Orange County Bankruptcy 1994: When a Treasurer Bet Against the Fed

On the afternoon of Tuesday, 6 December 1994, lawyers for the County of Orange, California, walked into the federal bankruptcy court in Santa Ana and filed a Chapter 9 petition on behalf of a county with a triple-A credit rating, a $3.7 billion annual budget, and per-capita income 18 percent above the California median. The filing listed liabilities of roughly $2 billion and unwound losses on a county investment pool that only four months earlier had been described to supervisors as worth $7.4 billion. By 8 December, with Credit Suisse First Boston preparing to seize and dump the pool's collateral, the realised loss reached $1.7 billion. It was the largest municipal bankruptcy in United States history — a record that stood until Detroit's filing in July 2013.

Orange County was not a distressed industrial city. It was the suburban tax base south of Los Angeles, a Republican stronghold of 2.6 million people, home to Disneyland, to large aerospace contractors, and to a white-collar workforce that commuted into greater LA. The county had no fiscal emergency, no revenue shortfall, no underfunded pension crisis of the kind that would later push Detroit under. What it had was a single elected official — Treasurer-Tax Collector Robert Citron — who had turned the county's cash-management operation into the largest leveraged bond fund ever run by an American local government, and a Federal Reserve chairman who in February 1994 began raising interest rates against a book engineered to profit only if rates fell.

Twenty-four years of the 9 percent pool

Robert Lafee Citron was first elected Treasurer-Tax Collector in 1970. The post in California is constitutionally independent: the treasurer reports to no supervisor, sets investment policy within broad statutory limits, and stands for re-election every four years. Citron, a Democrat in a county that had voted for every Republican presidential candidate since 1936, was re-elected five times. The reason was straightforward — his pool consistently returned more than the state-run Local Agency Investment Fund, and by the early 1990s it was beating that benchmark by roughly 200 basis points a year.

The Orange County Investment Pool was the commingled cash account for the county general fund, for the 31 cities inside the county that chose to deposit their operating cash with him, for the 30-plus school districts, for community-college districts, water districts, sanitation districts, and transportation agencies — 187 participating entities by late 1994 (Baldassare, 1998). Of the roughly $7.5 billion of participant capital, the county itself contributed about 30 percent; the balance was other people's money. School districts and special districts made up the majority, and many had issued short-term tax and revenue anticipation notes specifically in order to deposit the proceeds with Citron, borrowing at 4 percent to earn 9 percent from his pool. That carry trade — municipal entities borrowing to invest with a county treasurer — would later be scrutinised as a warning sign that went unheeded by the county's own board of supervisors and by the credit-rating agencies that stamped the note issues investment grade.

Citron's edge was not skill in security selection. It was structure. By 1993 he had leveraged the $7.5 billion of pool capital into roughly $20.5 billion of gross bond positions, financed through reverse repurchase agreements with a small group of Wall Street dealers. Merrill Lynch was by a wide margin the largest counterparty, providing financing and selling the pool a steady flow of structured notes — inverse floaters, index amortising notes, constant-maturity Treasury floaters, collateralised mortgage obligation residuals — that paid above-market coupons as long as short rates stayed low or fell. Merrill's five-year run of structured-note sales to Citron has been estimated at roughly $13 billion of face value, earning the firm approximately $100 million in fee revenue from the account (Jorion, 1995). The two Merrill bankers most associated with the relationship were Michael Stamenson, a senior institutional salesman based in San Francisco, and Chriss Varelas, who worked on the structured-note desk in New York.

The anatomy of the bet

Reverse repo is ordinary plumbing in fixed income. The pool posts a bond to a dealer, receives cash, and agrees to buy the bond back a short time later at a slightly higher price — the difference being the financing rate. Citron used the cash he raised this way to buy more bonds, which he then posted as collateral to raise more cash, which he used to buy yet more bonds. By late 1993 the pool's leverage ratio stood at roughly 2.7 to 1. The strategy worked as long as the yield curve sloped upward (short-term financing cheaper than long-term bond coupons) and as long as rates did not rise.

The second layer was derivative-embedded. Rather than holding plain Treasury or agency paper, Citron concentrated the pool in instruments whose coupons moved inversely to short-term rates. An inverse floater pays, for example, 10 percent minus LIBOR. When LIBOR falls, the coupon rises; when LIBOR rises, the coupon falls and can reach zero. Index amortising notes lengthen in duration as rates rise, locking investors into low coupons exactly when they would most prefer to be out. The pool's effective duration — how much its value would fall for a one-percentage-point move in rates — was close to 7.4 years at the end of 1993, more than twice what a prudent short-term cash-management fund would carry, and the leverage amplified that duration roughly threefold on an economic basis (Miller and Ross, 1997).

