line.
In the special case of the world CAPM, equation (14) can be stated asE (rit+1|It, yit = 1,
yit+1 = 1) = E (rMt+1|It) [pitbi1 + (1 - pit) bi2] + gi1hit, (15)
where E (rMt+1|It) is the conditional risk premium on the world market portfolio,and the
betas are the world CAPM betas for the two regimes. As discussed above,taking account
of the peso problem requires measurement of bi2 and pit. In practise, estimating bi2 from
returns during a regime-switch is infeasible as there are very few, if any, in available
return data. Hence, in the empirical work, we will assumethat
bi1 = bi2 = biM. This gives us the following cross-sectional implications
E (rit+1|It, yit = 1, yit+1 = 1) = lMtbiM + gi1hit, (16)
where lMt = E (rMt+1|It). In the empirical work we also consider time-variation inbetas.
Allowing for this time-variation is straightforward and does not affect any of the
derivations above.
Finally, note that for high survival probabilities, the hazard rate in equation (13)is almost
linear in the probabilities. Under the assumption that the probabilitiesabout expropriation
in equity markets and sovereign debt markets are highly re-lated, it is straightforward to
show that pit can essentially be backed out from observed sovereign bond spreads (see
Appendix A). The premise that probabilitiesof bond default and expropriation in equity
markets are related is supported bythe events in Malaysia in 1998 and the more recent
events in Argentina. Note thatfor small default probabilities, the hazard rate is almost
linear in sovereign bondspreads. As discussed and documented later, at least for the few
sovereign spreads that we observe, the spreads are highly correlated with observed
measures of country ratings. Hence, we can use the more extensively available data on
country ratingsto measure the hazard rates themselves.
3. Data
We collect monthly return data on 46 developed and emerging markets from
Datastream.According to International Finance Corporation (IFC) of the World Bank, 21
of these markets are classified as developed and 25 as emerging markets. The
underlyingsources of the data are Morgan Stanley Capital International (MSCI)
fordeveloped markets and IFC for emerging markets. The returns from IFC are the
investable returns that incorporate foreign investment restrictions (including special
classes of shares, sector restrictions, single foreign shareholder limits,
restrictionsallowing only authorized investors, company statues, and national limits). We
also consider the return on the MSCI world market portfolio. All returns are in U.S.
dollars,
and excess returns are calculated by subtracting the one-month Eurodollar rate =pp for
each month.The sample period is January 1984 to November 2000. It is, however, well
known that many emerging markets only were accessible for international investors
beginning in the late 1980s and the early 1990s. This is reflected in our data base. Data
for emerging markets are included as and when they open up. We let the opening date of
an emerging market be the date when IFC begins to record investable returns. The
inclusion date for each market is shown in Table 1. The inclusion dates are similar to
what other studies have considered to be the financial market liberalization dates (see, for
instance, Kim and Singal, 2000, Bekaert and Harvey, 2000, and Henry,2000). Our
empirical results are not sensitive to using alternative choices of liberalizationdates. The
total number of observations for developed markets is 203 and for emerging markets the
number of observations varies between 90 and 144.In Table 1 we report summary
statistics of the monthly dollar returns. The averagereturns across developed and
emerging markets are about the same, 1.32%and 1.34% per month, respectively.
However, the average standard deviation ofemerging markets is about twice as high as
for developed markets. It also seemsto be greater dispersion in returns and return
volatilities of emerging economies.The correlation with the world market return is much
higher for developed markets
than for emerging markets.Table 2 presents information regarding various attributes of
the countries. Theseattributes are used in our cross-sectional analysis of risk premia. The
Real GDP perCapita attribute is the real GDP per capita in constant dollars in 1990
(expressed ininternational prices, base 1985). The Trading Activity attribute is the sum of
exportsand imports divided by GDP in 1990. The real GDP per capita and trading
activityattributes are collected from the World Penn Tables. The Economic Rating and
theFinancial Rating attributes refer to the average country ratings from inclusion dateto
November 2000, and is provided by the International Country Risk Guide (ICRG).
The economic risk rating is meant to measure an economy’s current strengths
andweaknesses, whereas the financial risk rating is meant to measure an
economy’sability to finance its official, commercial, and trade obligations. More
specifically,the variables determining the economic rating include a weighted average of
inflation,debt service as a percent of exports, international liquidity ratios, foreign
tradecollection experience, current account balance, and foreign exchange market
indicators.In the empirical work our measure of reputation is the financial rating, whichis
a weighted average of loan default, delayed payment of suppliers’ credit, repudiationof
contracts by government, losses from exchange controls, and expropriation of private
investment. The country ratings are published on a scale from 0 to 50 where a higher
number indicates lower risks. We have re-scaled the ratings to
between 0 (low) and 100 (high). A rating of 0 to 49 then indicates a very high risk;50 to
59 high risk; 60 to 69 moderate risk; 70 to 79 low risk; and 80 or more verylow risk. The
country rating are used by Erb, Harvey, and Viskanta (1996) in theirstudy of the time-
series predictability of future returns. La Porta, Lopez-de-Silanes,Shleifer, and Vishny
(1998) use these ratings to study investor protection and ownership structure across
countries. In this paperwe use the ratings to measure sampleselectivity.
Finally, we report betas versus the MSCI world market portfolio. The betas are,on
average, about the same for developed and emerging markets. However, thedispersion in
betas is much larger across emerging markets ranging from 0.07 to1.80, whereas they are
all about one in the developed markets.
It is evident from Table 2 that the emerging economies are economies with relativelylow
GDP per capita. Further, emerging economies have a much lower countryratings than
developed economies. In fact, the correlation between the real GDP percapita and the
ratings are 70% (economic rating) and 80% (financial rating). Thetrading activity
attribute has a lower correlation with the real GDP per capita (about20%). The
correlations between trading activity and the ratings are about 20% and40%. There are a
few outliers (notably Hong Kong and Singapore), but excludingthem does not affect the
correlation between trading activity and credit rating significantly.We also collect
sovereign spreads for nine emerging economies from J.P. Morgan.2 These are economies
with Brady bonds (restructured dollar-denominateddebt). We argue that the country
ratings contain much of the cross-sectional information in the spreads. For each month,
we computed the correlation between the sovereign spreads and the composite country
ratings. The correlations variedfrom -95% to -46% with an average of -72%. That is,
sovereign nations with a highspread on their dollar-denominated debt tend to have a low
country rating. This isalso highlighted in Figure 1, which shows the spreads versus
country ratings afterthe averages of the variables for each month have been subtracted.
That is, the variablesare measured as deviation from month averages to sweep out time
effects. Thecorrelation is about -58% and is highly significant (a p-value close to zero).
Similar results are obtained with either financial or economic country ratings.Our sample
begins in 1984 for developed markets, and in the late 1980s andearly 1990s for emerging
markets. Consequently, only brief data histories are available,particularly for emerging
economies. This makes it difficulty to solely rely ontime-series methods for measurement
and statistical inference. For this reason, weextensively use pooled cross-sectional
methods in the estimation. Importantly, therelative rankings of the attributes do not vary a
lot over time, indicating that mostof the information is in the cross-section. We typically
rely on the time series to
2The nine economies are Argentina, Brazil, Colombia, Korea, Mexico, Peru, Poland,
Turkey, andVenezuela.estimate exposures to risk sources, but evaluates the asset pricing
implications inthe cross-section. Increasing the sample for developed markets (going
back to 1976)does not change our results qualitatively and are therefore not reported.In
some specifications we allow the beta of a market versus the world market portfolio to
vary according a conditional information variable, namely the world excess dividend
yield (i.e., the dividend yield on the world market portfolio in excess of the one-month
Eurodollar deposit rate). These series are collected from Datastream.
