Political Risk, Economic Risk, and Financial Risk

Author(s): Claude B. Erb, Campbell R. Harvey and Tadas E. Viskanta

Source: Financial Analysts Journal , Nov. - Dec., 1996, Vol. 52, No. 6 (Nov. - Dec., 1996),
pp. 29-46
Published by: Taylor & Francis, Ltd.

Stable URL: [Link]

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@[Link].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
[Link]

Taylor & Francis, Ltd. is collaborating with JSTOR to digitize, preserve and extend access to
Financial Analysts Journal

This content downloaded from

[Link] on Fri, 10 Nov 2023 10:11:25 +00:00
All use subject to [Link]
 Political Risk, Economic Risk, and Financial Risk
Claude B. Erb, Campbell R. Harvey, and Tadas E. Viskanta

Given the increasingly global nature of investment portfolios, an understanding of

country risk is very important. This article addresses the economic content of five
different measures of country risk:four measuresfrom the International Country
Risk Guide's political-, financial-, economic-, and composite-risk indexes and one
from Institutional Investor's country credit ratings. We explored whether any of
these measures contain information about future expected stock returns. We
conducted time-series/cross-sectional analysis linking these risk meastures toffuture
expected returns. Finally, we analyzed the links between fundamental attributes
such as book-to-price ratios within each economy and the risk measures. The results
suggest that the country-risk measures are correlated withfuture equity returns. In
addition, such measures are highly correlated with equity valuation measures. This
finding provides some insight into the reason that value-oriented strategies generate
high average returns.

W hat is country risk, and how should it affect that the financial-risk measure contains the most
global investment strategies? We explored information about future expected returns and that
five measures of country risk. Three-political risk, political risk contains the least.
economic risk, and financial risk-are from Political We next investigated the link between these
Risk Services' International Country Risk Guide country-risk measures and some more-standard
(ICRG). The ICRG also reports a measure of com- measures of risk. We investigated, for example,
posite risk, which is a simple function of the three whether a country's beta is correlated with the Mor-
base indexes. The fifth measure is Institutional Inves- gan Stanley Capital International (MSCI) World In-
tor's (II) country credit ratings (CCR). The informa- dex. Although this index is a standard risk measure
tion content of these indexes was examined in a for integrated capital markets, many researchers
number of ways. have found the world beta model inadequate to
We initially investigated whether the risk in- characterize risk in emerging markets. As an alter-
dexes contain information about future expected native, we also investigated the relation between
returns.1 This analysis was conducted in two ways. the country-risk measure and equity volatility.
First, we formed a portfolio of countries that expe- We then explored the interface between
rienced a decrease in risk rating (became more country-risk analysis and investment strategies
risky) and a portfolio of countries that experienced based on country fundamental information such
an increase in risk rating (became less risky). We as book-to-price ratios. We found that the risk
formed the portfolios after the risk information was indexes are highly correlated with the fundamen-
available and rebalanced them every six months. tal attributes. This finding provides some eco-
We found that these measures do, indeed, provide nomic insight as to why value-oriented strategies
information about expected equity returns. We earn higher returns than other strategies-they
supplemented this analysis with time-series/cross- reflect higher risk exposure.
sectional regressions that measure the amount of Relatively little research in finance has focused
information contained in each metric. We found on the economic content of various country ratings.
Political-risk measures were studied in Harlow
(1993) and Diamonte, Liew, and Stevens (1996).
Erb, Harvey, and Viskanta (1995) examined the
Claude B. Erb is a managing director at First Chicago information in credit-risk measures. In this study,
NBD Investment Management Company in Chicago. we investigated a broad cross-section of different
Campbell R. Harvey is a professor of finance at the
risk measures.
Fuqua School of Business, Duke University, and a
research associate at the National Bureau of Economic
Research in Cambridge, Massachusetts. Tadas E. MEASURING COUNTRY RISK
Viskanta is a vice president at First Chicago NBD
Many services measure country risk, including2
Investment Management Company in Chicago.
* Bank of America World Information Services,

Financial Analysts Journal * November/December 1996 29

This content downloaded from

[Link] on Fri, 10 Nov 2023 10:11:25 +00:00
All use subject to [Link]
* Business Environment Risk Intelligence (BERI) try groups. The ranking of factors affecting OECD
S.A., (Organization for Economic Cooperation and De-
* Control Risks Information Services (CRIS), velopment) country ratings appear to have been
* Economist Intelligence Unit (EIU), the most turbulent during the 15-year period from
* Euromoney, 1979 to 1994.

* Instituitional Investor,
* Standard & Poor's Rating Group (S&P), International Country Risk Guide
* Political Risk Services: International Country ICRG compiles monthly data on a variety of
Risk Guide, political-, financial-, and economic-risk factors to
* Political Risk Services: Coplin-O'Leary Rating calculate risk indexes in each of these categories, as
System, and well as a composite-risk index. Five financial, 13
* Moody's Investors Service. political, and 6 economic factors are used. Each
Each of the index or rating providers must factor is assigned a numerical rating within a spec-
amalgamate a range of qualitative and quantitative ified range. The specified allowable range for each
information into a single index or rating. This sec- factor reflects the weight attributed to that factor.
tion reviews in detail the methodologies used by A high score indicates low risk.
two of the foremost providers of risk ratings: Insti- Political-risk assessment scores are based on
tuitional Investor and International Country Risk subjective staff analysis of available information.
Guide. Economic-risk scores are based on objective analy-
sis of quantitative data, and financial-risk scores are
Institutional Investor based on analysis of a mix of quantitative and
Institutional Investor credit ratings are based on qualitative information.

a survey of leading international bankers, who are Calculation of the three individual indexes is
asked to rate each country on a scale from 0 to 100 simply a matter of summing up the point scores for
(100 represents maximum creditworthiness). II av- each factor within each risk category. The compos-
erages these ratings, providing greater weights to ite rating is a linear combination of the three indi-
respondents with greater worldwide exposure and vidual indexes' point scores. Note that political risk
more-sophisticated country analysis systems. (100 points) is given twice the weight of financial
Whenever a survey or expert panel is used to and economic risk (50 points each). ICRG, as well as
rate creditworthiness subjectively, the parameters many of the other providers, thinks of country risk
taken into account are difficult to define exactly. At as being composed of two primary components:
any given point in time, an expert's recommenda- ability to pay and willingness to pay. Political risk is
tion will be based upon factors the expert believes associated with willingness to pay, and financial
are relevant. and economic risk are associated with ability to pay.
To identify the factors that its survey partici- The specific formulas for these calculations are
pants have taken into consideration in the past, II as follows: PR = >PRi, ER = ZERi, FR = XFRI, and
asks them to rank the factors they consider in pre- CR = 0.5(PR + ER + FR), where PR is political risk,
paring country ratings. Table 1 presents the results ER is economic risk, FR is financial risk, and CR is
of this survey. Note that the bankers rank factors the composite-risk rating. The specific factors taken
differently for different groups of countries and into account for each risk index are detailed in
that rankings have changed over time within coun- Table 2.

