On May 1, 2000, we asked the VaR community to voice its opinion on the argument, as put forth in Nicholas Dunbar's book, Inventing Money, that VaR was in part responsible for the market events during the LTCM/Liquidity Crisis of 1998. Below are the fascinating results.

If this is all new to you, you may first read the details of the VaR/LTCM argument. And if this is not enough for you, peruse the list of additional resources.
Survey results: Culprit: 1, Bystander: 38, Neither: 8

Four themes explaining the crisis emerged from the responses of members of the GloriaMundi email list receiving the survey. These are, market panic was the real culprit; too much credit exposure was the issue; VaR was indirectly responsible by permitting complacency about risk; because of the way VaR is calculated and used, VaR would not have come into play.

Below are comments selected from the many great contributions made by survey participants:

1. MARKET PANIC WAS THE CULPRIT

  • Positions may have been cut so that the trading desk would not exceed it's loss limit (unrelated to VaR); Because the unfamiliar market dynamics created "Knightean" uncertainty, traders responded by fleeing to the sidelines (unrelated to VaR); Reduced risk tolerance (at any given level of VaR) led traders to reduce risky positions; Traders rushed to the door just to exit before the general panic set it, but of course, creating the general panic in the process (possibly related to VaR).
  • I was on an Emerging Market Fixed Income trading desk during the Russian meltdown. As the markets began to express skepticism in the country's ability to finance itself through new issues of GKOs, demand evaporated, and Russia could not rollover short-term debt. Trading desks began to hedge GKO positions with anything that ostensibly would offer a hedge, e.g., $-Russian paper, S&P futures, Brazilian securities (because everyone believed Brazil was next). Traders began to crumble under the weight of heavy selling by (highly leveraged) hedge funds. Two new Global Eurobond issues took place in the eye of the storm. The market already was flush with too much paper. Yes, VaR numbers began to increase substantially, but this was more of a result of rather than a cause of the selling. Yes, some controllers used VaR to regulate positions but I think by the time VaR numbers were run either mid-day or day's end, the damage had already been done.
  • Dunbar's argument that liquidation was driven by a fear of an increase in the multiplier from 3 to 4 is silly i.e. it's fear of the losses.
  • If the banks ordered trading desks to cut positions "not just in Russia, but everywhere", this is a strategy induced by panic and not VaR.

2. CREDIT RISK WAS THE REAL ISSUE

  • Many groups were over exposed to Russia, which caused a mass exodus and a domino-effect of liquidations. Perhaps we should blame the credit departments which made the margin calls and turned healthy hedge funds into non-existent or severely-damaged hedge funds. A snap credit squeeze from funding institutions over all 'leveraged' clients was responsible. The large lending institutions knew that a snap squeeze of credit might create liquidations, but were of the opinion that a first-mover approach [over increasing margins] would benefit themselves over the next lender as their clients' were forced to liquidate positions.

3. VAR WAS INDIRECTLY RESPONSIBLE BECAUSE IT MADE THE MARKET COMPLACENT

  • In a sense VaR was the culprit, because of its failings: (a) it does not take into account liquidity risk; (b) it ignores distributional stability issues and outlying, catastrophic risks, and the fact that financial risk is not stationary.
  • The sense in which VaR was a culprit: too many people who look at VaR and think that is how bad it can get - rather than planning what their strategy will be once they have lost two or three times the VaR.
  • For all LTCM's genius, their highly leveraged spread trades went the wrong way and they didn't have enough capital for an event that wouldn't be captured in a 99% confidence VaR.
  • The distinction about whether VaR was the culprit is not material. The risk management "culture" shared by many firms (and of which VaR is a part, if not the cause) created correlations between markets and risks that can not be traced to structural dependencies among those markets and risks.