4. Estimation and Methodology
In this section we present the estimation approach and discuss testable implications in the
time series as well as in the cross-section. We employ the generalized methodof moments
(GMM) of Hansen (1982) to estimate all parameters simultaneously asin Cochrane
(2001), and similar to Bansal and Dahlquist (2000) and Jagannathan andWang (2002). In
this framework, specific distributional assumptions of the asset returnsare not required,
and wedo not need to work in a normally independently and identically distributed
setting. We can handle both conditional heteroskedasticityand serial correlation in pricing
errors. The approach is different from traditional approaches as we avoid the problem of
generated regressors, and it is not necessary
to develop further methods and corrections as in two-step procedures.We have to deal
with missing data as the dollar return series for emerging markets
handle the missing data as in Bansal and Dahlquist (2000). The idea is to balance the data
set, and then apply the asymptotic results in the standard GMM framework. This is
further discussed below. We are interested in estimating the risk exposures and risk
premia simultaneously.Consider N markets (i = 1, 2, . . . , N), each with T observations (t
= 1, 2, . . . , T).Recall that the emerging markets have different lengths of histories. We
describe the
estimation approach for the world CAPM with time-varying betas.As in Jagannathan and
Wang (1996), Cochrane (1996), amongst others, we evaluate
the implications of the model in the cross-section as their is considerable cross- sectional
variation in the mean returns. Consider the cross-sectional risk premium implications in
equation (16). In addition, allow for the market beta of an asset to be time-varying
according to biM + biMzzt, where zt is a variable known at time t capturing time
variation in the market beta. The implications for the cross-section
of unconditional mean excess returns can then be written as
E (rit+1) = lMbiM + lMzbiMz + gihi, (17)
where lM = E (lMt) and lMz = E (lMtzt). The biMs and biMzs are the standard
time series projection coefficients. Hence, our first sets of moment conditions, for
each market i, are
E [(rit+1 - aiM - biMrMt+1 - biMzrMt+1zt) yityit+1] = 0, (18)
E [(rit+1 - aiM - biMrMt+1 - biMzrMt+1zt) rMt+1yityit+1] = 0, (19)
E [(rit+1 - aiM - biMrMt+1 - biMzrMt+1zt) rMt+1ztyityit+1] = 0. (20)
These moment conditions are exactly identified. We have 3N moment conditionsand the
same number of parameters. The point estimates from these moment conditions
correspond to the usual least squares estimates. We follow the literature and add
constants, or alphas. In the world CAPM, the aiMs should be equal to zero.Indeed, we
will evaluate the CAPM by checking whether the alphas are all equal tozero in the time
series. Our focus, however, is on the ability of the various modelsh and without sample
selectivity) to explain the cross-section of risk premia.
Note that we use the regime indicator variable to make our unbalanced panel abalanced
panel as in Bansal and Dahlquist (2000). That is, the moment conditions are multiplied
with the product of the regime indicators at time t and t + 1, yityit+1. The product
yityit+1 selects returns when markets are open both at time t and t + 1. In essence, this
procedure treats missing observations as zeros. This has a practical advantage since the
usual moment conditions which contain missing data can be filled with zeros, and then
standard GMM routines can be utilized.3
The sample selectivity part in equation (17) is gihi. As noted in the discussion of equation
(14), under simplifying assumptions, the probability of default can be recovered from the
sovereign bond spread. Further, this spread can be used to 3Hayashi (2000) considers,
also in an analysis of panel data, a similar approach. Stambaugh (1997) presents an
alternative approach to address this econometric issue.
14
completely characterize the hazard rate at time t. However, the data on sovereign interest
rate spreads are not available for many economies in our sample period. As shown
earlier, there is a high negative correlation between the country ratings and the spreads in
the cross-section (for economies where sovereign spread data are available). That is, a
country with a low rating tends to have a high spread (a high probability of default).
Consequently, to characterize the cross-section of hazardrates, we model the hazard rate
for market i as follows
gihi = (g00 + g01Ci) Ai, (21)
where Ai proxies for hi in the cross-section. For example, we let Ai equal the countryis
economic rating which then captures the cross-sectional variation in the hazardrate.
Further, to allow for controlled cross-sectional heterogeneity in gi, we model it as
gi = g00 + g01Ci, where Ci denotes a country-specific attribute such as its return
volatility, financial rating, or its trading activity.
The cross-sectional parameters (i.e., the risk premium parameters and the g00and g01
parameters) are then identified in the last set of moment conditions for each asset i
E [(rit+1 - lMbiM - lMzbiMz - g00Ai - g01AiCi) yityit+1] = 0. (22)
We also consider a specification with a constant term
E [(rit+1 - l0 - lMbiM - lMzbiMz - g00Ai - g01AiCi) yityit+1] = 0. (23)
The constant term l0 should be zero according to theory, and a non-zero constant
indicates that a model cannot price the assets on average. Alternatively, a non-zero
constant can be interpreted as a zero-beta rate different from the riskfree rate thatis
imposed. Note that all parameters including the betas and the cross-sectional parameters
l0, lM, lMz, g00, and g01 are jointly estimated using GMM. Details of the estimation are
given in Appendix B.
Results
This section presents the empirical results. Recall, that we earlier reported that the cross-
sectional dispersion in the average returns is fairly large for emerging markets and small
for developed markets. This cross-sectional dispersion poses a serious challenge to asset
pricing models. Variables that characterize the selectivity bias, such as country ratings,
have very little time-series variation, but considerablecross-sectional variation. Hence,
the effects of selectivity are primarily identifiable in the cross-section. Given the large
cross-sectional dispersion in the data along with the short data histories for many
emerging markets, we, as in Black, Jensen, and Scholes (1972), Fama and MacBeth
(1973), and Jagannathan and Wang (1996), focusprimarily on the explaining the cross-
sectional differences in risk-premia.
We first discuss the ability of the various models to capture the cross-section of average
returns through only systematic risk. We then include sample selectivity in the cross-
section. Finally, we discuss the results and provide further interpretations
of the results.
Evidence in the Absence of Selectivity
In Table 3, we provide evidence from the cross-section of asset returns. The estimated
risk premium for the market portfolio is negative, as can be seen in row 2 of Panel A.
This is a standard finding (see, for instance, Jagannathan and Wang, 1996). The ability of
the CAPM with constant betas to explain the cross-section of average returns is basically
zero as indicated by the adjusted R-square. In Panel B, we consider the CAPM where the
market betas are allowed to be time-varying. The model fails to capture the cross-
sectional dispersion in average returns in this specification as well The adjusted R-square
is only about 8%. The theoretical restriction that l0 = 0 can be rejected at the 5%
significance level. However, the constant term, as discussed below, is not particularly
relevant when sample selectivity is included in
the model. For completeness, we have conducted the time-series tests for both the
constant beta and timevaryingbeta versions of the CAPM (not reported in a table). We
find that the joint test of zero alphasis rejected in both cases. The rejections seem to be
primarily due to abnormal returns in emergingmarkets—this is consistent with Harvey
(1995) who also shows that CAPM implications are rejectedin emerging markets data.