Table 1. Rankings of Critical Risk Factors in Institutional Investor's Country

Credit Ratings by Rankings, 1979 and 1994
OECD Emerging Rest of World

Factor 1979 1994 1979 1994 1979 1994

Economic outlook 1 1 2 3 3 4
Debt service 5 2 1 1 1 1
Financial reserves/current
account 2 3 4 4 4 3
Fiscal policy 9 4 9 7 6 6
Political outlook 3 5 3 2 2 2
Access to capital markets 6 6 7 9 8 9
Trade balance 4 7 5 5 5 5
Inflow of portfolio
investment 7 8 8 8 7 8
Foreign direct investment 8 9 6 6 9 7

30 ?Association for Investment Management and Research

This content downloaded from

[Link] on Fri, 10 Nov 2023 10:11:25 +00:00
All use subject to [Link]
Table 2. Critical Factors in the ICRG Rating Table 3. ICRG Risk Categories
System Risk Category Composite Score Range

Percentage Percentage Very high 0.0-49.5

of Indivi- of High 50.0-59.5
Factor Points dual Index Composite Moderate 60.0-69.5
Low 70.0-84.5
Political
Very low 85.0-100.0
Economic expectations
versus reality 12 12 6
have significant similarities. Most of the providers
Economic planning
failures 12 12 6 transform widely used quantitative economic indi-
Political leadership 12 12 6 cators in roughly the same manner. The important
External conflict 10 10 5
differences are found in the degree of and specific
Corruption in government 6 6 3
factors included in the qualitative component of the
Military in politics 6 6 3
Organized religion in risk index measures.
politics 6 6 3 Table 4 provides a comparison of S&P and
Law and order tradition 6 6 3
Moody's ratings, by country, with the II and ICRG
Racial and national tensions 6 6 3
ratings as of October 1995. The S&P and Moody's
Political terrorism 6 6 3
ratings have a close correspondence with the II
Civil war 6 6 3
credit-risk measure (rank order correlation of 95
Political party development 6 6 3
percent) and with the ICRG financial rating (rank
Quality of the bureaucracy 6 6 3 order correlation of 90 percent). The correlations
Total political points 100 100 50 are weaker for the other measures. For example, the
Financial rank order correlation of the Moody's rating and
Loan default or the ICRG economic rating is only 68 percent.
unfavorable loan
restructuring 10 20 5
Delayed payment of SUMMARY DATA ANALYSIS
suppliers' credits 10 20 5
This section explores the time-series and cross-
Repudiation of contracts
sectional patterns in the country-risk measures.
by governments 10 20 5
Then, the correlations between risk measures are
Losses from exchange
controls 10 20 5 examined to assess whether each measure contains
Expropriation of private unique information. Finally, we provide some
investments 10 20 5
analysis of portfolio strategies that use the informa-
Total financial points 50 100 25
tion in the country-risk ratings. We track the
Economic performance of a portfolio that includes countries
Inflation 10 20 5 that have been downgraded recently, as well as a
Debt service as a portfolio containing upgraded countries.
percentage of exports
of goods and services 10 20 5
International liquidity Variation in Risk Measures
ratios 5 10 3
Foreign trade collection
Our analysis focused on 117 countries for which
experience 5 10 3 we have all five risk indexes.3 We segmented the
Current account balance countries into five groups: all countries, countries
as a percentage of goods
and services 15 30 8
with equity markets, developed countries with eq-
Parallel foreign exchange uity markets, emerging countries with equity mar-
rate market indicators 5 10 3 kets, and countries without equity markets.
Total economic points 50 100 25
Figure 1 presents time-series graphs of the
Overall points 200 100
equal-weighted risk indexes for three groups of
ICRG also groups country composite scores country types for the January 1984-July 1995
into ordinal risk categories to facilitate quick inter- period: developed countries with equity markets,
pretation and comparison of country scores. This emerging countries with equity markets, and all
categorization scheme is presented in Table 3. other countries.
The equal-weighted measures for the devel-
Index and Rating Provider Comparison oped countries (top panel) exhibit remarkably little
Although the factors taken into account by variation through time. The ICRG financial and
each provider of ratings and the audience it seeks economic measures remain about the same through-
to inform vary, the methods these providers use out the sample, as does the II country credit rating.

Financial Analysts Journal * November/December 1996 31

This content downloaded from

[Link] on Fri, 10 Nov 2023 10:11:25 +00:00
All use subject to [Link]
 Table 4. Comparison of Sovereign Country Ratings and Other Risk Attributes, October 1995

Country S&P Moody's II CCR ICRGC ICRGP ICRGF ICRGE

Argentina BB- BI 38.8 70.0 74.0 34.0 31.5

Australia AA Aa2 71.2 82.5 83.0 44.0 37.5

Austria AAA Aaa 86.2 84.0 81.0 47.0 39.5

Belgium AA+ Aal 79.2 83.0 79.0 46.0 41.0

Brazil B+ B1 34.9 62.5 64.0 33.0 28.0
Canada AA+ Aa2 80.3 83.0 81.0 46.0 39.0
Chile A- Baal 57.4 79.5 74.0 43.0 42.0
Colombia BBB- Baa3 46.5 68.0 60.0 40.0 35.5
Czech Republic A Baal 58.4 82.0 80.0 42.0 41.5
Denmark AA+ Aal 79.9 87.5 84.0 48.0 42.5

Finland AA- Aa2 71.4 84.5 87.0 43.0 39.0

France AAA Aaa 89.1 82.0 80.0 44.0 40.0

Germany AAA Aaa 90.9 84.5 83.0 47.0 39.0

Greece BBB- Baa3 50.0 75.0 75.0 38.0 36.5
Hong Kong A A3 67.0 81.0 72.0 46.0 43.5
Hungary BB+ Bal 45.0 72.5 78.0 39.0 28.0
India BB+ Baa3 46.1 69.0 63.0 37.0 37.5

Indonesia BBB Baa3 52.4 69.5 63.0 39.0 37.0

Ireland AA Aa2 73.4 84.0 85.0 44.0 38.5
Italy AA Al 72.3 77.0 75.0 41.0 38.0
Japan AAA Aaa 91.6 86.0 80.0 48.0 44.0
Malaysia A+ Al 69.1 80.5 76.0 43.0 42.0
Mexico BB Ba2 41.8 66.0 65.0 37.0 30.0

The Netherlands AAA Aaa 89.3 86.0 84.0 47.0 40.5

New Zealand AA Aa2 69.4 83.5 84.0 46.0 36.5
Nigeria NR NR 15.8 52.5 52.0 26.0 26.5
Norway AAA Aal 81.6 87.0 83.0 46.0 44.5

Pakistan B+ Bi 30.7 59.5 54.0 33.0 31.5

Peru NR NR 25.8 60.0 56.0 31.0 33.0

The Philippines BB Ba2 36.8 67.5 62.0 37.0 35.5

Poland BB Baa3 37.6 78.0 79.0 40.0 37.0
Portugal AA- Al 68.4 80.0 75.0 43.0 41.5
Singapore AAA Aa2 84.0 86.0 80.0 48.0 44.0
South Africa BB+ Baa3 45.2 76.5 75.0 41.0 36.5
South Korea AA- Al 72.2 82.0 77.0 46.0 41.0
Spain AA Aa2 73.7 74.0 69.0 41.0 38.0
Sweden AA+ Aa3 74.1 82.0 81.0 43.0 39.5
Switzerland AAA Aaa 92.2 89.0 85.0 50.0 43.0
Taiwan AA+ Aa3 79.9 84.5 77.0 48.0 44.0

Thailand A A2 63.8 77.0 69.0 43.0 41.5

Turkey B+ Ba3 40.9 62.5 59.0 36.0 30.0
United Kingdom AAA Aaa 87.8 79.5 78.0 46.0 35.0
United States AAA Aaa 90.7 83.0 80.0 48.0 38.0
Venezuela B+ Ba2 31.4 66.5 65.0 34.0 34.0
Zimbabwe NR NR 31.0 64.5 66.0 31.0 31.5
S&P rank correlation 95.2% 87.6% 77.0% 90.2% 72.4%
Moody's rank correlation 95.1% 87.5% 79.5% 89.8% 67.6%

NR = not rated.

