CUSTOMIZED, SYSTEMATIC COUNTRY RISK ASSESSMENT IS CRITICAL FOR
COMPANIES THAT CONTEMPLATE ACTIVITY ABROAD.
Country risk analysis (CRA) attempts to identify imbalances that
increase the risk of a shortfall in the expected return of a
cross-border investment. This paper describes the general process used
to create risk measures and discusses some of the weaknesses of this
process. It then examines the degree of association of six measures and
analyzes the ability of these measures to predict returns for a
manufacturing investment. The paper concludes that company analysts may
improve the performance of risk measures available from commercial
services by adjusting risk measurement to fit the company's
specific type of foreign direct investment.
Introduction
All business transactions involve some degree of risk. When
business transactions occur across international borders, they carry
additional risks not present in domestic transactions. These additional
risks, called country risks, typically include risks arising from a
variety of national differences in economic structures, policies,
socio-political institutions, geography, and currencies. Country risk
analysis (CRA) attempts to identify the potential for these risks to
decrease the expected return of a cross-border investment.
"Risk" implies that an analyst can identify a
well-defined event drawn from a large sample of observations. A large
sample contains enough observations to develop a statistical function
amenable to probability analysis. An event that lacks these requirements
moves toward uncertainty on the continuum between pure risk and pure
uncertainty. For example, the probability of death from an auto accident
classifies as a risk; the probability of death from a nuclear meltdown
falls into uncertainty, given a lack of nuclear meltdown observations.
Many of the individual events investigated by country risk analysis fall
closer to uncertainties than well-defined statistical risks. This forces
analysts to construct risk measures from theoretical or judgmental,
rather than probabilistic, foundations.
Uncertainty makes CRA more similar to a soft art than a hard
science. Analysts deal with the soft nature of CRA in different ways,
which can result in widely varying views of the risk level of a country.
For this reason, users of risk measures developed from commercial
country-risk services must understand analysts' construction
methods if they wish to analyze a company investment risk appropriately.
As demonstrated in the sections below, company analysts should be able
to improve upon outside measures by adapting risk systems to their
specific company investments.
Theory vs. Practice
Country risk analysis rests on the fundamental premise that growing
imbalances in economic, social, or political factors increase the risk
of a shortfall in the expected return on an investment. Imbalances in a
specific risk factor map to one or more risk categories. Mapping all the
factors at the appropriate level of influence creates an overall
assessment of investment risk. The mapping structure differs for each
type of investment, so an imbalance in a given factor produces different
risks for different investments.
This fundamental premise provides a simple theoretical underpinning
to CRA. Unfortunately, no comprehensive country risk theory exists to
guide the mapping process. [1] In practice, most country-risk services
create risk measures using an eclectic mix of economic or sociopolitical
indicators based on selection criteria arising from their analysts'
experiences and judgment. The services usually combine a variety of
factors representing actual and potential imbalances into a
comprehensive risk assessment that applies to a broad investment
category. Most CRA literature emphasizes a number of common points, then
slips into a detailed discussion of ways the respective authors
enumerate risk for various investments. The best authors emphasize the
necessity to adapt their analyses for a specific investment decision
given the judgmental nature of their methods.
Country Risk Categories and Measurements
Analysts have tended to separate country risk into the six main
categories of risk shown below. Many of these categories overlap each
other, given the interrelationship of the domestic economy with the
political system and with the international community. Even though many
risk analysts may not agree completely with this list, these six
concepts tend to show up in risk ratings from most services.
I. Economic Risk
II. Transfer Risk
III. Exchange Rate Risk
IV. Location or Neighborhood Risk
V. Sovereign Risk
VI. Political Risk
Economic Risk is the significant change in the economic structure
or growth rate that produces a major change in the expected return of an
investment. Risk arises from the potential for detrimental changes in
fundamental economic policy goals (fiscal, monetary, international, or
wealth distribution or creation) or a significant change in a
country's comparative advantage (e.g., resource depletion, industry
decline, demographic shift, etc.). Economic risk often overlaps with
political risk in some measurement systems since both deal with policy.
Economic risk measures include traditional measures of fiscal and
monetary policy, such as the size and composition of government
expenditures, tax policy, the government's debt situation, and
monetary policy and financial maturity. For longer-term investments,
measures focus on long-run growth factors, the degree of openness of the
economy, and institutional factors that might affect wealth creation.
