Kidder Peabody Reported Profits Its Own Software Invented
A bond trader generated hundreds of millions in reported profits that did not exist, exploiting how the firm's accounting system handled forward settling Treasury trades.
The Instrument
Treasury strips are created by separating a bond's interest payments from its principal, so each cash flow trades as its own zero coupon security. Reconstitution is the reverse, combining the pieces back into a whole bond.
These reconstitution trades could be entered as forward transactions, settling at a future date rather than immediately.
The System Flaw
The firm's accounting system valued a forward reconstitution trade by comparing the price of the strips to the price of the resulting bond. Because a forward settling transaction has a different present value than an immediate one, the system recorded a profit at the moment the trade was entered.
That profit was not economically real. It reflected the time value difference between now and settlement, which would decay to zero as the settlement date approached. The trade earned nothing.
Each trade booked an immediate profit that would erode to nothing by settlement. Rolling it forward before that happened booked the profit again.
The Loop
The trader, Joseph Jett, entered these trades in enormous volume and continuously rolled them forward before settlement. Each roll generated a fresh recorded profit while the previous one had not yet fully decayed.
Sustaining reported profits required ever larger volumes, because the decay on the existing book had to be outrun. Reported positions grew to very large notional amounts, and the firm's balance sheet reflected billions in Treasury positions supporting profits that did not exist.
When the pattern was identified in 1994, the firm reversed reported profits of roughly 350 million dollars. The parent company, General Electric, took a substantial charge and subsequently sold the business.
The Dispute About Intent
The case remains genuinely contested on the question of intent. Internal investigation concluded the trader had deliberately exploited a known flaw. He maintained that he traded within the firm's systems, that his profits were calculated by the firm's own software, that supervisors reviewed and approved his activity, and that he was made responsible for an institutional failure.
Regulatory proceedings produced a mixed outcome. He was found liable on a books and records charge while some more serious allegations did not succeed.
The ambiguity is itself instructive. Where a system generates false profits and a trader maximizes them, separating exploitation from participation in an institutional error is genuinely difficult.
What Should Have Caught It
The controls that would have prevented it are conceptual rather than technical. A strategy reporting large profits should have an explainable economic source. Reconstituting Treasury strips is close to a mechanical arbitrage and should generate small, competitive returns, not extraordinary ones.
Independent validation of pricing models, performed outside the trading function, is the specific control. Someone whose job is not to generate profit should verify that the system's valuation reflects economics rather than a modelling convention.
The Bottom Line
Kidder's profits were a decaying time value effect recorded as income and regenerated by rolling trades forward. If nobody can explain where a profit economically comes from, the system rather than the market may be producing it.