Key: II CCR = Institutional Investor Country Credit Ratings

ICRGC = International Country Risk Guide Composite Index
ICRGP International Country Risk Guide Political Index
ICRGF = International Country Risk Guide Financial Index
ICRGE = International Country Risk Guide Economic Index

32 ?Association for Investment Management and Research

This content downloaded from

[Link] on Fri, 10 Nov 2023 10:11:25 +00:00
All use subject to [Link]
 Figure 1. Analysis of Equal-Weighted Average The cross-sectional behavior of the risk mea-
Risk Ratings, January 1984-July 1995 sures is demonstrated in Figures 2 to 6. For the
countries with equity returns and for all countries
Developed Countries
(with and without equity markets), the figures show
100
the January 1984 risk level against the change in the

80 .. risk level up to July 1995. The figures analyze all five

measures: II country credit risk (Figure 2) and the
5 60 - ICRG composite (Figure 3), political (Figure 4), fi-
nancial (Figure 5), and economic (Figure 6) risk. The
> 40 ______________________ cross-sectional mean reversion in the political-risk
measure that Diamonte, Liew, and Stevens (1996)
20 - documented also occurs in the other risk measures.
20
Those countries that began with a very low risk
rating have tended to improve; the countries with a
high rating have tended to deteriorate.
Emerging Countries Mean reversion is particularly evident for the
100
financial- and composite-risk measures for coun-

80 -
Figure 2. Mean Reversion in Risk Levels: 11
Country Credit Rating, January 1984-
60 -
July 1995

- 40
Countries with Equity Markets
30
20
20 - ~.

X 20 -
20~~~~
C) S

- 10 *
All Other Countries
100
Z~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~. .. ....
% o . s .. o .. : A r .... ... i ............... ............ 9 .............. ... , n *. ...............................

80-

-10_..
60 -

-20 0
> 40 -

20 - -30llll

0 L _
84 85 86 87 88 89 90 91 92 93 94 95
All Countries
30
Year

20 -
CCR ...... ICRG Composite ICRG Political
P4
-300~~~~ %
--- ICRG Economic - - - ICRG Financial
-20 -

-30*

The ICRG political rating shows a small decline.

The analysis for the emerging countries and cu -40 l l
(Jnur 1 .084)
all other countries (the center and lower panels)
is similar. Many of the other countries have
similar economic characteristics to the emerging 0 20 40 60 80 loo
country sample. Generally, all of the risk ratings
increase over the sample period, particularly from
1988 to 1993.

Financial Analysts Journal * November/December 1996 33

This content downloaded from

[Link] on Fri, 10 Nov 2023 10:11:25 +00:00
All use subject to [Link]
Figure 3. Mean Reversion in Risk Levels: ICRG Figure 4. Mean Reversion in Risk Levels: ICRG
Composite Rating, January 1984- Political Rating, January 1984-
July 1995 July 1995

Countries with Equity Markets

Countries with Equity Markets
40
40

30 -
30 - %

20 -

20 -
X 10 _ *
0~~~~
-100

0 ~ ~ ~ 0
-10

-10~ ~ ~ ~ ~~~~*. -201

All Countries All Countries

40 50

30 _ <1 40 -

10 -~~~~ -
X .0*
-20 I 30 - .
30
A * ~0. *0.
-10~~~ * 20 -

- 20 10- * **. #
0~~~~0
U-1?L0 =-0*0

-30 -20
20 40 60 80 100 20 40 60 80 100
Risk Level Risk Level
(January 1984) (January 1984)

tries with equity markets. Also, political and eco- the different countries. The upper triangle of the
nomic risks exhibit the same type of behavior. The matrix reports the correlation based on changes in
weakest evidence of cross-sectional mean reversion rating, and the lower triangle reports the correla-
is found for credit risk. tion of the levels.

A number of countries show sharp improve- The correlations are not as high as one might
ment in their risk ratings. For example, Argentina expect. Obviously, the correlation between the
began with a ICRG composite rating of 40.0 in composite and the political ratings is the highest
January 1984 and rose to 71.5 by July 1995. Similar because, by definition, the political rating is 50
improvements are evident for many of the other percent of the composite. The highest cross-
Latin American countries. In contrast, some devel- correlation of the levels of the three ICRG compo-
oped markets have shown deterioration. For exam- nents, however, is 35 percent. Interestingly, the
ple, Switzerland dropped from an ICRG composite correlation between the ICRG financial measure
beanalal wit a seiCRGua compsite rtheg cofr40.0tions
rating of 95.0 in January 1984 to 88.5 in July 1995. and the II country credit rating is only 26 percent.
Similar declines
iprovemtent were found
are baevdeont for Japan,
formianyua the United
ofbtervothers Table 6 documents the degree of predictability
Kingdom,
LTin and the
Amerclatin Unitedin
coaunries States.
contas-egt,oe devel-s in the risk measures. The levels of the measures are
Table 5 details the correlations of the various very persistent. The table shows the average auto-
risk measures. Because the credit ratingl is onla correlations of the changes in the risk measures. The
change in the II country credit rating is the most
predictable. The average first-order autocorrelation
is 24 percent (20 percent in developed markets and

34 ?Association for Investment Management and Research

This content downloaded from

[Link] on Fri, 10 Nov 2023 10:11:25 +00:00
All use subject to [Link]
 Figure 1. Trends in the Average Variance of Figure 2. Trends in the Variability of the Global
Individual Equities Market, Countries, and Global Industries
(excludes October 1987) Factors, Worldwide Sample
(excludes October 1987)
Worldwide Model
1.6 Worldwide Model-Global Market Factor
.i 10
1.2 .
. 8

0.8 *v- J *~ . , 6

"g 4
o 0.4

8 2 _
0 l l l l l
12/82 12/84 12/86 12/88 12/90 12/92 12/94
12/82 12/84 12/86 12/88 12/90 12/92 12/94
Coefficient t-Statistic

Intercept 1.0411 38.0938 Coefficient t-Statistic

Slope -0.0008 -2.4858 Intercept 0.6694 7.2679

Slope 0.0005 0.4496

EU-Only Model
Worldwide Model-Country Factors
1.6
. 10

~ 1.2 See 0 .# 8

g 2
0.8 .. 0 s** 00?0
?~~~~~ ~~~ 2ttDo'; *1 s g
o 0.4
S~~~~~'o Ot S I I I I

12/82 12/84 12/86 12/88 12/90 12/92 12/94 12/82 12/84 12/86 12/88 12/90 12/92 12/94

Coefficient t-Statistic
Coefficient t-Statistic
Intercept 1.0849 33.3070
Intercept 4.3338 32.3642
Slope -0.0014 -3.7505
Slope -0.0025 -1.6013

importance of global influences in the worldwide Worldwide Model-Industry Factors

model is weak at best. b 10
Figure 3 repeats the same tests for the EU-only
sample. Here, the results are clearer and statistically .g 8 *
significant: The EU-only global market factor in-
creases in importance through time, the country fac-
tors lose importance, and the global industry factors 4
increase in importance. All of these effects point
2> 2-_
toward increasing market integration within the EU.
Another measure of the trend in European So 0 I I I
integration is based on the EU country factor re-
12/82 12/84 12/86 12/88 12/90 12/92 12/94
turns from the worldwide model. We estimated
the correlations of the EU country factor returns Coefficient t-Statistic
over four subperiods of 37, 36, 36, and 36 months.
Intercept 5.8305 45.8838
Under the null hypothesis of no increased integra- Slope 0.0023 1.5340
tion, the true correlation matrix should be the same
in the subperiods. Under the alternative that the