Transfer Risk is the risk arising from a decision by a foreign
government to restrict capital movements. Restrictions could make it
difficult to repatriate profits, dividends, or capital. Because a
government can change capital-movement rules at any time, transfer risk
applies to all types of investments. It usually is analyzed as a
function of a country's ability to earn foreign currency, with the
implication that difficulty earning foreign currency increases the
probability that some form of capital controls can emerge. Quantifying
the risk remains difficult because the decision to restrict capital may
be a purely political response to another problem. For example,
Malaysia's decision to impose capital controls and fix the exchange
rate in the midst of the Asian currency crisis was a political solution
to an exchange-rate problem. Quantitative measures typically used to
assess transfer risk provided little guidance to predict Malaysia's
actions.
Transfer risk measures typically include the ratio of debt service
payments to exports or to exports plus net foreign direct investment,
the amount and structure of foreign debt relative to income, foreign
currency reserves divided by various import categories, and measures
related to the current account status. Trends in these quantitative
measures reveal potential imbalances that could lead a country to
restrict certain types of capital flows. For example, a growing current
account deficit as a percent of GDP implies an ever-greater need for
foreign exchange to cover that deficit. The risk of a transfer problem
increases if no offsetting changes develop in the capital account.
Exchange Risk is an unexpected adverse movement in the exchange
rate. Exchange risk includes an unexpected change in currency regime
such as a change from a fixed to a floating exchange rate. Economic
theory guides exchange rate risk analysis over longer periods of time
(more than one to two years). Short-term pressures, while influenced by
economic fundamentals, tend to be driven by currency trading momentum
best assessed by currency traders. In the short run, risk for many
currencies can be eliminated at an acceptable cost through various
hedging mechanisms and futures arrangements. Currency hedging becomes
impractical over the life of the plant or similar direct investment, so
exchange risk rises unless natural hedges (alignment of revenues and
costs in the same currency) can be developed.
Many of the quantitative measures used to identify transfer risk
also identify exchange rate risk since a sharp depreciation of the
currency can reduce some of the imbalances that lead to increased
transfer risk. A country's exchange rate policy may help isolate
exchange risk. Managed floats, where the government attempts to control
the currency in a narrow trading range, tend to possess higher risk than
fixed or currency board systems. Floating exchange rate systems
generally sustain the lowest risk of producing an unexpected adverse
exchange movement. The degree of over- or under-valuation of a currency
also can help isolate exchange rate risk.
Location or Neighborhood Risk includes spillover effects caused by
problems in a region, in a country's trading partner, or in
countries with similar perceived characteristics. While similar country
characteristics may suggest susceptibility to contagion (Latin countries
in the 1980s, the Asian contagion in 1997-1998), this category provides
analysts with one of the more difficult risk assessment problems.
Geographic position provides the simplest measure of location risk.
Trading partners, international trading alliances (such as Mercosur,
NAFTA, and EU), size, borders, and distance from economically or
politically important countries or regions can also help define location
risk.
Sovereign Risk concerns whether a government will be unwilling or
unable to meet its loan obligations, or is likely to renege on loans it
guarantees. Sovereign risk can relate to transfer risk in that a
government may run out of foreign exchange due to unfavorable
developments in its balance of payments. It also relates to political
risk in that a government may decide not to honor its commitments for
political reasons. The CRA literature designates sovereign risk as a
separate category because a private lender faces a unique risk in
dealing with a sovereign government. Should the government decide not to
meet its obligations, the private lender realistically cannot sue the
foreign government without its permission.
Sovereign-risk measures of a government's ability to pay are
similar to transfer-risk measures. Measures of willingness to pay
require an assessment of the history of a government's repayment
performance, an analysis of the potential costs to the borrowing
government of debt repudiation, and a study of the potential for debt
rescheduling by consortiums of private lenders or international
institutions. The international setting may further complicate sovereign
risk. In a recent example, IMF guarantees to Brazil in late 1998 were
designed to stop the spread of an international financial crisis. Had
Brazil's imbalances developed before the Asian and Russian
financial crises, Brazil probably would not have received the same level
of support, and sovereign risk would have been higher.