The failure of the CAPM can also be seen in Figure 2 where we plot the averagereturns
against the predicted expected returns from the model. A true model would,ignoring
estimation errors, produce observations along the 45-degree line. The figure reveals that
there is almost no dispersion in predicted expected returns. Hence, the model does not
capture the large cross-sectional variation in average returns.
Evidence with Selectivity Included
The model specifications with sample selectivity are also reported in Table 3. Inthese
specifications the selectivity is modelled as (g00 + g01si)Ai, where si is the annual return
volatility for market i, and Ai is defined as the economic rating for country i less the
economic rating for the U.S. The expression (g00 + g01si) is the gi for country i. The
proxy for the hazard rate for country i is Ai, based on thereasoning provided earlier. Note
that Ai is negative for emerging economies and close to zero for developed economies.
This specification captures the intuition that as economies improve their economic rating
they become akin to developed marketsand the sample selectivity term would fall.When
sample selectivity is incorporated in the standard CAPM (in Panel A), thecross-sectional
R-square rises to 41% and the parameters associated with the selectivityterm are
significant (a p-value of 3%). The time-varying beta based CAPMwith sample selectivity
included is reported in Panel B. This specification does quitewell in capturing the cross-
sectionalvariation in risk premia, and has an adjustedR-square of 61%. The magnitudes of
the parameters that govern the selectivity bias(that is, g00 andg01) are similar across
different specifications. They are in all casesjointly significant at usual significance levels
(see the column labelled “Test ofJointSignificance”). The last two rows of Panel B
highlights the relevance, or the lackthereof, of the constant term l0. The empirical results
across thetwo cases (includingl0, or not) are comparable. Hence, as suggested by theory,
the constant term isnot particularly important.Table 4 provides themagnitudes of the
overall risk premia explained by systematicrisk and by sample selectivity. Economies
with poor economic rating have alarger andpositive selectivity bias. For developed
economies the variable Ai is essentiallyzero and hence the effect of selectivity on their
mean returns is absent.
the constant beta CAPM, the systematic risk contribution is about 0.45% per month for
both emerging and developed economies. However, the selectivity premium is0.50% per
month for emerging markets and close to zero for developed markets. Inthe model with
time-varying betas (see PanelB),the fraction of the emerging marketreturn attributed to
the selectivity bias is somewhat higher, and now stands at 0.58%per month. For
emergingmarkets more than 1/2 of the ex-post risk premium can beattributed to
selectivity. That is, sample selectivity seems to be the dominant influenceon the measured
risk premiums in emerging economies. Sample selectivity isnot an important dimension
for understanding measured risk premiain developedmarkets.The higher explanatory
ability of the world CAPM with time-varying betas andsample selectivity can be seen in
Figure 3which displays the average returns againstpredicted expected returns. The
improvement in fit is visible and the model is ableto produce the highdispersion in
average returns.We also considered alternative specifications for the parameter gi. In
particular,we replaced si with a reputationalvariable—the financial rating of an economy
i lessthe comparable rating for the U.S. The ability of this specification in terms of
capturingthecross-sectional variation in risk premia (i.e., adjusted R-square) is about30%.
This R-square is quite high relative the specifications withouttheselectivityeffects. As
shown in Table 4 the average emerging markets risk premium is still predominantlydue
to selectivity bias. Yet anotherchoice for the specification of gi, thetrading activity
variable, produces again similar results.Finally, we consider a specification where we use
thespreads (short samplesavailable for nine economies) on the Brady bonds to measure
the hazard rates directly.The approach is as follows. We projecttheaverage spreads on the
averagecountry ratings and use these projection coefficients to infer spreads and
defaulprobabilities (under theassumption of zero recovery as in Appendix A) for
allemerging markets. Recall that we are making the assumption that the probabilityof
default indebt markets coincides with the probability of expropriation. Fromthe
probabilities we can then compute the hazard rates (i.e., the his). For developedmarkets
we assume a zero default probability (and hence a zero hazard rate). Withthis measure of
the hazard rate, and the same specificationforthe gi as before, we estimate the cross-
sectional regression in equation (22). This specification capturesabout 36% of the cross-
sectional variation in risk premia. We find thatthe selectivity18term, that is gihi, is about
the 0.63% per month. The average probability of the defaultis about half a percent per
month. Hence, the bias in the mean return of about 0.63% per month can be supported by
rather small probability of default (risk of expropriation).Note that the empirical evidence
for this specification is quite similarto that discussed in the time-varying beta case in
Panel B of Table 3.
5.3. What Drives the Selectivity Bias?
In Panel A of Table 5 we inquire what economic variables can explain the
crosssectionaldispersion in the selectivity bias for emerging markets. In particular, weare
interested in whether the measured selectivity premium is related to trading activityand/or
measures of reputation. To do so, we considerthe measures of theselectivity premium
based on the specification where gi = (g00 + g01si), and therelative economic rating is
the proxy for thehazard rate. This specification was reportedin Table 3. The reputational
variable (the financial rating of country i lessthe comparable rating for theU.S.)is able to
explain about 32% of the dispersion inthe selectivity premium. This regression also
shows that the selectivity premiumrises as thecountry’s financial rating falls. Similarly,
when we use the tradingactivity variable, this explains about 19% of the dispersion in the
selectivitypremium.Economies with larger trading activity have a smaller selectivity
premium.In essence our evidence suggests that both trading activityand reputational
considerationsare important for explaining the selectivity premium.Allowing the
selectivity premium to depend on the volatilityinthe cross-sectionis motivated by
arguments presented in Brown, Goetzmann, and Ross (1995). Ourevidence indicates that
this attribute is notuniquely important to capture the crosssectionaldifferences in risk
premia. Indeed, the trade activity and financial reputationvariables do, at least
ineconomic terms, a comparable job of explaining the cross-sectional differences in the
risk premia. Thus, it seems to us that this is dueto the fact that the volatility of returns are
related to these variables.Thisis shownin Panel B of Table 5: return volatility is
decreasing in both trading activity andfinancial reputation. These variables, based on the
work of Eaton and Gersovitz(1981), and Bulow and Rogoff (1989a, 1989b) should matter
to the compensationthat emerging markets have toadditionallypay, due to risks of
expropriation. Wefind that this indeed is the case.
Conclusion
In this paper we show that the cross-sectional differences in the equity returns
acrosssovereign economies is determined by two features—systematic risk and a
selectivitypremium. We show that the selectivity premium captures more than 1/2 of
theaverage risk premium in emerging markets. The equity riskpremia in
developedmarkets seems to be driven solely by systematic risk. The main economic
implicationof this result is that after taking account ofselectivity premium all
internationalequity returns reflect systematic risk, as predicted by theory.Our empirical
work also shows that sovereigns thathave better financial marketreputations and trade
more actively have to pay a smaller selectivity premium. Thisempirical evidence lends
support to theview that both reputations and fear of tradesanctions are important in
determining the cost of equity borrowing for a sovereignnation.
Measuring Hazard Rates From Sovereign Spreads
This Appendix shows how the hazard rate can be measured from sovereign bondspreads.
