While interpreting violent market movements can potentially be illuminating, many experienced finance practitioners shy away from this exercise, having recognized the difficulty of the task. At the same time, the majority of them appear to have accepted the premise that supply-and-demand dislocations have emerged as increasingly influential forces driving the fixed income markets. Their impact on interest rates, credit spreads, and implied as well as realized volatilities rivals that of geopolitical events, economic fundamentals, and credit crises. Paradoxically, the adoption of similar risk management practices by a large proportion of the financial system has become an extremely potent technical factor that may exacerbate market volatility and result in financial losses. One cannot help but wonder whether the creation of the field of risk management, directed toward protecting institutions from financial losses and decreasing the overall volatility, has at times become a self-defeating prophesy. Obviously, the realization of risk management's potential to affect market behavior, particularly as it applies to dynamic hedging strategies, is not new.
During the credit and liquidity crisis of 1998, all investors simultaneously tried to reduce total risk per unit of capital and raise cash to cover margin calls. First, they naturally attempted to sell illiquid positions. After discovering there was no a bid a fire-sale of liquid positions ensued. This was compounded by lenders increasing haircuts on illiquid leveraged positions, thus forcing additional liquidations and further depressing the value of illiquid positions, which exacerbated margin calls. In general, such behavior may significantly increase correlations among asset classes with dramatically different fundamentals.
During last year's corporate governance crisis that followed accounting fraud and landmark corporate defaults, extreme aversion to credit risk resulted in investors and lenders significantly cutting exposures to short-maturity credit risk. Unable to borrow in the commercial paper market and via bank lines of credit, many corporations were forced to seek financing by issuing long-term debt. This led to a dramatic lengthening of the duration of liabilities that needed to be hedged. The attendant receiving fixed on interest rate swaps significantly tightened swap spreads, hence compressing the nominal spreads to Treasuries on a variety of products that use swaps as a benchmark.
The behavior of the fixed income markets in the wake of the Sept. 11 attacks and subsequent major market dislocations followed a different pattern. Due to a wide-spread adoption of sophisticated stress testing techniques, better leverage management, and other enhancements, it seemed that the impact of risk management on financial market behavior was better anticipated and incorporated into investment and business decisions. It has been argued that the change in the market dynamic was a direct consequence of the lessons learned from previous market crises. Extreme market movements in both the U.S. and Japan this summer, however, may indicate otherwise.
The Impact Of The Mortgage Market
Over the last three years interest rates in the U.S. have declined to 50-year lows, resulting in four consecutive waves of mortgage prepayments progressively larger in magnitude. The resulting unprecedented compression of the coupon concentration of the mortgage universe created a significant exposure to duration extension in the event of a rapid interest rate sell-off. Note that the vast majority of financial institutions in the U.S., including both absolute and relative-return asset managers, actively manage durations of their portfolios. Whenever these durations exceed pre-defined risk limits, investors hedge increasing exposures to interest rates by paying fixed on swaps or by selling various securities, such as Treasuries. It is therefore not surprising that the rise in interest rates from unsustainably low levels in mid-June was greatly amplified by mortgage duration hedging, which, in turn, led to a dramatic widening of swap and mortgage spreads, an increase in implied and realized volatilities, forced selling, and illiquidity.
The Impact Of Value-At-Risk
The extreme volatility of interest rates in Japan in June, when the 10-year Japanese Government Bond yield doubled in less than three weeks, ignited conflicting interpretations of what happened. News stories and commentaries by Wall Street strategists and economists, however, presented a stark contrast with the views of many Japanese institutions who believed that non-economic factors, in particular, the prevalent use of rigid risk management policies based on Value-at-Risk, have greatly exacerbated the sell off and volatility. In this regard, irrespective of the fundamental or technical drivers, a prospect of a further violent interest rate sell-off was extremely worrisome. The Bank of Japan's Governor Toshihiko Fukui even went as far as to identify huge amounts of JGBs "buried like mines" as a significant risk facing the Japanese economy and its banking system.