Political Risk concerns risk of a change in political institutions
stemming from a change in government control, social fabric, or other
noneconomic factor. This category covers the potential for internal and
external conflicts, expropriation risk and traditional political
analysis. Risk assessment requires analysis of many factors, including
the relationships of various groups in a country, the decision-making
process in the government, and the history of the country. Insurance
exists for some political risks, obtainable from a number of government
agencies (such as the Overseas Private Investment Corporation in the
United States) and international organizations (such as the World
Bank's Multilateral Investment Guarantee Agency).
Few quantitative measures exist to help assess political risk.
Measurement approaches range from various classification methods (such
as type of political structure, range and diversity of ethnic structure,
civil or external strife incidents), to surveys or analyses by political
experts. Most services tend to use country experts who grade or rank
multiple socio-political factors and produce a written analysis to
accompany their grades or scales. Company analysts may also develop
political risk estimates for their business through discussions with
local country agents or visits to other companies operating similar
businesses in the country. In many risk systems, analysts reduce
political risk to some type of index or relative measure. Unfortunately,
little theoretical guidance exists to help quantify political risk, so
many "systems" prove difficult to replicate over time as
various socio-political events ascend or decline in importance in the
view of the individual analyst.
Aggregate Risk Measures
Country risk analysis in the 197Os and 1980s tended to focus on the
risk a private lender such as a bank incurred when it made a hard
currency loan to a sovereign government outside its home country. Risks
were segmented to identify potential shortfalls in either the foreign
currency value of the investment or in the investor's home currency
(returns hold up in local currency, but decline when measured in the
investor's own currency). Quantitative risk analysis generally
focused on factors related to a country's ability to earn foreign
currency to repay the debt. Qualitative analysis attempted to ascertain
a country's willingness to repay the debt. This type of analysis
tended to focus on the sovereign, transfer, and short-term exchange rate
risk categories. With minor adjustments, this analytical approach also
was used to assess risk in short-term investments in foreign private
financial assets.
A multinational enterprise (MNE) that builds a plant in a foreign
country faces different risks than a bank lending to a foreign
government. The MNE must consider a longer time horizon and risks from a
much broader spectrum of country characteristics. Some categories
pertinent to a plant investment contain a much higher degree of risk
simply because the MNE remains exposed to risk for a much longer period
of time.
Table 1 gives the author's subjective view of the impact of
the six risk categories on different types of investments. The
investments are all assumed to be made in the foreign currency. The risk
impacts would change somewhat if the investments were denominated in own
currency (e.g., dollar costs of equipment made in the United States).
While all major categories potentially pose some degree of risk for
each type of investment, the longer time horizon for a direct investment
produces high impacts from a greater number of risk categories.
Specifically, economic, political, and location risks become more
problematic for a fixed investment lasting twenty or more years.
Transfer risk, on the other hand, can pose less of a risk to a long term
fixed investment since capital restrictions are unlikely to last for the
entire period of the investment. Countries typically impose capital
restrictions to help manage temporary foreign exchange shortages. MNEs
often can reinvest profits locally and wait out the restrictions without
severe negative impacts on the return to the investment over the
project's full life.
Companies can acquire country risk measures from a large number of
sources. [1] The Handbook of Country and Political Risk Analysis (Coplin
& O'Leary, eds. 1994) describes calculation methods and risk
information available from ten services. Other risk information
services, including Standard & Poor's DRI and The WEFA Group,
also describe their risk construction methods in detail. While a
comprehensive review of each measurement's construction is well
beyond the scope of this article, a brief description of the development
of a generic risk measure for a direct investment will give an
understanding of the process most services use to create their risk
measure.
Typically, a risk analysis team begins by specifying the type of
investment the system will measure, in this case a manufacturing plant.
The team then decides the relative influence of each of the six country
risk categories exert on the plant's operation and output. Using
Table 1 as a guide, the team devises an approach that gives the greatest
weight to economic risk, with slightly lower weights assigned to
exchange, location, and political risk. Since the plant's working
life exceeds twenty years, the team gives transfer risk a low weight. It
drops sovereign risk because the plant will be financed privately and
sell its output into private markets.