Consider a dollar denominated pure discount bond issued by a country.The payoff is
equal to one if there is no default, and µb + bbet+1 +hbt+1 if the countrydefaults. The
payoff process can thus be written as qbt+1= ybt+1 + (1 - ybt+1) (µb + bbet+1 + hbt+1) .
(24)
For simplicity, we assume that bb = 0. That is, we assume that the recovery valueof the
bond is not related to the systematic risk in the world economy.Further, theexpected
payoff on this bond in default is less than one (i.e., µb < 1). Valuing thispayoff using the
stochastic discount factor implies that
1/Rbt = [pbt + (1 - pbt) µb] /Rf t. (25)
Solving for the probability of no default, we obtain
pbt =Rf t - µbRbtRbt (1 - µb). (26)
Assume that the probabilities of default for the bond correspond to the probabilityof a
market shut-down, that is, pbt = pit. Under the further assumptionthat therecovery rate is
zero, we can directly recover the probability of default. Further,given the normal
cumulative distribution function wecan completely characterizethe hazard rate.The above
expression can also be used to compute the ex-ante beta on market i.We denote this with
bit =pitbi1 + (1 - pit) bi2. If we assume that the ratio of thebetas across the two regimes is
equal to a constant c and µb = 0, it follows that
bit = bi1 _Rf tRbt + c _Rbt - Rf tRbt __. (27)
Note that Rf t and Rbt can be observed directly from U.S. Treasuries and
Sovereignbonds, or as we demonstrate, approximated with a country’s relativecountry
rating.Hence, conditional on c, one can estimate the model with both a peso problem
andsample selectivity. In the special case with c = 1,there is no peso problem and wehave
that bit = bi1 = bi2.21B. Estimation DetailsThis Appendix shows the estimation in more
detail. Let q0 denote thetrue parametervector that we want to estimate. The typical
elements in q0 are aiM, biM, and biMz that are specific to each market, and the common
parameters l0, lM, lMz, g00and g01. By stacking the sample counterparts of the moment
conditions in (18) to (20), and (23), we have a vector of moment conditions
gT (q) =1T Tåt=1f (Xt, q) , (28)
where Xt summarizes the data used to form the moments conditions. The vectorgT (q)
has the dimension 4N. The moment conditions, given by (18) to (20), exactlyidentify the
aiM, biM, and biMz parameters. However, the moment conditions, givenby (23), is
overidentified. We have N moment conditions, but only 5 parameters (l0,lM, lMz, g00
and g01).We estimate the parameters by setting linear combinations of gT equal to zero.
That is, the moment conditions can be written as
ATgT = 0, (29)
where AT is a (3N + 5) × 4N matrix. In particular, our choice of AT is designed toensure
that the point estimates are the ones given by ordinary least squares.
Let AT
be the product of two matrices denoted by A1T and A2T (that is, AT = A1TA2T).
Thefollowing matrices result in least square point estimates
A1T =26666666664I3N 03N · · · 03N003N 1 · · · 1003Nˆ b1M · · · ˆ bNM003Nˆ b1Mz ·
· · ˆ bNMz003N A1 · · · AN003N C1A1 · · · CNAN, (30)
where I3N is the identity matrix with dimension 3N, 03N is a 3N vector of zeros,0N is an
N vector of zeros, and A2T is a diagonal matrix with typical element equalto 1/ åTt=1
yit+1. The ˆ biMs and ˆ biMzs are estimates of biMs and biMzs, and they are
given in the estimation. The ˆ biMs and ˆ biMzs are exactly the least square
estimatesobtained in a regression of the assets’ excess returns on the market excess
returnand scaled market excess returns as in (18) to (20). Further, the estimates of l0,
lM,lMz, g00, and g01 coincide with the least square estimates obtained in a regressionof
average returns on the betas and the proxies for sample selectivity. Our choice ofAT
ensures that ATgT (qT) = 0.
Based on Hansen (1982) we know that when linear combinations of gT are setequal to
zero as in (29), the asymptotic distribution of the point estimator qT is given
by
pT (qT - q0) d
! N _0, (A0D0)-1 _A0S0A00_(A0D0)-10_, (31)
where D0 is the gradient of the moment conditions in (28), and where S0 is the
variance-covariance matrix of the moment conditions and given by
S0 =¥åj=-¥E hf (Xt, q0) f _Xt-j, q0_0i. (32) The sample counterpart ST is estimated
using the procedure in Newey and West(1987) with four lags. D0 and A0 can be
estimated by their sample counterparts DTand AT. Note that the standard errors based on
(31) are robust to heteroskedasticity
and serial correlation in the moment conditions.
Table 1: Summary Statistics of Global Equity Returns
Mean Standard Deviaiion coreallation With world nclusion
Panel A. Developed Markets
Australia 1.07 6.84 0.52 84-01 203
Austria 1.16 7.33 0.34 84-01 203
Belgium 1.60 5.55 0.64 84-01 203
Canada 0.99 5.13 0.70 84-01 203
Denmark 1.18 5.66 0.53 84-01 203
Finland 1.91 8.62 0.54 88-01 156
France 1.59 6.07 0.70 84-01 203
Germany 1.38 6.27 0.60 84-01 203
Hong Kong 1.84 8.76 0.53 84-01 203
Ireland 1.03 5.73 0.65 88-01 156
Italy 1.46 7.51 0.51 84-01 203
Japan 1.02 7.36 0.76 84-01 203
Netherlands 1.56 4.73 0.75 84-01 203
New Zealand 0.30 7.02 0.47 88-01 156
Norway 1.09 7.26 0.58 84-01 203
Singapore 0.85 8.05 0.54 84-01 203
Spain 1.81 7.03 0.66 84-01 203
Sweden 1.65 6.92 0.63 84-01 203
Switzerland 1.51 5.38 0.66 84-01 203
U.K. 1.33 5.41 0.76 84-01 203
U.S. 1.38 4.37 0.79 84-01 203
Average 1.32 6.52 0.61
Panel B. Emerging Markets
Argentina 3.89 23.46 0.06 89-01 144
Brazil 3.16 19.61 0.31 89-01 144
Chile 1.83 7.73 0.24 89-01 144
China 0.26 13.41 0.27 93-01 96
Colombia 1.43 10.91 0.10 91-03 118
Greece 2.18 12.22 0.19 89-01 144
Hungary 1.61 13.15 0.47 93-01 96
India 0.36 8.68 0.13 92-12 97
Indonesia -0.08 15.13 0.41 90-10 123
Jordan 0.69 4.85 0.22 89-01 144
Korea 0.37 14.15 0.39 92-02 107
Malaysia 0.23 10.12 0.39 89-01 116
Mexico 2.04 10.18 0.42 89-01 144
Pakistan 0.99 12.56 0.08 91-04 117
Peru 0.75 9.14 0.34 93-01 96
Philippines 0.42 11.42 0.40 89-01 144
Poland 3.24 17.91 0.37 93-01 96
Portugal 0.96 6.91 0.51 89-01 144
South Africa 0.98 8.34 0.53 93-01 96
Sri Lanka -0.22 10.07 0.31 93-01 96
Taiwan 0.72 10.50 0.37 91-02 119
Thailand 0.38 12.55 0.43 89-01 144
Turkey 2.48 19.39 0.16 89-09 136
Venezuela 3.00 17.43 0.02 90-02 131
Zimbabwe 1.80 12.81 0.21 93-07 90
Average 1.34 12.50 0.29
Panel C. World
World 1.21 4.24 1.00 84-01 203
This table presents summary statistics of monthly dollar returns in global equity markets
from nclusion date to November 2000. Panels A, B and C show statistics for developed
markets, emrging markets and theWorld, respectively. The labels Average in Panels A
and B refer to theaverage (equally-weighted) across developed and emerging markets,
respectively. The meansand standard deviations are expressed in % per month.
