RiskMetrics
introduced RiskGrades to address investors' need for a consistent and
reliable way to measure market risk. RiskGrades are a new statistic that
allows for risk comparison across all asset classes. Like standard
deviation, RiskGrade quantifies the volatility of financial assets.
RiskGrades, however, are calibrated to be more intuitive and easier to
use than standard deviations. RiskGrades can range from 0 to over 1000,
where 100 corresponds to the average risk of a diversified market-cap
weighted index of global equities.
RiskGrades
are a standardized measure of volatility, and therefore allow an "apples
to apples" comparison of investment risk across all asset classes and
regions. Thus, we can say that a Brazilian stock with a RiskGrade of 300
is six times as risky as an Asian Bond Fund with a RiskGrade of 50.
Furthermore, RiskGrades capture all components of market risk: equity,
interest rate, currency, and commodity risk.
The
RiskGrade scale below compares the relative risk of some major asset
classes. Conservative investors usually don't venture much beyond major
government bonds, which typically have a RiskGrade below 30, while
aggressive investors are comfortable with high tech stocks like Yahoo!,
which often have RiskGrades in excess of 300.
RiskGrades are not constant through time, and
adjust to current market conditions: during turbulent times, such as
the Asian Crisis or the Russian devaluation, RiskGrades of major stock
markets can easily escalate beyond 200 to reflect higher risk, while
calmer markets could yield RiskGrades below 80.
Evolution of RiskGrades through time |
RiskGrades change constantly to reflect current market risk. During
crises, RiskGrades spike up and then slowly settle down as markets
calm.
RiskGrades dynamically gage market volatility.
It's immediately clear from the graph below that Brazilian stocks (BOVESPA
index) are generally far riskier than US stocks (S&P 500 index).
Ranging from 110 to 550, BOVESPA RiskGrades are consistently higher
than S&P 500 RiskGrades (90 to 180 range). As an Emerging Market,
Brazil is very sensitive to exogenous market shocks, such as the '97
Asian Crisis, and '98 Russian Devaluation (notice the spikes in
RiskGrades). Not surprisingly, Brazil's risk peaked with its
currency devaluation on Jan 13 1999. Notice also that all shocks
resulted in a rapid escalation of risk: events rarely come alone,
but rather are followed by a rapid succession of aftershocks. We
call this effect volatility
clustering.
As the graph above shows, risk is fluid and can spread rapidly like
a global storm, encompasing both emerging and developed markets.
Given a significant event, such as a major currency devaluation or
stock market crash, RiskGrades could easily double overnight.
Investing in any given asset does not always carry the same risk.
For example, holding U.S. stocks during the Russian crisis in
October 1998 was over 50% riskier than in April 1998 or April 1999,
when markets were closer to normal (i.e., with S&P 500 RiskGrades
around 100). Emerging markets in particular (e.g., Latin America,
Eastern Europe, Southeast Asia) are prone to extreme risk
escalations.
RiskGrades can help investors monitor their ongoing exposure to
market risk, which is ever-changing. RiskGrades are a benchmark for
the level of current market risk (e.g., is it above or below
average?). And while RiskGrades cannot predict unforseen events,
they quickly adapt to reflect impending aftershocks (i.e.,
volatility clusters). |
Below you
can plot the evolution of RiskGrades for a diversified equity index (S&P
500), a Blue Chip stock, General Electric (GE), and a technology stock
(Yahoo!):
Mouse-over the line chart above to see each asset's historical RiskGrade
measurement
Want to
know
how we calculate RiskGrades?
|
Historical volatility |
Similar to standard deviation
|
Exponential weighting
|
Calibration
Transparent risk benchmark
|
|
How we calculate RiskGrades |
RiskGrades are forecasts based on the analysis of historical
market volatility. Rather than predict which way the market
moves (e.g., whether the S&P 500 moves up or down next month),
we forecast how large market movements are likely to be (e.g.,
what is the chance of the S&P 500 moving by more than 10% next
month). Risk estimates are generally much more stable and
reliable than directional forecasts, and allow high-confidence
risk forecasts (between 95% to 99% confidence).
|
As a measure of volatility, the calculations behind RiskGrades
are similar to standard deviation. RiskGrades are derived by
observing how much past returns have deviated from their mean.
However, there are two main differences between RiskGrades and
simple standard deviations:
- The first is that RiskGrade
estimates are based on
exponential weighting
of historical data, which makes them more adaptive to current
market conditions than plain standard deviations. When J.P.
Morgan released the RiskMetrics methodology, it revealed a
series of studies that demonstrated that exponential weighting
significantly improved forecasting accuracy and responsiveness
in extreme market conditions.
- The second difference is that
RiskGrades have been calibrated for easier interpretation by
the general public. A RiskGrade of 100 is scaled to reflect
20% annualized standard deviation, which was the average
volatility of (market-cap weighted) global equities from
January 1995 to December 1999. Hence a RiskGrade of 100 is
generally our baseline for normal equity market risk, and can
be used to compare all other investments (e.g., a bond fund
with a RiskGrade of 33 is about a third as risky as a
diversified global portfolio of stocks during normal market
conditions).
To promote RiskGrades as a transparent
global benchmark for measuring risk, RiskMetrics publishes the
complete methodology, available in the form of a
RiskGrades
Technical Document, for free
download over the Internet. |
|
|