Next, the team selects indicators and measurement methods for each
risk category. It decides to use a common scaling system (risk factor
scales of 1 to 5, lowest to highest) and develops a scheme to classify
imbalances in each indicator. This scheme takes the form of a numeric
scale for quantifiable factors (e.g., assign a rating of 3 to current
account deficits greater than X% of GDP, a 4 to deficits greater than
(X+2)%, etc.). For non-quantifiable indicators, the team relies on a
judgmental assessment from its political expert who is tasked with
scaling political risk factors (e.g., assign a rating of 3 for unstable
political situation in a democracy, a 5 for a change to an autocratic
government). The team creates a measure for each of the five risk
categories by combining the individual scale values (e.g., current
account deficit rating plus import coverage rating plus debt service
rating equals transfer risk measure).
The team calculates its total country risk measure as a weighted
index using the weighting scheme dictated by the relative importance of
each risk category. By using the same weighting scheme to create
measures for each country, team members can compare the relative risk
their manufacturing plant faces in different countries.
The brief overview indicates some of the reasons risk measures for
a country may differ from one source to another. Analysts assessing the
same investment risk may use different indicators in their risk
categories, weight them differently in their final measures, or classify
individual factors as risky at different levels. For example, while
almost every system uses the current account deficit as a percent of GDP
as an indicator, the level at which it begins to increase risk varies
widely.
To demonstrate the variability of risk measures, the author
examined risk measures from six sources for thirty-eight countries. [3]
The measures come from four risk services (S-I to S-IV), the
S&P's long-term sovereign local currency debt rating, and the
author's company-specific manufacturing risk measure. S-II and the
author specifically measure risk for a manufacturing investment. S-I
measures five-year ahead direct investment risk for all industries,
S-III measures a composite investment risk, and S-IV measures economic
risk. Five of the measures were released in the first half of 1996, S-II
was released in the fourth quarter of 1996. Risk measures created in the
first quarter of 1995 were also available from S-I, the author, and, for
a 26-country subset, the S&P.
The team calculates its total country risk measure as a weighted
index using the weighting scheme dictated by the relative importance of
each risk category. By using the same weighting scheme to create
measures for each country, team members can compare the relative risk
their manufacturing plant faces in different countries.
The calculation methods used by the services, described in either
Coplin & O'Leary (1994) or provided to the author by the
respective services, indicated ordinal scaling was the highest level of
measurement appropriate for a six-way analysis. Countries were ranked 1
to 38, from lowest to highest risk, with ties receiving the average of
the ranks. The measures were then analyzed using nonparametric methods
appropriate for ordinal data (see Siegel (1956) for an excellent
discussion of nonparametric statistics). Table 2 displays the results of
the analysis.
The Kendall Coefficient of Concordance W measures the degree of
association among all risk rankings. At W=.73, the coefficient is
significant at the .001 level, so the six rankings as a group exhibit a
high degree agreement in the ranking of investment risk. On the other
hand, pair-wise correlation coefficients range from a high of .86 to a
low of .43, so this agreement is not exact. Some of the correlation
differences may be attributed to different emphasis of risk categories.
For example, S-IV and the author weigh economic factors heavily in their
measures while S-II, S-III, and the S&P focus more on transfer and
exchange risk categories. While all pairs possess statistically
significant correlation coefficients, the average pairwise correlation
of .64 indicates a great deal of variability arising from the different
measures of country risk.
In the smaller sample of twenty-six country risk measures prepared
in early 1995, S-I and the author's measure had a low but
statistically significant correlation coefficient of .44. Neither S-I
nor the author's risk measure demonstrated a significant
relationship with the S&P sovereign risk measure (.33 and .30
correlation, respectively). This smaller sample supported the general
results of the larger sample.
Predictability and Returns
A firm making a plant investment overseas needs its risk analysis
to identify economically detrimental developments, not necessarily
relate closely to other risk measures. The relationship between risk
measures from Table 2 and a measure of returns earned by U.S.
manufacturing firms on their direct investments abroad provides one
measure of effectiveness relevant for a manufacturing firm.
One would expect returns in any given year to be negatively
correlated with forward-looking country risk. To see this, assume all
countries had identical past risk outlooks, so investments were made
with the expectation of earning identical returns in all countries.
Then, in the current period, imbalances develop to increase risk in some
countries, while in others conditions improve to lower risk. By design,
the imbalances signaling higher future risk should suppress the current
income received from past investments. Conversely, low risk measures
imply a more favorable environment for current income.
As time passes, returns should gradually shift to align positively
with the current period's risk measure because manufacturers
earning low or negative returns in high-risk countries will abandon
business, and new investments into high-risk countries will be made only
if manufacturers can earn higher returns. Investments into low-risk
countries will require lower returns, gradually reducing the average
return in low-risk countries. This dynamic will show up as a gradual
shift from a negative risk-return relationship to a positive
relationship between the current risk measure and future returns.