The composition of the $20.5 billion portfolio at its peak, reconstructed from the county's post-bankruptcy disclosures and a subsequent academic post-mortem (Jorion, 1995), looked like this:

Citron was, functionally, running a hedge fund inside a county treasurer's office — one with no investment committee oversight, no independent risk manager, no mark-to-market reporting to the participants, and no leverage cap other than what his dealers were willing to extend. The pool's own quarterly reports carried securities at amortised cost, meaning that the supervisors reading them saw a book value that did not move with the market. When Citron told the board in mid-1994 that he did not believe the pool had material losses, he was technically correct under the accounting framework his office chose to apply.

Greenspan's turn

On 4 February 1994 the Federal Open Market Committee raised the federal funds target rate by 25 basis points to 3.25 percent, ending a three-year easing cycle and surprising a market that had priced no tightening at all for the first half of the year. Alan Greenspan's Humphrey-Hawkins testimony later that month framed the move as pre-emptive — "it was prudent to move toward a less accommodative stance in order to sustain and enhance economic expansion" — but the market read it as a regime shift. Six further hikes followed, including a 75-basis-point move on 15 November, taking the target from 3.00 percent at the start of the year to 5.50 percent by November. The two-year Treasury yield rose from 4.25 percent to above 7.7 percent across the same nine months. For a levered inverse-floater book with duration of seven years, each 25-basis-point move cost the pool roughly $350 million of mark-to-market value.

Citron's response was to double down. In spring and summer 1994, rather than de-risking, he purchased additional inverse floaters on the theory that rates would soon peak and reverse. A Wall Street Journal piece that September quoted a Merrill Lynch internal memo noting that Citron's positions had grown by roughly $3 billion since January and that the county's pool remained "by a substantial margin" the largest single buyer of inverse floaters in the municipal market. A Merrill compliance review the same summer flagged concentration concerns (Weiss, 1994). No action followed.

Moorlach's warning

In the June 1994 primary election for the treasurer's office, a Costa Mesa accountant named John Moorlach ran against Citron on a single issue — the risk of the investment pool. Moorlach studied the pool's published holdings, consulted fixed-income professionals, and campaigned on the argument that the pool's leverage and derivative concentration exposed participants to catastrophic losses if rates rose. "Mr Citron is borrowing short and lending long on a massive scale," Moorlach told the Orange County Register in May 1994, "and the entire structure will come apart if interest rates rise even one percentage point." Citron, who refused to debate, won 61 to 39 on the strength of incumbency and the pool's still-reported returns. Moorlach's warning was printed verbatim in a campaign flyer and ignored by the county establishment.

By October 1994, an internal audit commissioned by the county administrator revealed a mark-to-market loss of approximately $1.5 billion on the pool's holdings. The figure was not disclosed publicly. Citron briefed the board of supervisors in closed session and insisted that the losses were paper only, that the securities would mature at par, and that the pool had sufficient liquidity to hold through the storm. On 1 December 1994, Moorlach — now treasurer-elect following Citron's December 1 announcement that he would resign — held a press conference warning of imminent collapse. Short-term lenders immediately refused to roll reverse-repo financing. Credit Suisse First Boston, which held roughly $2 billion of pool collateral against its repo loans, demanded additional margin that the county could not post.

The filing and the cram-down

On 6 December 1994, Orange County filed Chapter 9. It was the first time a county of its size had invoked the municipal bankruptcy code, and the legal machinery was poorly tested — Chapter 9 had been used in decades prior almost exclusively for small utility districts and one mid-sized city (Bridgeport, Connecticut, in 1991, dismissed for cause). Two days later, with no bankruptcy stay capable of protecting the pool's repo collateral (derivatives and repo had already been carved out of automatic-stay protection under 1984 amendments to the Bankruptcy Code), CSFB liquidated. Other dealers followed. The pool lost a further $600 million in two days of forced selling into a market that knew exactly who the seller was. Total realised loss: $1.7 billion, or 23 percent of pool capital. Losses were allocated to participants pro-rata, with the county taking roughly $500 million of the hit and the 186 other municipal entities absorbing the remainder.

