The authors introduce a risk management tool called “market exposure”
to help money managers determine how risky a portfolio is relative to a
benchmark. They focus on the measurement and management of market risk for a
global bond portfolio. The manager of such a portfolio should define a set of
risk factors and exploit correlations between markets. The authors explain how
to incorporate volatilities and correlations to more accurately measure market
exposure and hence improve overall portfolio management.
The primary task facing money managers today is the quest to beat some specified
benchmark. Popular benchmarks include pension liabilities, a market index, or cash.
Managing the risk of a portfolio is becoming increasingly difficult because portfolios
are including more global securities and derivatives are more widely used. With this
increased complexity comes the need for more powerful and more accurate risk management
tools.
The authors develop a risk management tool called “market exposure” to
address how risky a portfolio is relative to its benchmark. Market exposure is defined
as the sensitivity of a portfolio to moves in the overall market. (Market here is used
to mean the typical basket of permissible securities.) Mathematically, market exposure
is the slope coefficient of a linear regression of the return on the manager's portfolio
on the return of the appropriate market portfolio.
The market portfolio for a domestic equity investor is represented by a domestic equity
index, and the market portfolio for a domestic bond investor is represented by a
domestic bond index. For the bond market, the market exposure is approximately the
duration of the investor's portfolio divided by the duration of the benchmark portfolio
adjusted for differences in the respective volatilities.
Asset returns are typically decomposed into exposures to underlying fundamental risk
factors. Examples in the bond market of relevant risk factors include credit rating and
yield-curve information (level, slope, and curvature). By defining risk factors,
managers are able to attribute any deviation from the benchmark to differences in risk
factor exposures plus a residual term that is hopefully small. The portfolio's risk can
be decomposed into two parts: exposures that affect the overall market and residual
exposures that affect the portfolio but not the market.
Global bond fund managers seek to provide superior returns relative to the appropriate
global bond index. The fund manager must deviate from the allocations implied in the
appropriate global bond index to provide superior performance. Hence, the bond fund
manager takes on risk. The level of risk an active manager assumes can be measured by
the standard deviation of the return differences of the portfolio compared with the
index. This risk measure is called tracking error, which can be decomposed into two
components: market exposure resulting from different allocations from the index and
residual risk resulting from risks that do not affect the index.
The expected portfolio return is equal to the market exposure measure times the index
return because residual risks should not contain any risk premium. Residual risk can be
totally diversified.
The authors examine their approach to risk management—the market exposure
tool—by using daily data from February 1988 through March 1995, and they
downweight at a rate of 10 percent a month. They document that volatilities vary among
maturities and among countries. Also, most major bond markets are not highly correlated
with each other or with currencies. Hence, simply relying on relative durations gives
misleading results because the duration approach assumes that all markets are perfectly
correlated. Using a historical example, the authors show that the market-exposure
approach is much more accurate than other approaches.
The global bond trader typically seeks to be market neutral and takes advantage of
inappropriate relative prices. The authors argue that a portfolio is market neutral if
and only if the market portfolio has no expected excess return.
The authors demonstrate that one can determine a set of implied views for any given set
of selected weights (assuming the selection is optimal). They review several different
approaches for designing a market-neutral trade, including using strictly duration,
using regression weights to minimize volatility, and using volatility-weighted
durations. A zero duration strategy typically has a directional bias because yield-curve
movements are not parallel. Rather than pursue zero duration, the authors recommend
defining a market-neutral trade as one with zero market exposure.
The active bond manager should focus on opportunities that add residual risk while
maintaining an appropriate level of market exposure. Using the authors' measure of
market exposure that explicitly incorporates volatilities and correlations, bond
managers can improve their portfolio management.