To test this assumption, annual manufacturing returns were
developed from Bureau of Economic Analysis data for income and direct
investment position abroad (Survey of Current Business, October 1998).
The most recent data covers various sectors, including manufacturing,
for the years 1994 to 1997. The investment position abroad consists of
historical investments measured in current dollars. Income includes
income net of local withholding taxes in a given year. While not a
perfect measure of returns, [4] the ratio of income to investment gives
a useful approximation of the annual profitability of U.S. manufacturing
overseas investments.
Returns for the thirty-eight countries were ranked for 1996 and
1997. Table 3 displays nonparametric correlation coefficients between
the two annual return rankings with the six risk measures from above.
The table also includes the change in the correlation coefficient from
1996 to 1997.
The casual observer immediately notes there are no high correlation
coefficients, and no statistically significant correlation between risk
and return for any of the measures. Risk-return correlation in 1996 was
negative for only S-I, S-IV, and the author. Signs changed from negative
to positive between 1996 and 1997 for S-IV's and the author's
measures, and became more positive for S-II and S-III. The S&P
correlation weakened, perhaps reflecting a difference between sovereign
and manufacturing investment risk. While there may be a faint indication
that S-IV and the author's risk-return rankings behave as expected,
the results certainly are not conclusive on the ordinal ranking data.
To compare all six risk measures from Table 2, returns for
manufacturing were ranked. Ranking the return measures ignores ratio
information in the data, and forces countries with very similar returns
into an order that may not be accurate given the weaknesses in the
measurement of the underlying return data. Figure 1 displays the
distribution of 1996 manufacturing returns by country. Ranking
uncertainty potentially exists among countries showing returns of around
ten, twelve and fifteen percent:
The author's risk measure attempts to overcome some of the
limitations of traditional risk service measures with a fuzzy logic
approach to risk measurement. [5] Risk measurement with this fuzzy logic
system produces an interval scale that enables some parametric analysis.
When valid, parametric analysis takes advantage of the greater
information embedded in interval data, specifically, the closeness of
some of the country returns. Table 4 provides a parametric risk-return
analysis of S-I, S&P, and the author's measures for the S&P
twenty-six country subset.
The insignificant S&P results are not surprising given a high
number of identical risk measures among the countries in this sample (18
developed countries in the sample had AAA ratings). S-I also had a high
concentration of equivalent risk ratings among developed countries. The
author's measure, on the other hand, produces results consistent
with expectations. Significant negative risk-return correlation in the
current years (1994,1995) moves in the direction of a positive current
risk-future return relationship (declining negative correlation with
1996 and 1997 returns).
Adding risk measures for fourteen more developing countries rated
by both the author and S-I dilutes the strength of the correlation
results a little (Table 5). The author's measures, however,
continue to show the expected sign and movement from negative
correlation to positive correlation.
Figure 2 demonstrates graphically the relationship between the
author's 1995 January risk measures and manufacturing returns from
1994 to 1997. The patterns in the figures show the expected shift in the
relationship as demonstrated by the declining slope of the trendlines
through the years. Obviously, high dispersion exists in relationship,
but Figure 2 demonstrates that the risk measure specifically designed to
capture the author's manufacturing company investments behaves as
expected. Its superior performance to any other measure in this study
also supports the value of a company-specific risk measure to improve
the company's understanding of the investment environment, rather
than rely solely on outside measures.
Concluding Remarks
Risk measures examined here displayed general agreement concerning
the relative risk in thirty-eight countries. Risk measures from external
services, however, performed poorly as predictors of one-
to-two-year-ahead manufacturing foreign investment returns. Some of that
inability may be caused by differences in the specific investment for
which the risk measures were created, some may be caused by weaknesses
in the risk measurement system. The best performing measure, created by
the author, was specifically designed to measure longer-term direct
investment risk for a manufacturing firm. The better results of this
measure give some indication that an analyst should be able to add value
by adapting external information to a company's specific investment
type.