Financial Analysts Joumal: March/April 1996 35

This content downloaded from

[Link] on Fri, 10 Nov 2023 10:11:25 +00:00
All use subject to [Link]
 Table 5. Correlations of Risk Measure Levels and Changes, Semiannual
Observations, January 1984-July 1995

Source II CCR ICRGC ICRGP ICRGF ICRGE

II CCR -0.03 0.01 0.03 -0.09

ICRGC 0.35 0.79 0.54 0.43
ICRGP 0.30 0.83 0.25 0.06
ICRGF 0.26 0.60 0.35 0.05
ICRGE 0.10 0.52 0.24 0.25

Table 6. Persistence of Risk Measures, First-Order Autocorrelations of Log

Rating Changes, Semiannual Observations, January 1984-July 1995

All Countries

Developed Country Emerging Country

Source Average Minimum Maximum Average Average
II CCR 0.24 -0.31 0.77 0.20 0.26
ICRGC 0.04 -0.92 0.93 -0.01 0.07

ICRGP -0.01 -0.84 0.51 0.03 -0.04

ICRGF 0.10 -0.60 0.75 0.08 0.11

ICRGE -0.18 -0.72 0.67 -0.17 -0.19

We also assessed the impact of the change in That is, their portfolios are only investable if you
the II country credit rating on the next month's know in advance what next month's rating will be.
change in the ICRG rating. There is little informa- An alternative approach is an event strategy,
tion here. Only one of the regressions, that on finan- which was pursued by Harlow (1993). He exam-
cial risk, has a coefficient that is significantly ined the cumulative returns of countries up to six
different from zero. The coefficients are all negative, months after political rating increases or decreases.
which makes little sense (increased II CCR predicts This strategy is investable, given that countries are
lower ICRG ratings). We concluded that the ICRG purchased or sold after the rating changes.
contains information that predicts II CCR but the Our strategy was to form two portfolios: up-
reverse is not true. grades and downgrades. The portfolios are rebal-
anced every six months. If the rating does not
An Initial Portfolio Strategy
change, the country stays in its respective portfolio.
Table 7 suggests a relation between average
This strategy reduces transaction costs and increas-
returns and average ratings. One way to analyze
es the number of countries in the portfolios.
this relation is to form portfolios based on rating
The top panel of Table 9 complements the
changes. One version of this approach is analyzed
Diamonte, Liew, and Stevens results. The upgrade
in Diamonte, Liew, and Stevens (1996). They
portfolios have higher average returns than the
formed two portfolios: upgrade and downgrade,
downgrade portfolios. This result is true not only
based on the ICRG political-risk measure. Impor-
for political risk but also for the other risk measures.
tantly, their approach is ex post rather than ex ante.
Indeed, of the four ICRG risk measures, political
risk is never the most important one. In the all-
Table 7. Sample-Period Correlation between
country group and the emerging markets, the
Average Risk Measures and Price
composite risk is more important than the others in
Moments
the sense that it implies more profit. In the
Country Sample II CCR ICRGC ICRGP ICRGF ICRGE developed countries, the financial-risk measure
All countries has the most ability to discriminate between high-
Geometric return -0.23 -0.15 -0.13 -0.16 -0.16 and low-return portfolios.
Volatility -0.52 -0.45 -0.31 -0.49 -0.59 The lower panel of Table 9 presents the results
Beta - MSCI World 0.24 0.43 0.44 0.40 0.30 from an investable strategy. Portfolios are based on
Developed countries a previous rating change and held for six months.
Geometric return 0.18 -0.15 -0.28 -0.08 0.21 Each country is equal weighted in the portfolios.
Volatility -0.46 -0.41 -0.38 -0.47 -0.15 The results suggest that the composite-risk
Beta - MSCI World 0.09 -0.15 -0.24 -0.04 0.06 measure has considerable power to identify high-
Emerging countries and low-return portfolios. On a risk-adjusted basis,
Geometric return -0.26 -0.06 -0.02 -0.08 -0.12 the hedge strategy (buy upgrades and sell
Volatility -0.16 -0.08 0.20 -0.16 -0.45 downgrades) based on composite risk has an alpha
Beta - MSCI World 0.03 0.42 0.46 0.35 0.20 of 1,140 basis points a year when all countries are

36 ?Association for Investment Management and Research

This content downloaded from

[Link] on Fri, 10 Nov 2023 10:11:25 +00:00
All use subject to [Link]
 Figure 7. Equity-Risk and Country-Risk Ratings by Exposure to World Market, January 1984-
July 1995

Institutional Investor CCR ICRG Composite Rating

1.5 *1.5

1.0 - 1.0 -

0.5 - 0~~~~~~S.5 -0
u u~~~~~~~

0 C

-0.5. -0.5 I
0 20 40 60 80 100 0 20 40 60 80 100
Average Institutional Investor CCR Average JCRGC

ICRG Political Rating ICRG Financial Rating

1.5 *1.50

1.0 - %~~~~~~~* * * A.0 * 00 0.

-,0.5 - * ~~0.5-

0~~~~~~~~~

-0.5. 1 -0.5 II
0 20 40 60 80 100 0 10 20 30 40 50
Average ICRGP Average ICRGF

ICRG Economic Rating

1.50

C 0~~~~~~~~~~~~~~~~~~6
40 0
Q),0.5-

0~~~~~~~~~

-0.5 II
0 10 20 30 40 50
Average ICRGE

examined. The political-risk measure has an alpha is executed for developed countries, the alpha for
of -160 basis points a year. The portfolios formed the composite risk is 840 basis points. For economic
on economic risk have an alpha of 750 basis points. risk, the alpha is 540 basis points. Similar to the
The portfolios based on financial and credit risk overall sample, the financial, political, and credit
fare worse than those based on political risk. measures fare poorly.
Perhaps the most interesting aspect of the When the emerging markets are examined, the
analysis is that this phenomenon does not apply alpha on the composite risk measure is 860 basis
only to emerging markets. If the portfolio strategy points. For economic risk, the alpha is 880 basis

Financial Analysts Journal * November/December 1996 37

This content downloaded from

[Link] on Fri, 10 Nov 2023 10:11:25 +00:00
All use subject to [Link]
Figure 8. Equity-Risk and Country-Risk Ratings by Annualized Standard Deviation, January 1984-
July 1995

Institiutional Investor CCR ICRG Composite Rating

120 120

- ? 100 * 1 100

Average Institutional Investor CCR AverageICRGCO

60 - UZ 60-
> >~~~~~~~~~~~~~~~~N*
40 - 40 -
0 0 4 080 10- 0 0 6 80 -0
120 -1200

0 0
0 20 40 60 80
100 0 20 40 60 80 100
Average Institutional Investor CCR Average ICRGC

ICRG Political Rating ICRG Financial Rating

120 120

- 100 _-100 0

-80 _80

60 - 60

40
20
10
-
2 40 24
IC20R*E

Average ICRGP ~~~~~~~~Average ICRGF

> >

ICRG Economic Rating

120 a lO la

-; 100

~80-

> 60
40~~~~~~~~~~
20 *-: .*
gO 0

0 1 1
0 10 20 30 40 50

Average ICRGE

points. Consistent with the developed market ilar results. For example, the risk-adjusted return of
analysis, financial, political, and credit risk are the composite-risk, upgrade-downgrade portfolio
unable to distinguish between high and low is 1,170 basis points a year for the capitalization-
returns in a portfolio strategy. weighting scheme, compared with 1,140 basis
The performance of these strategies is robust to points for equal weighting (these results are
different country-weighting schemes. As an alter- available on request). Within the country group-
native to the equal weighting of countries, we used ings, capitalization weighting makes the economic-
a capitalization-weighting strategy and found sim- risk measure more important in developed markets

38 ?Association for Investment Management and Research

This content downloaded from

[Link] on Fri, 10 Nov 2023 10:11:25 +00:00
All use subject to [Link]
Table 8. Predicting Changes in Risk Attributes rating and decreased risk rating. Much information

Source Coefficient t-Statistic R2 is potentially lost with such a coarse aggregation.