Correlation with World refers to thecorrelation coefficient with the world market
portfolio. The inclusion date (year-month) is thefirst month with observations of
investable returns. The last observation of Malaysia is August1998. T refers to the
number of observations for each market.
Table 2: Country Attributes
Real GDP pr Capita
Trading
Activity
Economic
Rating
Financial
Rating Beta
Panel A. Developed Markets
Australia 14,445 34.43 82.3 75.8 0.84
Austria 12,695 79.18 91.1 80.3 0.57
Belgium 13,232 144.96 87.6 78.6 0.83
Canada 17,173 51.24 89.2 78.5 0.85
Denmark 13,909 65.26 86.4 79.2 0.71
Finland 14,059 47.67 84.5 76.3 1.16
France 13,904 45.16 86.3 77.4 1.00
Germany 14,628 58.03 93.0 81.9 0.88
Hong Kong 14,849 262.96 83.7 78.4 1.11
Ireland 9,274 114.60 85.3 80.5 0.93
Italy 12,488 41.46 84.9 76.0 0.90
Japan 14,331 20.92 96.1 84.4 1.32
Netherlands 13,029 103.72 90.2 83.8 0.84
New Zealand 11,513 55.34 84.3 75.4 0.82
Norway 14,902 81.11 92.6 86.5 0.99
Singapore 11,710 373.26 89.2 83.9 1.03
Spain 9,583 37.52 81.2 75.1 1.08
Sweden 14,762 59.46 85.8 78.1 1.04
Switzerland 16,505 72.73 98.0 85.5 0.84
U.K. 13,217 51.48 90.7 73.2 0.97
U.S. 18,054 21.50 92.3 76.7 0.82
Average 13,727 86.76 88.3 79.3 0.93
Panel B. Emerging Markets
Argentina 4,706 15.18 63.4 60.6 0.31
Brazil 4,042 12.66 65.7 57.0 1.46
Chile 4,338 65.46 81.0 73.7 0.46
China 1,324 25.42 82.1 74.7 0.97
Colombia 3,300 35.38 75.0 67.4 0.31
Greece 6,768 54.16 68.9 68.5 0.54
Hungary 5,357 60.67 76.4 66.0 1.66
India 1,264 18.76 73.8 67.4 0.31
Indonesia 1,974 52.61 75.2 66.7 1.67
Jordan 2,919 144.21 64.0 71.4 0.26
Korea 6,673 62.48 85.9 78.8 1.52
Malaysia 5,124 154.20 83.5 81.1 1.04
Mexico 5,827 32.72 74.4 63.1 1.04
Pakistan 1,394 35.01 60.6 62.0 0.28
Peru 2,188 26.80 68.5 65.4 0.84
Philippines 1,763 61.48 64.0 66.7 1.12
Poland 3,820 45.84 77.8 72.1 1.80
Portugal 7,478 75.20 82.2 79.2 0.87
South Africa 3,248 47.22 75.3 71.2 1.18
Sri Lanka 2,096 67.37 69.2 68.1 0.86
Taiwan 8,063 89.88 92.6 86.5 1.03
Thailand 3,580 75.83 81.1 75.3 1.32
Turkey 3,741 41.99 60.3 55.4 0.73
Venezuela 6,055 59.64 72.6 64.5 0.07
Zimbabwe 1,182 59.00 55.7 56.0 0.74
Average 3,929 56.77 73.2 68.8 0.90
This table lists country attributes. Panels A and B show the attributes for developed
markets andemerging markets, respectively. The labels Average in Panels A and B refer
to the average (equallyweighted)across developed and emerging markets, respectively.
Real GDP per capita refers to RealGDP per capita in constant dollars (expressed in
international prices, base 1985). Trading Activityrefers to exports plus imports over
nominal GDP. Real GDP per capita and trading activity aretaken from the Penn World
Table for the year of 1990. Economic and Financial Ratings refer tothe average financial
and economic country rating provided by International Country Risk Guidefrom
inclusion date to November 2000. Beta refers to the slope-coefficient in a regression on
amarket’s excess return versus the excess return on the MSCI world market portfolio.
Table 3: Cross-Sectional Estimates of Risk and Sample Selectivity Premia
World Market Sample Selectivity Adjusted Tests of Joint Significance
R-square
Constant Market Conditional Relative
Economic Rating
Volatility × Relative
Economic Rating
World
Market
Sample
Selectivity All
l0 lM lMz g00 g01
Panel A: World CAPM with Constant Betas
0.75 -0.22 [0.12] [0.12]
(0.49)
1.08 -0.26 0.01 [0.69] [0.69]
(0.61) (0.65)
0.49 7.47 -0.82 0.32 [0.25] [0.03] [0.07]
(0.32) (4.81) (0.34)
0.74 -0.18 7.90 -0.80 0.41 [0.79] [0.03] [0.07]
(0.51) (0.67) (4.69) (0.32)
Panel B: World CAPM with Time-Varying Betas
0.78 1.36 -0.16 [0.11] [0.11]
(0.46) (0.66)
1.09 -0.18 0.13 0.08 [0.48] [0.48]
(0.55) (0.59) (0.82)
0.51 1.16 7.30 -0.86 0.48 [0.12] [0.03] [0.02]
(0.41) (0.62) (4.88) (0.35)
0.82 -0.18 0.26 7.92 -0.85 0.61 [0.27] [0.02] [0.06]
(0.47) (0.62) (0.85) (4.86) (0.33)
This table presents results from estimations of the asset pricing models. All markets are
estimated in one common system which includes both a timeseries estimation of betas as
well as a cross-sectional estimation of risk premia and sample selectivity premia. PaneA
and B show the results for the worldCAPM with constant and time-varying betas,
respectively. Autocorrelation and heteroscedasticityconsistent standard errors for the
estimatedcoefficients are reported in parentheses. The Adjusted R-square reports the
adjustedcoefficient of determination between fitted returnsgenerated by the model and
actual realized returns on the assets. Average pricing errorrefers to the cross-sectional
average (equally-weighted) ofthe pricing errors, and is reported for Developed and
Emerging markets.Tests of joint significance report p-values from tests of jointly
significant riskpremia associated with the market, sample selectivity, andall premia.