Companies investing overseas should consider country risk in a
systematic approach consistent with the types of investments they are
making. If they use an external service, measures from that service must
be tested for relevance. Ideally, the measures should be recombined in a
system that better relates to a company's specific investment
needs. Some external systems invite such recombination by making all of
their individual risk measures available. In any case, a company needs
to examine the relationship between risk and its businesses to make sure
risk measures actually help the company improve its business decisions.
Finally, the weakness of the results also reflects the weakness of
the state of country risk analysis. The field would benefit greatly from
additional research into the theoretical and quantitative relationships
between risk and the returns earned in cross-border investments.
Duncan H. Meldrum is Corporate Economist, Air Products and
Chemicals, Inc., Allentown, PA. This paper reflects the opinions of the
author and not those of Air Products and Chemicals, Inc.
ENDNOTES
(1.) The July 1999 issue of Business Economics presents the
author's attempt to integrate elements of new growth theory into a
longer term, theoretically-based measure of country risk ("Country
Risk and a Quick Look at Latin America").
(2.) The Economist Intelligence Unit (EIU), Political Risk
Services' Risk Letter and International Country Risk Guide,
Standard & Poor's DRI and risk rating, The WEFA Group, Standard
& Poor's, Moody's, Euromoney, BERI, Rundt's, to list
a few.
(3.) Risk measures were obtained from marketing literature,
subscriptions by the author's company and a Duke University web
site (www.duke.edu/[tilde]charvey/ Country_risk/pol/poltab6.htm).
Services were provided anonymity by the author.
(4.) Valid criticisms include the fact that exchange rate
fluctuations affect historical investment positions. Also, because most
direct investments earn less in early years, returns in countries just
recently open to U.S. direct investment will most likely be understated
relative to countries with a long-established U.S. investment position.
(5.) See "Country Risk and a Quick Look at Latin America"
in the July 99 issue of Business Economics for a more detailed
discussion of the author's methods to calculate risk measures.
REFERENCES
Coplin, William D. and & O'Leary, Michael K., editors, The
Handbook of Country and Political Risk Analysis, East Syracuse, New
York: Political Risk Services, International Business Communications,
1994.
Siegel, Sidney, Nonparametric Statistics for the Behavioral
Sciences, New York: McGraw-Hill, Inc., 1956.
U.S. Bureau of Economic Analysis, Survey of Current Business,
October, 1998.
RISK IMPACT BY INVESTMENT TYPE
Direct Short Term Short Term Long Term
Risk Investment Financial Loan to Loan to
Category: Private Sector Private Sector Governmet Governmet
Economic High Low Low Low to Moderate
Transfer Moderate High High Moderate
Exchange High None to High Non to High High
Location High Moderate Low Moderate
Sovereign Low Low High High
Political High Low Moderete High
1996 INVESTMENT/ECONOMIC RISK
MEASURES (KENDALL COEFFICIENT OF
CONCORDANCE W = .73)
Spearman Pair-wise Rank Correlation Coefficients
(Rankings for38 Countries, Corrected for Ties)
Source: S-I S-II S-III S-IV S&P
S-I -
S-II .61 -
S-III .61 .80 -
S-IV .53 .43 .66 -
S&P .67 .82 .86 .59 -
Author .56 .46 .65 .77 .60
All correlation coefficients significant at .01 level
1996 RISK MEASURE VS
MANUFACTURING RETURNS
SPEARMAN RANK CORRELATION - 38 COUNTRIES
Returns for Change in
1996 Risk Measure from: 1996 1997 Coefficient
S-I -.05 -.03 +.02
S-II .07 .12 +.05
S-III .22 .27 +.05
S-IV -.10 .10 +.20
S & P .21 .11 -.10
Author -.08 .14 +.22
1995 RISK MEASURE VS
MANUFACTURING RETURNS
CORRELATION COEFFICIENTS - 26 COUNTRIES
Mfg. Returns in Year:
Jan 95 Risk: 1994 1995 1996 1997
S-I -.05 .05 .06 .06
S&P .28 .11 .09 .03
Author -.45 [*] -.51 [*] -.43 [*] -.34 [*]
(*.)Statistically Significant at .01 level
1995 RISK MEASURE VERSUS
MANUFACTURING RETURNS
CORRELATION COEFFICIENTS - 40 COUNTRIES
Mfg. Returns in Year:
JAN 95 Risk: 1994 1995 1996 1997
S-I .07 .22 .09 .15
Author -.33 [*] -.22 -.12 .01
(*.)Significant at .05 level