Citron resigned on 4 December. He was indicted on 27 April 1995 and pleaded guilty on 19 May 1995 to six felony counts including misappropriation of funds, filing false and misleading financial statements, and making false certifications on pool holdings. In a post-plea interview with the Orange County Register, he told the paper, "I was an investment novice. I relied heavily on the advice of market professionals. In retrospect, I wish I had more training in complex government securities." He was sentenced on 17 November 1996 to a $100,000 fine, one year in a work-release programme, and five years of probation. He served no prison time and died of cancer in 2013.

The recovery plan — the so-called Chapman plan, named after the court-appointed advisor Bruce Chapman — went through several iterations. A June 1995 ballot measure that would have raised the local sales tax by 0.5 percent to repay creditors was defeated 61 to 39, in a rebuke of county leadership as sharp as the original vote returning Citron to office. A cram-down confirmation plan was eventually approved in June 1996, combining cost cuts, asset sales, bond refinancing through a $880 million recovery bond issue, and a litigation recovery programme that targeted Wall Street counterparties. Orange County exited bankruptcy on 12 June 1996.

The suits and the settlements

The county sued Merrill Lynch in January 1995 for $2.4 billion, alleging that the firm had knowingly sold unsuitable derivatives to an unsophisticated buyer and had aided Citron in concealing losses. Merrill denied liability. After three and a half years of litigation and the depositions of Stamenson and Varelas, Merrill Lynch settled with Orange County for $437 million in June 1998 without admitting wrongdoing — the largest settlement ever paid by a brokerage to a municipal client. A parallel SEC administrative proceeding concluded in August 1998 with a $2 million fine against Merrill and no findings of fraud. Stamenson and Varelas faced no personal sanctions from the SEC. Separate settlements were reached with CSFB, Morgan Stanley, Nomura, and several other dealers, bringing total recoveries to approximately $875 million against the $1.7 billion loss.

The structural consequences reached further than the settlement cheques. The Governmental Accounting Standards Board issued Statement No. 31 in March 1997, requiring public investment pools to report holdings at fair value rather than amortised cost — directly addressing the reporting fiction that had allowed Citron to assure supervisors for months that paper losses were irrelevant. California's Government Code was amended to cap reverse-repo exposure at 20 percent of pool assets and to prohibit public agencies from purchasing inverse floaters, structured notes with coupons keyed to inverse formulas, and range notes. The Securities Industry Association revised its suitability guidelines for institutional municipal clients. Many other states followed with parallel restrictions.

Orange County was not the only 1994 casualty of the Fed's tightening cycle. The same rate move produced the Mexican peso crisis that followed in December 1994 and the bond-market rout that caught Procter & Gamble, Gibson Greetings, and a range of other corporate and municipal buyers of structured derivatives. The deeper pattern — leverage built against a narrow directional bet, enabled by permissive counterparties, concealed by marked-to-myth accounting — would recur in larger form when Long-Term Capital Management unwound in 1998 and on a systemic scale in 2008.

What Orange County was

The Orange County failure was not primarily a story of fraud, though fraud was charged and admitted. It was a story of agency costs in the public sector, of a narrow fiduciary concept treated as a permission to run duration risk for constituents who did not know they were exposed. The pool's participants — the school boards, water districts, and transit agencies — had been "complicit bystanders" (Baldassare, 1998): they had asked no questions about how Citron produced 200 basis points of excess return in a 5 percent environment, and they had written policies that let him keep producing it on their collateral. Citron's own testimony supported that reading. "Nobody ever asked me how I was doing it," he told investigators in 1995. "They only ever asked what the yield was."

Against that background, the Fed's tightening was not the cause of the failure. A portfolio that cannot survive a 250-basis-point move in the federal funds rate over nine months was not a portfolio that a county treasurer should have been running in the first place, and the post-mortem literature is unanimous that Citron's strategy would have failed against any number of paths that did not happen to take short rates from 3 to 5.5 percent in 1994. The Volcker disinflation of 1979–1982 had already shown what a policy-driven rate cycle could do to a levered book, and the savings-and-loan debacle of the 1980s had taught the same lesson about borrowing short to lend long in regulated public-sector balance sheets. The striking thing about Orange County was that the lesson had to be taught again, in the same country, within fifteen years.

Today, every municipal investment pool in the United States reports at fair value, publishes a weighted-average maturity, discloses duration, and is constrained by statute from the specific instruments that ate the Orange County pool. The 1994 failure is the reason those rules exist. Robert Citron's last public statement, made to a reporter at his Santa Ana home in the late 1990s, was a single sentence of the sort that bankrupt gamblers sometimes produce when the distance is finally long enough. "I thought I knew what I was doing," he said. "I didn't."