Table 4: Decomposition of Average Excess Returns
Average Systematic
Risk
Sample
Selectivity
Pricing
Error
Panel A: World CAPM with Constant Betas
Volatility Attribute
Developed Markets 0.79 0.45 0.07 0.27
Emerging Markets 0.89 0.44 0.50 -0.05
Financial Rating Attribute
Developed Markets 0.79 0.50 0.00 0.29
Emerging Markets 0.89 0.48 0.50 -0.09
Trading Activity Attribute
Developed Markets 0.79 0.50 -0.05 0.33
Emerging Markets 0.89 0.51 0.49 -0.09
Panel B: World CAPM with Time-Varying Betas
Volatility Attribute
Developed Markets 0.79 0.46 0.06 0.28
Emerging Markets 0.89 0.33 0.58 -0.01
Financial Rating Attribute
Developed Markets 0.79 0.50 0.01 0.28
Emerging Markets 0.89 0.36 0.59 -0.07
Trading Activity Attribute
Developed Markets 0.79 0.50 -0.05 0.34
Emerging Markets 0.89 0.37 0.58 -0.06
This table presents the decomposition of the measured excess returns for developed and
emerging markets (equally-weighted averages) generated by
models with sample selectivity. Panel A and B show the decomposition for the world
CAPM with constant and time-varying betas, respectively, as in Table3.
The specification with the volatility attribute is as reported in Table 3 (withno constant
term). The specifications with financial rating and trading activity
use these attributes instead of the volatility attribute. The decompositions are expressed in
% per month. Average refers to the the average excess return
from inclusion date. Systematic risk refers to the contribution of market
components.Sample selectivity refers to contribution due to sample selectivity.
Pricing error refers to the average pricing error.
Table 5: Sample Selectivity and Volatility Projections
Constant Economic
Rating
Financial
Rating
Trading
Activity Volatility Adjusted
R-square
Test of Joint
Significance
Panel A: Sample Selectivity Projections
-0.15 -6.16 0.49
(0.14) (1.43)
-0.24 -5.05 0.32
(0.17) (1.57)
1.00 -0.26 0.19
(0.29) (0.10)
-1.60 0.17 0.74
(0.27) (0.02)
-1.76 -1.46 -2.15 0.01 0.14 0.87 [0.00]
(0.26) (1.73) (1.32) (0.05) (0.02)
Panel B: Volatility Projections
(0.17) (1.16) (0.89) (0.04) (0.01)
10.08 -20.76 0.20
(0.98) (6.99)
14.65 -1.30 0.19
(1.90) (18.65) (16.24) (0.87)
This table presents the results of cross-sectional regressions for 25 emerging markets.
Heteroscedasticityconsistent standard errors for the estimatedcoefficients are reported in
parentheses.Panel A and B show the results for sample selectivity and volatility,
respectively, on variouscountryattributes. Sample selectivity is measured as the part of
the measured equity premiumdue to sample selectivity as obtained in the worldCAPM
with time-varying betas using relativeeconomic rating, and volatility × relative economic
rating as in Table 3 (with no constant term).Theattributes are the economic rating, the
financial rating, and the relative trading activity. Allattributes are relative the U.S. The
Adjusted Rsquarereports the adjusted coefficient of determination.Test of joint
significance reports p-values (within square brackets) from a test of jointlysignificant
coefficients.
Securities and Exchange Board of India Group
on Secondary Market Risk Management
1 Background
The SEBI Group on Secondary Market Risk Management discussed the
introduction of interest rate derivatives in India at its meeting on March
12, 2003. The Group’sdeliberations covered:
The time table for introduction of exchange traded interest rate
derivatives inIndia
The specification of the initial set of interest rate derivative contracts
to beintroduced
The road map for introduction of additional products
The risk containment systems for the initial set of derivatives
The road map for research in fixed income analytics and the resulting
refinementof product design and fine tuning of the margining
system.In line with the Group’s view that its conclusions on these
issues be put up for public comments, the Group has prepared this
consultative document.
2 Need for Exchange Traded Derivative Products
The Reserve Bank of India’s Working Group on OTC Rupee Derivatives
has stated the need for exchange traded interest rate derivatives
admirably well:
“While OTC derivatives market has traditionally played a dominant role
in debt markets globally and would continue to do so in future, it is
desirable to supplement the OTC market by an active exchange-traded
derivative market. In fact, those who provide OTC derivative products
can hedge their risks through the use of exchangetradedderivatives. In
India, in the absence of exchange-traded derivatives, the risk ofthe
OTC derivatives market cannt be hedged effectively. Exchange-
tradedderivative market has the following features: an electronic
exchange mechanic m andemphasises anonymous trading, full
transparency, use of computers for ordematching, centralisation of
order flow, price-time pr ority for order matching, largeinvestor base,
wide geographical access, lower cost of intermediation, settlement
guarantee, better risk management, enhanced regulatory discipline,
etc. At present, inIndia, there exists a reasonable OTC market for
interest rate products which raises the need for exchange-traded
interest rate derivatives products.” 2“Also, some of the features of OTC
derivatives markets embody risks to financialmarket stability, viz., (i)
the dynamic nature of gross credit exposures, (ii) information
asymmetries and lack of transparency, (iii) the high concentration
ofOTC derivative activities in major institutions, and (iv) the dominance
of OTC derivatives markets in the global financial system. Instability
arises when shocks,such as counterparty credit events and sharp
movements in asset prices that underlie derivative contracts, occur
which significantly alter the perceptions of current and potential future
credit exposures. When underlying asset prices change rapidly, the
size and configuration of counterparty exposures can become
unsustainably large and provoke a rapid unwinding of positions.”
(Paragraph 3.2) “The Group felt that there is a need for exchange-
traded interest rate derivatives (IRDs) as debtmarket volumes,
particularly in IRS, have been growing rapidly and exchange-traded
products would reduce the risk substantially through a clearing
corporation, novation, multilateral netting, centralised settlement and
risk management. The Group considered that India has already set up
mature institutional infrastructure for trading, clearing and settlement
in the equity markets which could be harnessed for the debt market. It,
therefore, proposes to allow trading in IRDs through the anonymous
order- driven screen-based trading system of the stockexchanges
which will facilitate participation byallclasses of investors and
increasemarket access across the country.” (Paragraph 3.3)“…interest
rate futures, interest rateoptions, interest rate swaps – both plain
vanilla swaps as well as swaps with embedded options like
caps/floors/collars, as well asstandardised repos may be allowed to be
traded on the stock exchanges.” (Paragraph 3.5)
It must also be added that interest rate risk is one of the most
pervasive risks in the
economy that affects not only the financial sector, but also the
corporate and householdsectors. The critical importance of interest
rate risk management for banks and financialinstitutions is well
understood, and its increasing importance for the corporate sector in
ahousehold financial savings on the assets side and the increasing
amount of housing loans on the liabilities side makes interest rate risk
increasingly important for the household sector. It is in fact likely that
for many households interest rate risk is vastly more important than
equity market risk. Moreover, because of the Fisher effect, interest
rateproducts are the primary mechanism available to hedge inflation
risk which is typicallynthe single most important macroeconomic risk
facing the household sector. In thiscontext, therefore, it is important
that the financial system provides the household sectorgreater access
to interest rate risk management tools through exchange traded
derivatives.Exchange traded derivatives are also potentially very
attractive to the corporate sector and to the financial sector.
