Feb 21

Amaranth - the updated case study

by David Harper, CFA, FRM, CIPM

FRM |

Last year's FRM exam assigned fully 13 learning outcomes to the Amaranth case study based on Ludwig Chincarini's December 2006 study. His study tried to infer the hedge fund's failed trades, but was hobbled by a lack of primary source data. In the meantime (June 2007), a US Senate Subcommittee, armed with subpoenas, published a report that detailed actual positions. Chincarini integrated the subcommittee's findings into his recently published "Lessons from the Collapse of Amaranth."

Calendar spread trade (long winter months + short non winter months)

The subcommittee findings basically confirmed Chincarini's original deduction that Amaranth was engaged in calendar spread trades:

"a general bet that winter natural gas prices would rise, while non-winter natural gas prices would decline, referred to as the long winter, short non-winter spread trade."

This calendar spread is remarkably simple: go long winter months (Jan,Mar) and short non-winter months (Nov,Apr). But on an epic scale.

However, Chincarini over-estimated NYMEX’s ability to limit Amaranth’s exposure. NYMEX was not aware of Amaranth's ICE positions. Indeed, in September even as Amaranth had very large NYMEX positions, they were deliberately moving "substantial positions" to the unregulated ICE.

Other highlights:

  • Although Amaranth had no formal stop losses or concentration limits, they had a robust risk system including: a risk manager for each trading book, daily reports (position, P&L, leverage, concentrations, premium at risk, industry exposures, VaR, stress reports and liquidity reports). The stress reports included scenarios: credit spreads increase, volatility contracts jump, yield curve shifts.
  • The Subcommittee report asserts Amaranth was contributing to artificial (i.e., non-fundamental) spreads throughout the summer and the price decline itself in September. Apparently, Amaranth was violating position limits throughout 2006, basically ignoring warnings, and were headed for trouble even in the summer of 2006 (when they had 50-80% of the open interest on futures contracts for many months!). There basically were the market and they couldn’t reduce even that summer without big losses.

Could any model have captured the loss potential?

Chincarini's original deduction was good here, too: where VaR (market risk) could not explain the massive losses, liquidity risk filled much of the gap.

His analysis classifies:

  • Market risk (quantified by VaR): volatility of returns
  • Liquidity risk (hard to quantify): degree of difficulty in exiting a given trading position, and
  • Funding risk: ability of Amaranth to meet margin calls on natural gas positions

In theory, VaR addresses the market risk. Chincarini figures a leveraged VaR model would have estimated about $1.4 billion loss at 99% confidence. But actual losses were more like $3.3 billion (assuming constant positions; their cumulative total loss was nearer to $6 billion!). On the one hand, this is a reminder that a 99% VaR is just where the extreme tail starts! But on the other hand, this implies a five standard deviation collapse, which impugns the VaR specification. In short, my opinion is that Amaranth is a fine example for either a supporter or a critic of VaR.

He shows that liquidity risk, while difficult to quantify, rivaled market risk (given they had 50% to 100% of the open interest). Calling it liquidity risk doesn’t do it dramatic justice, they were facing margin calls and free fall - gravity starting to take over at the start of September 2006.

And regarding funding risk, "even had Amaranth’s trade had been logical from a VaR perspective and a liquidity perspective, it would have not been logical or prudent from a funding risk perspective."

The morale of the story (aside from the need for inter-exchange transparency that paves the way for meaningful regulation) looks to be that VaR needs a liquidity risk complement and a stress testing complement.

Comments

  1. Be the first to leave a comment!

Leave a Comment