ICRG attributes as predictors of II CCR More information can be obtained by trying to

ICRGC 0.2120 7.59 5.0% predict both the cross-section and the time-series of
ICRGP 0.1244 5.67 2.8 expected returns based on the risk attributes.
ICRGF 0.0956 5.69 2.8 Table 10 presents attribute regressions in the
ICRGE 0.0833 4.65 1.9
following form:
II CCR as predictors of ICRG attributes
ICRGC -0.0115 -1.13 0.1% Rt = Co + ClAt- + EtA
ICRGP -0.0087 -0.62 0.0
ICRGF -0.0585 -4.09 1.4
where R represents a vector of six-month returns

ICRGE -0.0198 -0.99 -0.0

from July 1984 to June 1995 (some markets' returns
begin later) for all of the countries in our sample. A
Notes: In the top panel, the semiannual log change in the II
ratings is regressed on the lagged semiannual log changes of the represents the risk attribute that is lagged and
ICRG ratings. The sample is from March 1984 through October matched to the country. The full sample of all coun-
1995 for countries with equity markets. In the lower panel, the
tries has 884 observations: 441 for the developed
monthly log change in the ICRG ratings is regressed on the
lagged semiannual log change in the II CCR. The sample is from countries and 443 for the emerging equity market
February 1984 through September 1995 for countries with sample.
equity markets.
Table 10 reports coefficients and t-statistics for
five separate regressions of the returns on the
than in other markets. Capitalization weights make
attributes. The R2 from this regression and the R2
the political-risk measure more important in
from a similar regression with indicator variables
emerging markets.
for each country is also reported (fixed-effect
Some caution should be exercised in interpret-
regression, or FER2). Regressions are estimated on
ing our results, because transaction costs have not
the lagged level of the attribute, as well as lagged
been taken into account. We have taken two steps,
changes in the attribute. The number of countries
however, to minimize those costs. First, turnover is
in each cross-sectional regression grew from 28 in
minimized by keeping the countries with zero rat- March 1984 to 48 in March 1995.
ing change in their current portfolios. Second, we
The top panel of Table 10 reports the regres-
rebalanced only twice a year. Our portfolio strategy
sions using the lagged level of the attribute as an
is successful for developed as well as emerging
explanatory variable for the cross-section of expect-
markets. In many of the developed markets, trad- ed returns. With the full sample of 48 countries,
ing costs can be minimized by using index futures. each of the five risk attributes has a coefficient sig-
nificantly different from zero when the attributes
CROSS-SECTIONAL ANALYSIS OF are examined separately. In each case, the coeffi-
RISK cient is negative, implying that lower rating (higher

This section formalizes our investigation of the risk) is associated with higher expected returns. In

relation between expected returns and country-risk this analysis, the financial-risk variable is most im-
portant and the composite is second most impor-
measures using time-series/cross-sectional regres-
tant. When a multivariate regression is estimated
sion analysis. An examination of the link between
on four attributes, only the financial-risk variable
fundamental valuation measures and the country
enters the regression with a significant coefficient.
ratings indicates that the country ratings can ex-
Regressions were estimated separately on
plain a considerable amount of the variation in
developed and emerging countries, and the results
book-to-price measures. Finally, we address the
show that the developed countries sample is
problem that country attributes have different in-
driving the explanatory power of the entire sample.
terpretations in different countries.
The ICRG financial-risk variable enters the devel-
oped country regression with a t-ratio close to 4 in
Time-Series/Cross-Sectional Analysis
the univariate regression. In the univariate and
of Risk Attributes multivariate regressions, the ICRG financial- and
The analysis in Table 9 suggests that portfolio economic-risk variables both enter the regression
groupings by certain attributes produce positive but the political-risk and II CCR variables do not.
risk-adjusted profits on an ex ante basis. This simple In the emerging countries sample, no attribute
analysis was based on two portfolios: increased risk enters the regression with a significant coefficient;

Financial Analysts Journal * November/December 1996 39

This content downloaded from

[Link] on Fri, 10 Nov 2023 10:11:25 +00:00
All use subject to [Link]
Table 9. Changing Country-Risk Portfolio Strategy, January 1984-July 1995

Downgrade Upgrade Upgrade-Downgrade

MSCI MSCI MSCI

Portfolio Standard World Portfolio Standard World Portfolio Standard World
Sample Return Deviation Beta Return Deviation Beta Return Deviation Alpha

Contemporaneous rating changes and portfolio performance

All countries
II CCR 21.6% 20.6% 0.95 30.6% 24.4% 0.95 9.0% 14.4% 9.0%
ICRGC 16.7 24.3 1.07 33.4 21.9 0.97 16.7 14.5 16.6
ICRGP 19.3 21.5 0.95 34.3 24.3 1.13 15.0 16.1 11.2
ICRGF 18.1 30.2 1.21 28.5 22.6 0.99 10.4 23.0 11.4
ICRGE 25.5 26.6 1.30 26.6 18.9 0.78 1.1 10.9 8.0
Developed countries

II CCR 17.2% 19.8% 0.98 19.5% 20.3% 1.09 2.3% 13.5% 1.0%
ICRGC 15.0 19.7 1.05 21.3 21.2 1.20 6.3 9.8 3.7
ICRGP 15.9 17.7 0.98 24.4 27.6 1.52 8.5 16.6 1.8
ICRGF 14.5 21.1 1.12 22.1 20.2 1.12 7.6 10.0 6.9
ICRGE 16.1 17.8 0.95 19.2 18.4 1.02 3.1 8.4 1.1
Emerging markets

II CCR 22.2% 25.1% 0.87 36.7% 33.3% 0.83 14.4% 24.5% 15.3/
ICRGC 16.0 35.0 1.12 41.1 26.3 0.84 25.0 29.7 24.9
ICRGP 20.0 31.6 0.82 39.1 28.2 1.07 19.1 27.4 12.6
ICRGF 18.6 36.8 1.28 35.8 26.8 0.76 17.2 33.5 21.1
ICRGE 31.5 35.1 1.45 32.7 24.9 0.51 1.2 22.6 14.1
MSCI World 14.9% 15.5%

Predictive rating changes and portfolio performance

All countries
II CCR 29.1% 21.5% 0.76 28.6% 24.7% 1.04 -0.5% 20.2% -4.8%
ICRGC 21.3 22.3 1.02 30.8 22.0 0.85 9.5 16.6 11.4
ICRGP 26.1 18.9 0.85 26.8 21.9 1.00 0.7 10.7 -1.6
ICRGF 26.6 30.8 0.86 26.4 21.6 1.02 -0.2 24.5 -4.9
ICRGE 23.4 25.9 1.04 30.7 21.3 0.94 7.4 17.9 7.5
Developed countries
II CCR 19.9% 20.6% 1.03 20.6% 20.5% 1.18 0.7% 14.2% -1.9%
ICRGC 17.1 20.0 1.19 21.7 18.7 0.94 4.6 10.8 8.4
ICRGP 18.8 18.6 1.07 20.3 21.5 1.25 1.5 10.4 -1.2
ICRGF 21.8 20.9 1.09 19.9 19.3 1.12 -1.9 9.9 -2.5
ICRGE 14.8 18.5 1.04 19.9 17.6 0.99 5.1 9.0 5.4
Emerging markets
II CCR 35.4% 26.9% 0.61% 29.9% 30.8% 0.81 -5.5% 29.9% -7.7%
ICRGC 22.8 31.1 0.69 36.4 28.3 0.84 13.6 29.6 8.6
ICRGP 28.3 27.4 0.57 28.2 25.8 0.75 0.0 23.4 -3.9
ICRGF 27.2 37.1 0.76 33.8 26.6 0.89 6.6 33.5 0.8
ICRGE 27.2 37.3 0.90 38.0 28.8 0.83 10.8 36.3 8.8
MSCI World 17.5% 15.0%