4. VAR WAS UNRELATED TO THE PROBLEM

  • I believe that VAR made the situation more transparent and expedited a liquidation that would have only been worse if dragged out.
  • After the trading losses, the VaR associated with Russian positions could easily have fallen to near zero (rather than increased), because these securities were basically worthless.
  • No institution manages their daily positions solely with VaR. If anything, the use of this measure is a hard sell to trading desks.

5. OTHER COMMENTS

  • Even though Basle capital requirements apply to banks, many firms in the financial industry (e.g., credit union) calculate the Basle capital ratio. Thus, a lot of firms try to minimize risk weighted capital amount when adding new derivatives transactions or others transactions.
  • Dunbar's model seems to be theoretically valid at least for future events. Assuming that VaR models had a worldwide acceptance, the mechanism of these models might easily trigger or accelerate a financial crisis.
  • People can't excuse themselves by referring to some 'model'. Who is responsible for 'model risk'? We suggest that model risk is the responsibility of the decision taker.


THE ORIGINAL SURVEY FOLLOWS:

VaR's Role in the Liquidity Crisis of '98 - Culprit or Bystander?

One of the major arguments in Nick Dunbar's book, Inventing Money: The story of Long-Term Capital Management and the Legends behind it, is that the "liquidity crisis" of 1998 was caused, indirectly, by VaR models. This is a powerful claim that deserves some critical attention, because, if true, risk managers and regulators to re-evaluate the use of VaR. Below, I outline the argument and then list aspects of the argument that could be explored further.

Dunbar's Position on the Role of VaR in the '98 Liquidity Crisis: Culprit

Dunbar sets the stage for his argument by describing the events surrounding the default by Russia on August 17, 1998, on its short term debt obligations (GKOs). Then, beginning on page 202, Dunbar writes, "Now, with sudden mark-to-market losses, the Russian contribution to VAR rose rapidly. The effect was to cause many trading desks to breach their VAR limits. According to the Basle committee, once a breach took place so many times, more capital would have to be allocated or positions had to be cut."

Dunbar then states that risk managers at investment banks ordered trading desks to cut positions "not just in Russia, but everywhere." Hedge funds also had to cut "still-profitable positions" in swap and government bond markets in order to raise collateral for their money-losing positions with bank counterparties.

Because "everyone tried to do this at once," and because several of the biggest players were using similar proprietary trading strategies (specifically, arbitrage trades and convergence trades in fixed income, equity and fx markets)--which meant they all were trying to exit similar positions--market liquidity evaporated, prices gapped and spreads spread. Big losses resulted.

Next, the critical analysis and alternative interpretation of these events.

A Different Possible View of the Role of VaR: Bystander

Several points run counter to Dunbar's argument and create the possibility of a different interpretation of these events. Here's the critical analysis:

  • Contrary to Dunbar's statements, Basle capital requirements apply not at the trading desk level, but only at the top level of the firm and only to commercial banks, not investment banks or hedge funds.
  • Even for banks under Basle market risk rules, the rules say nothing about the relation between VaR limits and capital requirements; they establish requirements between the 60-day average aggregate VaR and capital.
  • It is my belief (I can't produce evidence in support of this) that in 1998 VaR limit structures were not very detailed at investment banks or hedge funds. That is, trading desk and book level VaR limits were not very common (especially at hedge funds), so it is not likely that "exit" trades were being induced by breaches of VaR limits, as Dunbar states. [Many commercial banks in contrast had very detailed VaR limit structures even in 1998.]

Given these points, it is possible to tell a different story without reference to VaR. Here's the proposed alternative: Hedge fund losses caused the banks to fear credit losses from their hedge fund counterparties. Hedge funds therefore had to exit trades so they could raise cash needed to post collateral demanded by their counterparty banks. The resulting exit trades moved market prices, because the positions were large. Those market moves caused investment banks and others to also exit their similar trading positions, because they were trying to avoid losses. The combined effects of these attempts to exit positions resulted in the crisis.

RELATED RESOURCES YOU MAY FIND INTERESTING: