Orange County, California and Robert Citron

A treasurer, a pool, and two decades of trust

Robert Citron had been Orange County's elected treasurer for two decades by the time the crisis broke. By the early 1990s he ran the Orange County Investment Pool (OCIP), a fund that took deposits not just from the county itself but from around 200 other local public agencies nearby, cities, school districts and water districts that were required or chose to place their cash with him. For years the pool paid returns well above what a plain savings account or money-market fund offered, and depositors kept coming back for more.

Where the returns came from

The returns came from leverage, not skill. Citron borrowed short-term cash through reverse repurchase agreements, in effect a chain of short-dated loans collateralised by the pool's own securities, and used the proceeds to buy more securities: medium-term Treasury and agency notes, and structured notes called inverse floaters, whose coupon rises when short rates fall and falls when short rates rise. By 1993 and 1994, the pool's book value had grown to over $20.6 billion, against about $7.6 billion of actual participant deposits, a leverage ratio of roughly 2.7 times. The SEC later found that over this period Citron and his assistant treasurer had leveraged the pool to levels ranging from 158% to over 292%.

1994 turns against the position

This kind of strategy only works while rates stay low and stable, and 1993 obliged: the pool earned close to 8% for the year. But in February 1994 the Federal Reserve began a tightening cycle, raising the federal funds rate through the year from a monthly average of about 3.05% in January to about 5.45% by December, a move of roughly 240 basis points, and adding a final 75 basis-point rise in the discount rate on 15 November 1994. Every rate rise pushed down the market value of Citron's long-dated, rate-sensitive holdings. The damage built steadily through the year: month-end mark-to-market figures show the pool's reported paper loss growing from about $26 million (0.45% of the portfolio) in January to over $443 million (5.24%) by June. As collateral values fell, the banks and dealers on the other side of the reverse repos made collateral calls and cut back how much they would lend against the same securities, over $873 million in reductions and calls in the first half of 1994 alone.

A public warning, and a re-election anyway

The warning signs were public well before the crisis broke, which is what sharpens this case beyond a simple story of a bad bet. In May 1994, John Moorlach, a Republican challenger running against Citron for county treasurer, obtained Citron's portfolio holdings through a public-records request and circulated the list to bond dealers for their opinion. He later said it was the worst portfolio they had ever seen. It made no difference at the ballot box: on 8 June 1994, Citron was re-elected for a seventh term.

Disclosure, resignation, bankruptcy

The reckoning came at the end of the year. On 1 December 1994, the county disclosed that the investment pool had suffered a paper loss of about $1.5 billion on the $20.6 billion portfolio. Citron resigned under pressure from the county's Board of Supervisors on around 4 December, before the new term he had just won even began. On 6 December 1994, Orange County and the investment pool each filed for Chapter 9 bankruptcy, the largest municipal bankruptcy in US history at the time. Between mid-December 1994 and 20 January 1995 the county was forced to liquidate the pool's holdings, in a market that knew exactly why it was selling. The participants' final, realised loss came to about $1.7 billion on their $7.6 billion of deposits, roughly 22.3%.

The aftermath

The aftermath ran on for years. Voters rejected a proposed sales-tax rise meant to help fund recovery in June 1995, so the county instead financed its exit from bankruptcy through a bond issue of about $880 million in June 1996, formally emerging from Chapter 9 that same month. The Securities and Exchange Commission filed civil and administrative actions in January 1996 over misleading disclosure in eleven municipal bond offerings worth over $2.1 billion. Citron pleaded guilty to six felony counts and was sentenced in November 1996 to a year served through work release and house arrest, plus a $100,000 fine. Merrill Lynch, the county's main securities dealer, settled the county's lawsuit against it for $400 million in June 1998 and paid a separate $2 million civil penalty to the SEC over its underwriting disclosures.

Two different loss figures appear in this case, $1.5 billion and $1.7 billion, and they describe two different moments rather than two conflicting stories. The $1.5 billion figure is the paper loss disclosed on 1 December 1994, before the portfolio was sold. The $1.7 billion figure is what participants actually lost once the county was forced to liquidate the whole book over the following seven weeks. Both numbers come from the same SEC finding of fact, and they sit close together for a simple reason: a forced sale in a market that already knows the seller is distressed rarely improves on the price that seller feared.

A forward rate is a breakeven that can be locked in today, not a forecast of what will actually happen, and Citron's strategy is the mirror image of that idea. By borrowing short through reverse repos and holding long-dated notes, he was implicitly short the forward rates embedded in the Treasury yield curve. The trade only paid off if realised future short rates stayed below what the curve's forwards already implied, in other words, a bet that the market's own breakeven was set too high. When the Federal Reserve raised rates in 1994, it did not prove the market wrong. It proved the forwards right, and Citron's position lost.

The second half of the story is leverage. Duration is what created the exposure to rates in the first place; a $20.6 billion book of medium-term notes would have lost value in 1994 whether or not it was leveraged. What leverage did was change the base the loss was measured against. Because the reverse-repo lenders stood ahead of the county's own deposits, they were repaid first and in full, out of collateral, while the loss landed entirely on the $7.6 billion of participant deposits that supported the whole $20.6 billion book. A market-value decline of only around 7% on the total portfolio became a loss of roughly a fifth of the county's own money. Government and agency paper carries close to no default risk, but it is not close to free of interest-rate risk once leveraged and mismatched in maturity like this.

The third mechanism is the one that turned a paper loss into an emergency. As the pool's securities fell in value, reverse-repo counterparties made collateral calls and reduced how much they would lend against the same collateral, an amount exceeding $873 million in the first half of 1994 alone. This is the same margin mechanic that recurs elsewhere in the course: a mark-to-market loss on a leveraged position becomes a demand for cash today, whether or not the position might recover later. Orange County did not have that cash on hand. The result was a rushed, forced liquidation in December 1994 and January 1995 that locked in a loss the paper marks had only threatened until then.