Notes: The upper panel portfolios are formed by equal weighting the returns for all countries that during the current period experienced
an upgrade or downgrade in their risk exposures. The lower panel portfolios are formed by equal weighting the returns for all countries
that during the previous period experienced an upgrade or downgrade in their risk exposures. Portfolios were reformed every six
months. MSCI World alphas are annualized. MSCI World returns in the two panels differ because the starting date in the upper panel
is January 1984 and in the lower panel is July 1984.

the closest are the ICRG financial- and economic- 10 have many similarities. In particular, in the de-
risk variables. veloped country sample, the ICRG financial vari-
The lower panel of Table 10 presents the re- able is by far the most important, with t-ratios close
gression coefficients of country returns on the to 2. In contrast to the upper panel, the political-risk
lagged change in the attributes. These results are variable is important-but only for the emerging
closer to the trading strategy results in Table 9, in market sample.
which the portfolios were constructed of countries Overall, the regression evidence complements
with changes in rating. Although the change re- the evidence from the portfolio strategies. The re-
gressions have less explanatory power than the gression results, however, do not seem as dramatic
lagged levels, the upper and lower panels in Table as profits from the portfolio strategies. This result

40 ?Association for Investment Management and Research

This content downloaded from

[Link] on Fri, 10 Nov 2023 10:11:25 +00:00
All use subject to [Link]
Table 10. Cross-SectionalTime-Series Importance of Country-Risk Attributes

Multivariate
Sample Regression II CCR R2 ICRGC R2 ICRGP R2 ICRGF R2 ICRGE R2 R2
Lagged level regressions
All Univariate -0.0011** 0.5% -0.0023*** 0.7% -0.0017** 0.4% -0.0043*** 1.2?% -0.0041** 0.4%
All Multivariate 0.0003 0.0016 -0.0082*** 0.0024 1.1%
Developed Univariate -0.0007 -0.1 -0.0019 0.0 -0.0007 -0.2 -0.0107*** 3.2 0.0071** 0.8
Developed Multivariate 0.0027* 0.0000 -0.0184*** 0.0109*** 6.3
Emerging Univariate -0.0009 -0.2 -0.0023 0.1 -0.0011 -0.2 -0.0035 0.2 -0.0050 0.2
Emerging Multivariate 0.0015 0.0011 -0.0046 -0.0042 -0.4

Lagged change in level regressions

All Univariate -0.0072 -0.1 -0.0046 0.0 -0.0076* 0.3 -0.0015 -0.1 0.0071 0.0
All Multivariate -0.0037 -0.0096** 0.0050 0.0092 0.2
Developed Univariate 0.0072 -0.1 -0.0014 -0.2 -0.0012 -0.2 -0.0181* 0.6 0.0051 -0.1
Developed Multivariate 0.0097 -0.0032 -0.0188* 0.0049 0.2
Emerging Univariate -0.0147 -0.0 -0.0089 0.1 -0.0135** 0.9 -0.0008 -0.3 -0.0019 -0.2
Emerging Multivariate -0.0147 -0.0185** 0.0123 0.0097 0.8

Notes: The upper panel results from time-series/cross-sectional regressions of semiannual returns against the lagged risk attribute or,
in the multivariate case, risk attributes. The lower panel results from time-series/cross-sectional regressions of semiannual returns
against the lagged log change in the risk attribute or, in the multivariate case, risk attributes.

* 10 percent level of significance.

** 5 percent level of significance.
1 percent level of significance.

could be driven by the fact that the value of the percent of the variation in the full sample of coun-
change variable is often zero. tries. The economic-risk measure enters with a neg-
ative coefficient. This result suggests that high
Risk Attributes and Fundamental ratings (low risk) are associated with low book-to-
price ratios.
Variables
Ferson and Harvey (1996) provided evidence When the data are segmented by developed
that fundamental attributes such as book-to-price, and emerging countries, the results are similar to
earnings-to-price, dividend-to-price, and price-to- those for the pooled regression. For developed
cash ratios are linked to the risk exposure of nation- countries, 18 percent of the cross-sectional variance
al markets. They proposed an asset-pricing frame- of the book-to-price ratios can be explained, to
work in which the fundamental attributes are which the economic-risk variable contributes 10
linked to dynamic country-risk (beta) functions. percent. For the emerging equity markets, 29 per-
We focused on a set of three attributes available cent of the variation can be explained in the multi-
for both the developed and emerging markets: variate model and the economic-risk variable
book-price, earnings-price, and dividend-price contributes 24 percent.
ratios. Table 11 provides cross-sectional regressions The risk variables also show some ability to
that use our country-risk measures to attempt to explain the cross-section of dividend yields. For the
explain the cross-section of the valuation attributes. dividend yield, however, the results are different
As in Table 10, Table 11 presents both univariate across the developed and emerging markets. For
regressions and multivariate regressions for three the developed markets, a total of 16 percent of the
different samples. Our discussion focuses on the cross-sectional variation can be explained. The
multivariate regressions. The number of countries ICRG economic-risk measure contributes 13 per-
in the cross-sectional regression varies from 18 in cent, and the financial-risk variable contributes 5
March 1984 to 47 in March 1995. percent (the contributions need not sum to the total
Table 11 presents a number of interesting re- because of correlation between the risk measures).
sults. First, the risk measures have some ability to In emerging markets, a similar amount of explana-
explain all three valuation ratios. They do the best, tory power (17 percent) is found. In this case, how-
however, in explaining the cross-sectional varia- ever, almost all of the explanatory power is coming
tion in the book-to-price ratios. In the full sample, from financial risk, which contributes 16 percent.
more than 25 percent of the variation can be ex- In both the developed and emerging market regres-
plained using all four risk measures. Of all the sions, the two risk variables enter with negative
component risk measures, the ICRG economic-risk coefficients, suggesting that higher ratings (lower
rating is the most important, accounting for 18 risk) are associated with lower dividend yields.

