Assessing the effect of a bear market

Another form of scenario testing that can be performed with historical simulation is
to measure the effect on a portfolio of an adverse run of price moves, no one price
move of which would cause any concern by itself. The benefit of using historical
simulation is that it enables a specific real life scenario to be tested against the
current portfolio. Figure 8.7 shows how a particular period of time can be selected
and used to perform a scenario test. The example portfolio value would have lost
$517 000 over the two-week period shown, far greater than the largest daily move
during the five years of history examined and 12 times greater than the calculated
95% VaR.

A bank must be able to survive an extended period of losses as well as extreme
market moves over one day. The potential loss over a more extended period is known
as ‘maximum drawdown’ in the hedge-fund industry. Clearly it is easier to manage
a period of losses than it is to manage a sudden one-day move, as there will be more
opportunities to change the structure of the portfolio, or even to liquidate the
portfolio.

Although in theory it is possible to neutralize most positions in a relatively short
period of time this is not always the case. Liquidity can become a real issue in times of market stress. In 1994 there was a sudden downturn in the bond markets. The
first reversal was followed by a period of two weeks in which liquidity was much
reduced. During that period, it was extremely difficult to liquidate Eurobond positions.
These could be hedged with government bonds but that still left banks exposed
to a widening of the spreads between government and Eurobonds. This example
illustrates why examining the impact of a prolonged period of market stress is a
worthwhile part of any stress-testing regime. One of the characteristics of financial
markets is that economic shocks are nearly always accompanied by a liquidity
crunch.

There remains the interesting question of how to choose the extended period over
which to examine a downturn in the market. To take an extreme; it would not make
sense to examine a prolonged bear market over a period of several months or years.
In such prolonged bear markets liquidity returns and positions can be traded out of
in a relatively normal manner. Thus the question of the appropriate length of an
extended period is strongly related to liquidity and the ability to liquidate a position
– which in turn is dependent on the size of the position.

Market crises, in which liquidity is severely reduced, generally do not extend for
very long. In mature and deep markets the maximum period of severely restricted
liquidity is unlikely to last beyond a month, though in emerging markets reduced
liquidity may continue for some time. The Far Eastern and Russian crises did see
liquidity severely reduced for periods of up to two months, after which bargain
hunters returned and generated new liquidity – though at much lower prices. As a
rule of thumb, it would not make sense to use extended periods of greater than one
month in the developed markets and two months in the emerging markets. These
guidelines assume the position to be liquidated is much greater than the normally
traded market size.

In summary, historical simulation is a very useful tool for investigating the impact
of past events on today’s portfolio. Therein also lies its limitation. If a bank’s stress
testing regime relied solely on historical simulation it would be assuming that past
events will recur in the same way as before. This is extremely unlikely to be the case.
In practice, as the dynamics of the world’s financial markets change, the impact of a
shock in one part of the world or on one asset class will be accompanied by a new
combination of other asset/country shocks – unlike anything seen before. The other
limitation is, of course, that historical simulation can only give rise to a relatively
small number of market shock scenarios. This is simply not a sufficiently rigorous
way of undertaking stress testing. It is necessary to stress test shocks to all combinations
of significant risk factors to which the bank is exposed.