The legal question that defined the unwinding of Bernard L. Madoff Investment Securities LLC ("BLMIS") was deceptively narrow. Under the Securities Investor Protection Act of 1970 ("SIPA"), 15 U.S.C. §§ 78aaa et seq., the trustee of a failed broker-dealer must calculate, and the Securities Investor Protection Corporation ("SIPC") must in part advance, each customer's "net equity" claim. When the broker-dealer in question never executed a single trade, what does "net equity" mean? That is the question Chief Judge Dennis Jacobs, writing for a panel that included Judges Reena Raggi and Denny Chin, answered in In re Bernard L. Madoff Investment Securities LLC, 654 F.3d 229 (2d Cir. 2011), cert. denied sub nom. Velvel v. Picard, 567 U.S. 934 (2012). The answer — dollars in, dollars out — is the doctrinal centerpiece of the Madoff aftermath and the most technically lawyerly holding in the entire affair.
The statutory question
SIPA was enacted in 1970 to do for brokerage customers something analogous to what the FDIC does for bank depositors: provide a limited federal backstop, funded by industry assessments, when a broker-dealer fails and customer property goes missing. SIPC administers the fund and, under 15 U.S.C. § 78fff-3(a), advances up to $500,000 per customer (no more than $250,000 in cash) to satisfy each customer's "net equity" claim. Net equity itself is defined at 15 U.S.C. § 78lll(11) as the dollar amount the broker would owe the customer "had the broker-dealer liquidated, by sale or purchase on the filing date, all securities positions of such customer," less any indebtedness of the customer to the broker. The distribution machinery in 15 U.S.C. § 78fff-2(c)(1) then directs the trustee to allocate the fund of "customer property" pro rata, based on those net equity claims.
The statute presupposes the existence of positions. BLMIS had none. The aggregate paper balances shown on the November 2008 customer statements totaled approximately $64.8 billion in fictitious securities. The principal that customers had actually wired in and never gotten back — the figure Irving H. Picard, the SIPA trustee, derived from BLMIS's own bank records — was approximately $17.5 billion. (Madoff's day-of-arrest estimate of $50 billion in losses, the figure that circulated in the press, was neither; the totals and their provenance are catalogued in the opening chapter of this volume (see [sec-oig-509-madoff-investigation]).) Between $64.8 billion and $17.5 billion lay roughly $47 billion in phantom gains, and the question of who was entitled to share in the recovered customer property turned entirely on which number the trustee was permitted to use.
Two constructions, two distributions
Two readings of § 78lll(11) presented themselves. Under the last-statement method, urged by an articulate group of long-tenured customers, the trustee was to honor the balances shown on each customer's most recent BLMIS statement — November 30, 2008 — as the measure of net equity. The statement reflected a regulated broker-dealer's own books and records; customers, the argument went, had every legal right to rely on them, and SIPA was designed precisely to vindicate that reliance when the broker collapsed. Under the cash-in/cash-out method — also called the Net Investment Method — urged by Picard and SIPC, each customer's claim equaled total principal deposited minus total amounts withdrawn, with no credit for paper profits.
The stakes ran in two directions. On the inbound side, the last-statement method would have produced roughly $64.8 billion in allowed customer claims and exhausted the customer-property fund several times over; the cash-in/cash-out method capped the universe at the $17.5 billion principal figure. On the outbound side — and this is the link that runs forward into the next chapter (see [sepinwall-righting-others-wrongs]) — the choice of method determined who counted as a net winner (someone whose withdrawals exceeded her deposits) and who as a net loser. Only net losers had allowed claims under the cash-in/cash-out approach, and net winners were not merely shut out of the fund; they were exposed to clawback litigation under 15 U.S.C. § 78fff-2(c)(3) and 11 U.S.C. §§ 547, 548, and 550 for the fictitious profits they had already received.
Chief Judge Jacobs's reasoning
Jacobs's opinion is a textual reading anchored to economic common sense. The "securities positions" referenced in § 78lll(11), he held, must be real. The trustee cannot "liquidate" by sale or purchase what was never owned, and to honor the November statements would require SIPC to pay claims keyed to invented trades at invented prices — including, the court noted, prices outside the day's actual high-low range, a hallmark of Madoff's back-dating. SIPA, Jacobs wrote, "does not require the Trustee to aggravate the injuries caused by Madoff's fraud." BLMIS, 654 F.3d at 235.
The opinion then dispatches the customers' principal counter-arguments. The plain language argument — that "securities positions" must include whatever the broker has booked — fails because the books were lies, and SIPA elsewhere distinguishes between securities the broker actually holds for the customer and mere bookkeeping entries. The legitimate expectations argument fails because, in Jacobs's much-quoted formulation, "an investor cannot expect a return on phantom securities that were never purchased." Id. A subjective expectation, however reasonable, does not conjure a real share certificate. The industry custom argument — that earlier SIPA liquidations had relied on statement balances — fails because no earlier SIPA liquidation involved an enterprise whose entire trading history was fabricated; the closest precedent, In re New Times Securities Services, Inc., 371 F.3d 68 (2d Cir. 2004), had distinguished real (if misrecorded) positions from wholly fictitious ones and pointed in the same direction Jacobs now took.
Equity, windfalls, and the shadow of fraudulent transfer law
The most consequential passages situate the cash-in/cash-out method within a broader bankruptcy tradition. In Ponzi-scheme bankruptcies generally, courts compute net loss precisely the way Picard proposed, because the alternative produces what the cases call the "windfall problem": late entrants and statement-rich holders would be paid in full for fictitious gains, while early-and-out customers — those who happened to redeem before the music stopped — would keep gains funded entirely by other victims' principal. The cash-in/cash-out method aligns customer-property distribution with the doctrine, traced to Cunningham v. Brown, 265 U.S. 1 (1924) — the original Ponzi case — that in a fraudulent enterprise, equity is measured in dollars in and dollars out.
That alignment matters because SIPA does not stand alone. Section 78fff-2(c)(3) explicitly empowers the trustee to invoke the avoidance powers of Bankruptcy Code §§ 544, 547, 548, and 550, and under § 548(a)(1)(A), Ponzi-scheme transfers are routinely treated under a Ponzi-scheme presumption as carrying actual intent to defraud. Fictitious profits paid to net winners are therefore avoidable as constructively fraudulent transfers; only the return of principal is protected by the § 548(c) good-faith-for-value defense. The largest exposure under that regime ran, naturally, through the feeder funds — Fairfield Sentry Limited ("Sentry") alone held a customer claim measured in the billions, and the Sentry litigation occupies its own chapter in this volume (see [clauss-roncalli-weisang-risk-lessons]).
Doctrinal takeaway
The rule of In re BLMIS is this: where a broker-dealer's records are wholly fabricated, the SIPA trustee must measure each customer's "net equity" under 15 U.S.C. § 78lll(11) by reference to actual cash deposited and withdrawn — not to the balances shown on the customer's most recent statement. The deeper lesson for the law-student reader is that "investment" in fraud-recovery proceedings is an equitable rather than contractual concept: when the underlying transactions are fictitious, the measure of loss tracks the bank-account ledger, not the brokerage-statement ledger. The principle reaches beyond Madoff. The Stanford Group SIPC litigation and the line of cases descending from SIPC v. Barbour, 421 U.S. 412 (1975), have asked variants of the same question, and the answer Jacobs gave — that statutory text written for ordinary brokerage failures cannot be stretched to validate an entirely invented one — is now the controlling answer in the Second Circuit and the persuasive answer elsewhere.