Financial Analysts Journal * November/December 1996 41

This content downloaded from

[Link] on Fri, 10 Nov 2023 10:11:25 +00:00
All use subject to [Link]
 The incremental contribution of the risk rat- The third panel of Table 13 implements the
ings relative to the book-price valuation measure Ferson and Harvey (1996) attribute-adjustment
is presented in Table 12. Univariate regressions are methodology. Ferson and Harvey made beta risk
estimated in the form (with respect to the MSCI world portfolio) a func-
tion of the attribute. We estimated the following
Rt = Co + ClAt_l + C2B/Pt-1 + Et. time-series regression for each country:

Table 12 reports the coefficients and t-ratios by

Rit = bio + biIRIV,t + bi2(R , t xA, t-_) + Et
risk attribute. The results indicate that the II CCR
and the ICRG political-risk measures provide little With the results of this regression, an adjusted
or no incremental information. The ICRG financial- attribute was formed:
risk measure is the most important variable, fol-
lowed by ICRG economic risk. Although both the
i, t-l I bi2 +bi2A.n-
economic and financial measures add important
incremental explanatory power to the developed The third panel of Table 13 reports regressions
country regressions, they have no ability to add to of the cross-section of returns on the cross-section
the emerging equity market regressions. For the of adjusted attributes. The positive coefficient sug-
emerging markets, the book-price valuation at- gests a positive relation between beta risk and ex-
tribute fully characterizes the information in the pected returns. Although many of the coefficients
risk ratings. are not significantly different from zero at conven-
tional levels, the explanatory power of the regres-
Trading Strategies Based on Risk sions uniformly improve over the raw attribute

Attributes (top panel) regressions.4

The bottom panel in Table 13 considers both
The time-series/cross-sectional methodology
the adjusted and unadjusted risk attributes. Note
has two disadvantages. First, in stacking the time-
that collinearity is no problem here because the
series of returns together, important information
adjustment factors are country specific. Consistent
regarding the cross-sectional correlation of the re-
with the results in the panel above, the beta risk
turns is eliminated, which could cause the standard
factor enters each regression with a positive
errors to be understated. Second, the time-series/
coefficient. Each of the II CCR and ICRG risk
cross-sectional methodology imposes the same
measures enters with a negative coefficient.
slope coefficient for all time periods, even though
Notable in this table is the large jump in
it could change through time. Ferson and Harvey
(1991 and 1993) found that the variation in the slope
explanatory power. The average cross-sectional R
for the II CCR is now 26 percent (regressing on
coefficients is to some degree predictable.
credit risk alone produces an 8 percent R2). The
The top panel of Table 13 reports the results of
economic- and financial-risk measures both show
estimating a cross-sectional regression at each six-
similar explanatory power.
month interval. The slope coefficients are averaged,
and the standard error of the average is also pre-
sented. The results are largely consistent with those CONCLUSIONS
reported in Table 10 for the univariate regressions. The goal of this research was to explore the economic
For each of the risk measures, the average slope content of five country-risk measures: Institutional
coefficient is negative. In all cases, the slope coeffi- Investor's country credit rating and the International
cient is greater than 1.5 standard errors below zero. Country Risk Guide's political-, financial-, economic-,
Although not reported here, we investigated and composite-risk ratings. Our analysis suggests
the pattern of estimated cross-sectional slope coef- that the ICRG composite, financial, and economic
ficients through time along with the time-series of ratings, in particular, contain considerable informa-
R2s. In all cases, the explanatory power of the risk tion. For example, for portfolios based on changes in
variables increases through time. the risk ratings, risk-adjusted abnormal returns are
The second panel of Table 13 assesses the in- in the range of 1,000 basis points a year. Trading on
cremental contribution of the risk attributes when the basis of the political-risk measure alone has no
the book-to-price ratio is included in the cross- ability to produce abnormal returns.
sectional regression. Although the coefficients are The cross-sectional regressions confirm the re-
negative for each of the risk attributes, only II CCR sults of the portfolio analysis. Some of the ICRG
and the ICRG composite and political ratings have risk measures-in particular, economic and finan-
coefficients that are more than one standard error cial risk-can predict the cross-section of expected
below zero. returns, which is most strongly evidenced in the

42 ?Association for Investment Management and Research

This content downloaded from

[Link] on Fri, 10 Nov 2023 10:11:25 +00:00
All use subject to [Link]
 00 ~~~l0 ~ N CC C0 I N It)
Lri y~ ri 6 t6icli o6 t<cl C ~ o 6 .

6-0 8-0~~~~~~~~~~~~~~~~ .