3 Time Table for Introduction
The Group is well aware that the publicly available fixed income
analytics in India is not adequate for the development of a vibrant
derivatives market. While recognizing thatfixed income analytics is
among the most intellectually challenging parts of modern finance, the
Group is of the view that the development of this field has been held
back in India less by a lack of supply and more by the paucity of
demand. Therefore, rapid development of new markets and products is
the best way to solve the “chicken and egg”problem of whether the
market comes first or the analytics comes first.The Group recommends
therefore that the first set of exchange traded interest ratederivatives
start trading almost immediately. In consultation with the exchanges,
the Group has arrived at April 21 as the most feasible launch date. As
discussed later, the Group desires and expects that the Exchanges
would spearhead a concerted research effort in fixed income analytics
over the next three months. Allowing for a month for regulatory review
and two months for software changes, the Group desires and expects
that the improved product designs and more fine tuned
marginingsystems arising out of this researchwould be implemented
within a period of six months. Early launch has three major
implications:
1. The primary implication of early launch is a significant degree of
over-marginingin the initial six month period until more refined models
are implemented. Sincethere can be no compromise on the issue of
market safety, the Group is compensating for model risk by aggressive
over-margining.
2. Another implication is that to the extent that the zero coupon yield
curve that ispublicly available currently is not fully accurate, there
would be a basis risk inhedging interest rate risk using products based
on this curve. Since all hedgesInvolve e some degree of basis risk, the
Group does not regard this as a showstopper if there is complete
transparency regarding the construction of the yieldcurve. The yield
curve provider has been given clear directions to achieve this.The
Group desires and expects that an improved yield curve be
implementedwithin six months. The yield curve provider has been
given quantitativebenchmarks in this regard. These issues are
discussed more fully in 4.3 below.
3. For software reasons, it would be possible to allow only two decimal
places in theprice quotes at the time of launch.
Exchanges will howevermodifytheir softwareto allow four decimal
places as soon as possible. The National Stock Exchangehas indicated
that this could beaccomplishedby midMay, that is to say, withinone
month of launch. It must also be pointed out that one paisa is a small
numbercompared with theoneday standard deviation of most bond
prices, so thelimitation of two decimal places does not significantly
detract from the utility ofthe product for hedging purposes.
4 Product Specification: Long Bond Futures
4.1 Maturity and Coupon
The Group discussed three issues in connexion with product design for
the futures on long maturity risk free bonds:
1. The maturity of the underlying long bond
2. Maturities of the futures contracts
3. The characteristics of the bond
Regarding maturity of the underlying, there was unanimity that the
most liquid segment in the government securities market is the ten
year maturity and that this should be the maturity of the underlying
long bond for interest rate futures. Regarding maturity of the futures
contracts, it was decided that exchanges would be free to introduce
contracts up to a maximum maturity of one year. Exchanges would be
free to decide whether to have quarterly contracts beyond the first
three months, and whether the quarters should be fixed months of the
year or rolling quarterly horizons from the Contract introduction date.
There was some discussion regarding the choice between whether the
underlying should be a coupon bond or a zero coupon bond. The Group
decided that the exchanges should have the freedom to offer either or
both of these products and also to choose the coupon rate in case of
the coupon bond. Exchanges indicated that the coupon rates could be
in therange of 6-8%.
4.2 Physical Settlement versus Cash Settlement
The Group deliberated at length on the merits of physical and cash
settlement. Several advantages were identified with physical
settlement:
1. Some members felt that physical settlement would improve the
linkages betweenthe derivative market and the underlying market.
2. Some members also felt that the requirement of physical settlement
would act as a restraining force on speculators and reduce volatility.
3. Some members also felt that physical settlement might reduce basis
risk.
On the other hand, several disadvantages of physical settlement were
also noted:
4. First and foremost was the problem of market manipulation. Relative
to global standards, Indian banks hold a large fraction of their assets in
government securities.
5. Moreover, the issue size of any single government security is also
relatively small. This means that the entire issued amount of even a
very liquid overnment security is sometimes less than 10% of the
aggregate government urities holding of a single large player. This
means that a large player couldeasily corner the entire floating stock
of any specific issue. Gradual opening up ofthe Indian financial sector
to global players adds to the vulnerability of mostgovernment
securities to market manipulations. Most physically settled long
bondfutures deal with the problem of squeezes by allowing several
different bonds tbe delivered with different conversion factors.
However, even in this system,there is a “cheapest to deliver” bond
that could become the target of manipulation.
The alleged squeeze1 of German government bond futures in March
2001 by alarge regulated entity in that country was also referred to as
a pointer to whatcould happen in India. The “cheapest to deliver” bond
that was allegedlysqueezed in that case had an issue size of over _6b,
much larger than the issuesizes in the Indian market, but this issue
size was still regarded as too small.
2. Some participants (particularly households) may not have ready and
easy accessto the government securities market to achieve physical
delivery at low cost. Thecurrent government securities market is not
transparently and publicly accessibleto many of these participants.
3. The absence of short selling makes physical settlement more
problematic.After carefully deliberating on both sides of the issue, the
Group decided that the ten yearlong bond futures should initially be
launched with cash settlement.
4.3 The Zero Coupon Yield Curve
The Group then discussed the issue of determining the settlement
price. It was decidedthat the settlement price should be based on the
value of the notional bond determinedusing the zero coupon yield
curve computed by the yield curve provider designated bythe
exchange.If the notional bond is zero coupon, the settlement price of
the bond is simply the presentalue of the principal payment (at the end
of 10 years) discounted for 10 years at the 10year zero coupon yield.
For example, suppose that on 18/1/2003, the ten year zero
yieldpublished by the yield curve provider was 5.9023%
(annuallycompounded) implying aprice for the ten year zero of
56.3568 (100 x 1.059023-10). On the next working day,20/1/2003,
suppose that the yieldwas 5.8492% (annually compounded) implying a
priceof 56.6401. A person who bought a ten year zero coupon future
on 18/1/2003 would have
a mark to market gain of 28.33 paise on 20/1/2003.If the notional bond
has a coupon, the present value of the bond is obtained as the sum
ofpresent value of the principal payment discounted at the 10 year
zero coupon yield andthe present values of the coupons obtained by
discounting each coupon payment for thetime period remaining till the
coupon payment at the zero coupon yield for that maturity.The Group
decided that the following obligations should be imposed on the
designated Yeld curve provider (which could be the exchange itself or
a third party):
1. The yield curve should be computed by a completely objective
process withoutany element of human judgement so that any market
participant could arrive atthe same yield curve by applying the
published computation algorithm to publiclyavailable data.
2. The computation algorithm must be fully disclosed to the public. The
onlyeffective way of disclosing a computation algorithm is to disclose
the source
code. The yield curve provider would therefore be required to make
the entiresource code for its algorithm available on the web site under
a GNU GeneralPublic Licence. This requirement extends also to the
source code of thealgorithms used to convert a traded bond into a
series of cash flows and any othersimilar pre-processing that may be
carried out prior to the actual estimation itself.It also extends to the
algorithms that convert yield curve parameters into actualzero coupon
yields, the algorithms to value a notional bond given the zero
couponyields and any other similar post processing that may be
carried out after theactual estimation itself.
3. The yield curve provider would be required to make available on the
web site aset of at least 25 trading days (i.e. one month) of data suites
for the input data.Each day’s data suite would include the traded
prices and other transaction data2that is input into the estimation
algorithm. The data suite would also include thesample output from
the algorithm on this data suite. Similar data suites andsample outputs
must also be provided for the pre- processing and post-
processingalgorithms referred to above.