Recall that Value-at-Risk (VaR) provides a conceptual framework for presenting financial risk as a single, intuitive summary measure. VaR estimates losses (described by the left (loss) tail of the probability distribution of returns on a security or portfolio) resulting from extremely large market movements for a given confidence level (e.g., 95%) and investment horizon (e.g., one month). Practical implementations of Value-at-Risk can be thought of as consisting of the following three steps:
1 Use valuation models to analyze the dependency of the
portfolio's market value on changes in underlying risk
factors
2 Estimate the joint probability distribution of these risk
factors
3 Construct the probability distribution of the portfolio's
market value by integrating results from steps one and two.
Step two is of particular interest to this article since, depending on the number of data points used to estimate probability distributions of risk factors, large market movements can dramatically change estimates of correlations and volatilities, causing Value-at-Risk measures to increase even if the composition of the portfolio remains unchanged. If Value-at-Risk is used as a dominant measure of risk (as it is believed to be the case among most depository institutions in Japan) and rigid policies force transactions when VaR exceeds predefined risk limits, an increase in market volatility may result in forced selling, exacerbating volatility and losses.
In order to understand the origins of the massive interest rate sell-off in June, let us describe some of the technical factors that contributed to the preceding interest rate decline. Back then, the Financial Securities Association toughened its stance on the capital adequacy of the largest Japanese banks. The regulator has put significant pressure on these institutions. In response, these institutions allegedly took on an increasing amount of risk in the JGB market in order to achieve earnings objectives. Smaller depositories then increased their holdings of JGBs in order to take advantage of seemingly never-ending price appreciation.
The sell-off started on June 18, when the FSA proposed a new system under which the banking industry would cover any losses the government might incur after having given a precautionary capital injection. Making banks collectively responsible for the government's losses was apparently translated by many as a softer stance prompting large banks to reduce holdings of JGBs. Reducing balance sheet risk appeared to be the prime objective; locking in gains was an important yet secondary consideration. This coincided with the auction of the 20 year bond on which the yield came at 0.8%. This was below the yield target of many insurance companies. In addition to losses, the increase in realized volatilities and correlations of various points on the JGB yield curve caused a sharp increase in Value-at-Risk estimates on depository portfolios, prompting risk managers to reduce positions, exacerbating the sell-off and volatility even further. Importantly, many of these JGBs resided in held-to-maturity accounts, making it virtually impossible for portfolio managers to sell them preemptively. These holdings could only be reduced by risk management departments in a number of circumstances pre-determined by the risk management policies of an institution, including, for instance, a rise in volatility in a sharp sell-off. Unless these practices are changed, a potential for risk management to exacerbate future sell-offs in Japan remains significant.
Conclusion
Given the generally more volatile global financial markets, rigorous risk management is critical for the viability of financial institutions. However, consequences of the adoption of similar risk management practices by a large proportion of the financial system must be analyzed via extensive stress testing of market, credit, liquidity, and financing risks. The implications of these stress tests should be incorporated into investment strategies and business decisions. Moreover, advantages and limitations of various measures of risk must be carefully examined to ensure that risk management policies do not automatically force financial institutions to transact during market dislocations. Accounting constraints must also be better understood in terms of their potential impact. The following important areas need specific improvement:
* Risk measures must be differentiated in terms of their
reliance on statistical models. Policies must be carefully
formulated in terms of hard (e.g., duration) versus soft
(e.g., Value-at-Risk) risk limits. In the vast majority of cases,
a breach of a risk limit should initiate an internal discussion
regarding the course of action rather than an automatic
transaction.
* Given the potential of delta-hedging to amplify market
movements in the same way as portfolio insurance, a larger
proportion of hedging strategies should involve options and
structured products. In that regard, maximum advantage
should be taken of opportunities to purchase options at the
asset/liability management level, e.g., via debt issuance.
The ability to learn from experience and incorporate adverse scenarios into investment decisions and risk management practices will increasingly influence the success, and even the survival of financial institutions. As we have learned over the past few years, another 'one-in-a-thousand-years' market crisis may be just around the corner.
This week's Learning Curve was written by Leo Tilman, head of institutional investment strategies at Bear Stearns in New York, Raymond Wong, senior managing director, and Misahiro Yamaguchi, managing director at Bear Stearns (Japan).