oc "C N 6 cl~ cON o6 rl 0

Lr)0 D N 0C0 c 6c 5-f . .F N Z)' 0 N

r4 I!i) .3 ON '.0 - N 00 ".) b

' O l CN I C) Lf C) ) N O - --4) Cl N 0 LC' 14 N crl '.0 O

0-1 0N80C
14 00 re) N m ~~~~~ ~ ~~~~~~~CIO 0 0CCN Cl
- ~~~~~~~~~~~~~~~C- C.)4r ~ C

(3 r D 4F4C )z c C: A5 c o C ON)0N C- a) (3 CC ' 0 C-) a) 0

Cl '.0 ( Cl '.0 C Cl 0 C CT CD CDN CT CD Cl . Z

-8-11~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~-

J9 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~CZ

~~4 r-I r-4 C) C) N C) C)~~~~~~~~~O rN l

CO) 0-4~~~~~~~~~~~~~~"00

(3 00 rz -L o 0 ClCN0 N 0 CDC 00 N 00 O(

a) ~~~~~~~~~~~~~~~~~~~~~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 0.~~~~~~~~~~~~~~~) C

.0~~~~~~~~~~~~~~

Cl () a aNL)0 Cc) C

ua U
- U ~ * *a)
w~~~~~~~~~~~~00*Cd
.-)*'.-a)(3
- . -C C

(3 > Cl- Io O Ul-NCO0 -

0 m 0 m 0 (Z 0 Cl00 et.0 a)
0.) COO.~~~~~~~~~~~~~~~~~~~~~~~~~()
'-C CO 6~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

0 ~~~~~~~~~~~~~LLNC'O'.0 - - .~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~1 - .
CU (a) a) tt a) LL 2- Cl a)

0 1-" C.)~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~> "41L

Ur.. ;4.c u u .u . .-0

I 5C0~~~~~~~~~~~~~~~~~~~~~~~~C

0 ~ ~ ~ 4 C)o>C)
0.) Co 00
0.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~c 3 Z

Financial Analysts Journal - November/December 1996 43

This content downloaded from

[Link] on Fri, 10 Nov 2023 10:11:25 +00:00
All use subject to [Link]
 0-1 00

z \0O ? 0 )cZ
u z

.0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Z N ) 00 0 Cl

00

ONON lNL. -r*-

0 0

0 Q) e- W~~ NC1 1 -

U U~~~~~~~~~~~~~~~~C l

U)~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~C

-4) C) C \- 0 CLC)r -4 0 \OLfTN 0)0 It ON LfN \0C C14 ) A

4 Z Zo 20 2

a) 0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~Ua a

, C C ) ) O z ) C0C CD. CD0 : 4 D O O C

cn 4.1~~~~~~~~N0O N C ~ ~ 0~~cI
.2 ~~~~0 ~ ~ 0 ~ ~~~ ClN~~~0~~ cr~~~ ClNc~~~~~NLf~~~Z .~~~ C~~~ON ~ 7

0)
a)~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

tJ~~~~ ~ a) a)

> 0-4 ~ ~ ~ ~ ~ 0.~ 0-4

o -J ~-oCl-0C~~000 U ~ 0.U ~N4U UONU0(U Ul~l~N 0

Ua)
)04 U
O.O U0 i
04 .O 0- ~0
n uU U UU
I-O 0- V) un0
.9-U U
- -4 04 - - -

44 ?Association for Investment Management and Research

This content downloaded from

[Link] on Fri, 10 Nov 2023 10:11:25 +00:00
All use subject to [Link]
developed markets in our sample. Change in polit- oriented strategies work.
ical rating also has some marginal explanatory Our final contribution is to bridge attribute-
power in emerging equity markets but not in de- oriented investment strategies with asset pricing.
veloped markets. Ferson and Harvey (1996) argued that popular val-
The country-risk ratings are correlated with uation attributes should enter each country's dy-
fundamental valuation attributes. For example, 25 namic risk function. We followed their suggestion
percent of the cross-sectional variation in book-to- and found a relation between dynamic risk with
price ratios can be explained by the risk ratings. respect to a world benchmark and expected re-
This explanatory power is largely driven by the turns. In addition, similar to Ferson and Harvey's
ICRG economic-risk variable, which alone can ex- results, the attributes are still important for the
plain 18 percent of the cross-sectional variation. alpha; that is, even after allowing for the attributes
These results shed light on the information that to influence the beta risk, they still have marginal
determines the fundamental valuation measures. cross-sectional explanatory power when included
We have provided insights on why global value- in the prediction exercise.5

NOTES

1. To ensure the widest possible dissemination of our 3. Time-series graphs of the risk indexes for each country are
methodology, we have established a country-risk home page available through the country-risk Web site.
on the World Wide Web: http: / /[Link]/ -charvey/ 4. For example, the explanatory power of the book-to-market
Country_risk/[Link]. This site includes information variable doubles after the Ferson and Harvey (1996) attribute
on 135 different countries that could not be included in this adjustment is implemented. This result appears to provide
manuscript. For example, the site contains equity return considerable support for the Ferson-Harvey method.
histograms for 48 countries, time-series graphs of the five risk 5. We appreciate the help of Chris Rath at Political Risk Services
measures for 117 countries, and summary statistics. in supplying us with the data. Rob Feldman provided
2. An appendix, available on request, provides information on valuable research assistance. John Liew and Ross Stevens
and comparisons of these providers. provided valuable comments on an early draft.

REFERENCES

Bekaert, Geert. 1995. "Market Integration and Investment Fama, Eugene F., and James D. MacBeth. 1973. "Risk, Return and
Barriers in Emerging Equity Markets." World Bank Economic Equilibrium: Empirical Tests." Journal of Political Economy, vol.
Review, vol. 9, no. 1:75-107. 81, no. 3 (May/June):607-36.

Bekaert, Geert, and Campbell R. Harvey. 1995a. "The Cost of Ferson, Wayne E., and Campbell R. Harvey. 1991. "The
Capital in Emerging Markets." Working paper, Duke University Variation of Economic Risk Premiums." Journal of Political
and Stanford University. Economy, vol. 99, no. 2 (April):385-415.

. 1995b. "Emerging Capital Markets and Economic _ . 1993. "The Risk and Predictability of International
Development." Working paper, Duke University and Stanford Equity Returns." Review of Financial Studies, vol. 6, no. 3:527-66.
University.
_ . 1994. "An Exploratory Investigation of the Fundamental
. 1996. "Emerging Equity Market Volatility." Journal of Determinants of National Equity Market Returns." In The
Financial Economics, forthcoming.
Internationalization of Equity Markets, edited by Jeffrey Frankel.
Black, Fischer. 1972. "Capital Market Equilibrium with Chicago: University of Chicago Press.
Restricted Borrowing." Journal of Business, vol. 45, no. 3
_ . 1996. "Fundamental Determinants of National Equity
(July):444-55.
Market Returns: A Perspective on Country Risk and Asset
Diamonte, Robin, John M. Liew, and Ross L. Stevens. 1996. Pricing." Journal of Banking and Finance, forthcoming.
"Political Risk in Emerging and Developed Markets." Financial
Analysts Journal, vol. 52, no. 3 (May/June):71-76. Gibbons, Michael R. 1982. "Multivariate Tests of Financial
Models: A New Approach." Journal of Financial Economics, vol.
Erb, Claude B., Campbell R. Harvey, and Tadas E. Viskanta.
10, no. 1 (March):3-27.
1994. "National Risk in Global Fixed Income Allocation." Journal
of Fixed Income, vol. 4, no. 2 (September):17-26. Harlow, W. Van. 1993. Political Risk and Asset Markets. Boston:
Fidelity Investments Monograph.
____. 1995. "Country Risk and Global Equity Selection."
Journal of Portfolio Management, vol. 21, no. 2 (Winter):74-83. Harvey, Campbell R. 1991. "The World Price of Covariance

.1996. "Expected Returns and Volatility in 135 Countries." Risk." The Journal of Finance, vol. 46, no. 1 (March):111-57.
Journal of Portfolio Management, vol. 21, no. 3 (Spring):46-58. _ . 1993. "Portfolio Enhancement with Emerging Markets
Errunza, Vihang R., and Etienne Losq. 1985. "International and Conditioning Information." In Portfolio Investment in
Asset Pricing under Mild Segmentation: Theory and Test." Developing Countries, edited by Stijn Claessens and Sudarshan
Journal of Finance, vol. 40, no. 1 (March):105-24. Gooptu. Washington: World Bank:110-44.

Financial Analysts Journal * November/December 1996 45

This content downloaded from

[Link] on Fri, 10 Nov 2023 10:11:25 +00:00
All use subject to [Link]
 . 1995. "Predictable Risk and Returns in Emerging . 1974b. "The International Pricing of Risk: An Empirical
Markets." Reviewu of Financial Studies, vol. 8, no. 3:773-816. Investigation of the World Capital Market Structure." Journal of

Harvey, Campbell R., Bruno H. Solnik, and Guofu Zhou. 1994. Finance, vol. 29, no. 2 (May):48-54.

"What Determines Expected International Asset Returns?" _ . 1977. "Testing International Asset Pricing: Some Pessi-
Working paper, Duke University. mistic Views." Journal of Finance, vol. 32, no. 2 (May): 503-11.

Harvey, Campbell R., and Guofu Zhou. 1993. "International _ . 1983a. "The Relationship between Stock Prices and
Asset Pricing with Alternative Distributional Assumptions." Inflationary Expectations: The International Evidence." Journal
Journal of Empirical Finance, vol. 1, no. 1:107-31. of Finance, vol. 38, no. 1 (March):35-48.

Liew, John M. 1995. "Stock Returns, Inflation, and the Volatility _ . 1983b. "International Arbitrage Pricing Theory." Journal
of Growth in the Money Supply: Evidence from Emerging of Finance, vol. 38, no. 2 (May):449-57.
Markets." Working paper, University of Chicago. Stambaugh, Robert F. 1982. "On the Exclusion of Assets from
Lintner, John. 1965. "The Valuation of Risk Assets and the Tests of the Two Parameter Model: A Sensitivity Analysis."
Selection of Risky Investments in Stock Portfolios and Capital Journal of Financial Economics, vol. 10, no. 3 (November):237-68.
Budgets." Reviezw' of Economics and Statistics, vol. 47, no. 1:13-37.
Stehle, Richard. 1977. "An Empirical Test of the Alternative
Ross, Stephen A. 1976. "The Arbitrage Theory of Capital Asset Hypotheses of National and International Pricing of Risky
Pricing." Journal of Economic Theory, vol. 13, no. 3:341-60. Assets." Journal of Finance, vol. 32, no. 2 (May):493-502.

Shanken, Jay. 1990. "Intertemporal Asset Pricing: An Empirical Stulz, Rene. 1981a. "On the Effects of Barriers to International
Investigation." Journal of Econometrics, vol. 45, no. 1 (July/ Investment." Journal of Finance, vol. 36, no. 4 (September):923-34.
August):99-120. _ . 1981b. "A Model of International Asset Pricing." Journal
Sharpe, William. 1964. "Capital Asset Prices: A Theory of of Financial Economics, vol. 9, no. 4 (December):383-406.
Market Equilibrium under Conditions of Risk." Journal of World Bank. 1993. Emerging Stock Markets Factbook. Washington:
Finance, vol. 19, no. 3 (September):425-42. International Finance Corporation.
Solnik, Bruno. 1974a. "An Equilibrium Model of the
Intemational Capital Market." Journal of Economic Theory, vol. 8,
no. 4:500-24.

46 ?Association for Investment Management and Research

This content downloaded from

[Link] on Fri, 10 Nov 2023 10:11:25 +00:00
All use subject to [Link]