4. The yield curve provider would be required to make available on the
web site thefull time-series of yield curve parameters for as long a
period as possible. Thisdata set must extend back at least to April 1,
1999. The term yield curveparameters is used because many
estimation methods represent the entire yieldcurve as a functional
form that depends on a small number of parameters. Forexample, the
Nelson-Siegel functional form has four parameters. A cubic splinecould
have as parameters the location of the knots and the coefficients of
thecubic polynomial over each segment; alternatively theparameters
may be thecoefficients over a family of basis splines.
5 Product Specification: Notional T-Bill Futures
The Group agreed that there should be a future on a short term
interest rate. The choicehere is between a contract on an inter-bank
rate and a contract on a risk free interest rate.It was generally agreed
that the liquidity in the inter-bank market (MIBOR and MIFOR)is much
higher than in the T-Bills and in dated securities with a residual
maturity of lessthan one year. A contract on the inter-bank rate would
therefore be highly desirable. TheGroup washowever informed that
some legal issues have been raised on whethercontracts on indices like
MIBOR would fall within the definition of derivativesunder theSCRA.
Since the Group does not profess any expertise in legal matters, it
decided tosidestep the legal risk by focussing on contracts on the risk
free interest rate. The productthat is proposed is futures on notional T-
Bills with a maturity of 91 days (three months).Many of the issues
hereare similar to that for long bond futures and the
Group’srecommendations are also similar:
1. The contract would be cash settled.
2. The settlement price would be based on the risk free zero coupon
yield curve. Allthe objectivity and transparency requirements discussed
in 4.3 above would applyin this case also.
3. Exchanges would be free to introduce contracts up to a maximum
maturity of oneyear. Exchanges would be free to decide whether to
have quarterly contractsbeyond the first three months, and whether
the quarters should be fixed months ofthe year or rolling quarterly
horizons from the contract introduction date.
Chapter 5 Project work
Calculation of Beta for Nestle (I) ltd.
SensexMkt. price Mkt.return
Stock ret. xy x2 y2
july,2006 x y4 10520.11 1029.955 10410.49 1066.4 3.53 -1.04 -3.67 1.08 12.466 10436.84 1091.3 2.33 0.253 0.59 0.06 5.427 10509.53 1106.9 1.43 0.7 1 0.49 2.05
10 10684.3 1108.05 0.14 1.66 0.23 2.75 0.01
Total 7.43 1.573 -1.85 4.38 19.94
For Nestle ltd. Beta is calculated as
Beta= n. sum. xy –(sum. x)(sum.y) / n. sumx2 –(sum x)2
Beta calculated by putting the values is -1.25 for nestle ltd.
Calcualtion of Beta for Sh. Cement ltd.
Sh. Cements
SensexMkt. price Mkt.return
Stock ret. xy x2 y2
x y10520.11 720.2510410.49 755.1 3.53 35.15 124 1.08 1235.510436.84 775.85 2.33 20.75 48.34 0.06 430.5610509.53 800.2 1.43 24.35 34.82 0.49 593
10684.3 840.2 0.14 40 5.6 2.75 1600
7.43 120.25 212.76 4.38 3859.06
The calculation farmula for the beta is
Beta= n. sum. xy –(sum. x)(sum.y) / n. sumx2 –(sum x)2
The beta for the Sh. Cement ltd. is 1.12
Calculation of beta for Jindal steel ltd.
Jindal industries ltd.
SensexMkt. price Mkt.return
Stock ret. xy x2 y2
x y10520.11 1409.1510410.49 1448.15 3.53 39 137.67 1.08 152110436.84 1467.15 2.33 19 44.27 0.06 36110509.53 1500.25 1.43 33.1 47.33 0.49 1095.61
10684.3 1523.25 0.14 23 3.22 2.75 529
7.43 114.1 232.49 4.38 3506.61
The formula for calculation of beta is
Beta= n. sum. xy –(sum. x)(sum.y) / n. sumx2 –(sum x)2
And the beta for jindal steel ltd. is -2.18
CONCLUSION
Investors want to maximize expected return subject to their tolerance for risk. Such
returns take the form of dividend and /or interest income and appreciation in the price for
the asset held.
The risk associated with holding common stock is really the likelihood that expected
returns will not materialize. If dividends or price appreciation fall swhort of expectations,
the investor is disappointed. The principle sources behind dividend and price appreciation
uncertainties are forces and factors that are either controllable or not subject to control by
the firm.
Uncontrollable forces, called sources of systematic risk, include market, interest rate and
purchasing power risks. Market risk reflects changes in investor attitudes toward equities
in general that stem from tangible and intangible events. Tangible events might include
expectations of lower expected profits and the resultant panic selling. Interest rate risk
and purchasing power risk are associated with changes in the price of money and other
goods and services. Increase in interest rates (the price of money) cause the prices of all
types of securities to be marked down. Rising prices of goods and services ( inflation or
purchasing power changes) have an adverse effect on security prices because the
postponement of consumption through any form of investment means less “real” buying
power in the future.
The principal sources of unsystematic risk affecting the holding of common stocks are
business risk and financial risk. Business risk refers to changes in the operating
environment of the firm and how the firm adapts to them. Financial risk is associated
with the debt debt and equity mix of financing the firm. Operating profits can be
magnified up or down, depending upon the extent to which debt financing is employed
and under what terms.
Total risk of aninvestment can be thought of as consisting of two components:
diversifiable and non diversifiable risk . the former risk can be almost entirely by holding
a large enough mix of carefully selected securities.
The only risk an investor is compensated for taking is thus non diversifiable risk. Beta
measures this risk and can be used to determine the appropriate required return on a
security.
SUGESSTIONS
1. Sharekhan ltd. should also issue an IPO so that the company can be promoted
much because of extra financial resources.
2. Sharekhan ltd. should reduce the Dmat opening charges.
3. It should lessen the brokerage charges so that it can cut the competitors easily.
4. It should also open its branches in small cities .
5. It should also adopt multimedia advertisement by way of T.V. and Mobile
Phones.
BIBLIOGRAPHY
Websites:
www.sharekhan.com /companyinfo
dates of site visited: 1st july,2006 , 2nd july 2006, 5th july 2006,
www.google.com / risk and return analysis.
dates of site visited: 1st july,2006 , 2nd july 2006, 5th july 2006
Book:
Security Analysis and Risk Management
Author: Donald E. Fischer
Publisher: Prentice Hall of India Pvt. Ltd.
Page no. reffered 90
Magazine:
Valueline
Publisher: ShareKhan
Newspaper
The Economic Times
Dates: july,2006 4,5, 6,7,10
Pages referred: Compulsory Rolling Stocks On BSE/NSE.
Market Intelligence.
Annexure 1
Market information of Nestle I)
Closing Prices
July,2006 Rs.
4 1029.95
5 1066.4
6 1091.3
7 1106.9
10 1108.5
Wk. ago 1003.35
Month ago 960960.95
Face Value Rs.10
P/E ratio 33.4
Market capitalization Rs.10684 cr.
52 week high 1348
52 week low 700
Annexure2
Market information of Jindal Steel ltd
Closing Prices
July,2006
4 1409.15
5 1448.15
6 1467.15
7 1500.25
10 1523.25
Wk. ago 1421.65
Month ago 1311.45
Face Value Rs.5
P/E ratio 8.2
Market capitalization Rs.4692 cr.