The Federal Reserve pumped tens of billions of dollars into the banking system to keep financial markets afloat after the September 11 attacks.
By Martin Mayer
January 2002
Institutional Investor Magazine
It needn’t have been so expensive.
No disaster plan is perfect. No matter how meticulously it’s drafted, it can’t account for the unthinkable, as the financial community has come to understand all too well since September 11. But common to all such plans is a postrecovery phase called “lessons learned,” which in this instance is yielding a university-level, continuing education.
Some postmortems are easy. Clearly, it’s not a good idea for all of a company’s phone lines to run through the same location. Many firms in Lower Manhattan tried to spread their risk by using at least two telecommunications carriers, but that provided false security when, as was often the case, all the lines ran through the World Trade Center. They won’t make that mistake again.
But other lessons, concerning essential components of the national and global financial infrastructure, are more complicated and costly and have only begun to be sorted out. These go to the heart of how markets operate, how institutions move money and how regulatory bodies, particularly the Federal Reserve System, react during times of extreme crisis.
The textbook case concerns the Bank of New York Co. and how its problems affected the functioning of securities markets. BoNY, a leading securities custody bank and therefore a major cog in the clearing and settlement of trades, had based its primary and secondary processing centers within a few hundred yards of each other, in close proximity to Ground Zero; both had to be evacuated, and one site remained off-limits for days. The bank has learned the obvious lesson and taken steps to disperse its facilities and improve its backup capabilities. But bigger flaws in, and risks to, the interbank clearing infrastructure have been exposed.
BoNY’s processing troubles and settlement delays complicated the recovery of the financial markets as a whole, which were devastated by the destruction of Cantor Fitzgerald and several other interdealer government securities brokers based in the World Trade Center. Indeed, the all-important U.S. Treasuries market, which was highly dependent on the pricing data that Cantor Fitzgerald supplied, came perilously close to total breakdown. Exacerbating the state of alarm was the fact that two banks, BoNY and J.P. Morgan Chase & Co., dominate the custody business. What if both somehow went out of commission? It’s no longer unthinkable.
“The scary thing is that with only a little extra effort - say, a car bomb in front of Chase’s facilities in Brooklyn - the terrorists could have taken out the entire financial system,” asserts Peter Vinella, chief executive officer of New York-based capital markets consulting firm PVA International. “We were pretty lucky,” he says, adding that the government securities market was back in business within a couple of days only because of “unprecedented acts of cooperation. People agreed that they would not allow any manipulation of the market at that vulnerable time.”
Adds Cristóbal Conde, president and chief operating officer of SunGard Data Systems, which provided backup facilities to many Wall Street firms after the September 11 attacks: “The settlement issue got taken care of only because of an inordinate amount of trust among participants. That’s not something that you want to rely on permanently.”
Permanent solutions require effective policies, particularly at the Federal Reserve, the central bank and lender of last resort. Unfortunately, the policy lesson in this instance has been at best ignored and at worst flouted. Although the Fed proved once again that it is capable of riding to the rescue with crisis-averting infusions of liquidity - just as it did in the aftermath of the 1987 stock market crash and during the Asian and Russian collapses of 1997-'98 - it followed central banking orthodoxy that some critics say is far too rigid and risky.
When financial institutions on September 11 had trouble settling their accounts on the Federal Reserve Communications System, the electronic payments network known as Fed Wire, the Fed advanced $45 billion from the discount window to keep the nation’s money flowing. And in the following days, when BoNY owed other institutions as much as tens of billions of dollars because of its inability to match and settle trades, the Fed tided it over and market participants rested easier.
But critics say there was a flaw in the Fed’s response: its reinforcement of bilateral net settlement arrangements, which have long prevailed in its own Fed Wire as well as in over-the-counter derivatives and repurchase agreement markets. In bilateral settlement each pair of counterparties add up their purchases and sales, and at day’s end one pays the other to cover its net debit position.
On September 11 bilateral systems crashed, leaving long trails of incomplete and questioned trades that had to be investigated and fixed. These were the underlying reason for the Fed’s extraordinary intervention. Says consultant Vinella: “The Federal Reserve came in and said, ‘We’ll give you a fungible piece of paper that can be any bond you need it to be.’ That kept the market functioning somewhat normally, although not in the way it usually operates.”
The alternative approach - multilateral settlement - isn’t nearly so vulnerable, and it proved its mettle in the wake of the terrorist attacks. In multilateral settlement all participants transact continuously, as in a bilateral system. But at the end of the day, all transactions are pooled and each institution’s ledger is reduced to a single credit or debit, which is settled through a deposit.
Examples of multilateral systems are the commodities clearinghouses of Chicago, which functioned normally through the crisis; the Depository Trust & Clearing Corp.'s National Securities Clearing Corp., which settled the stock trades of September 10 on the normal schedule, three days later; and the New York Clearing House Association’s Chips network for large-value interbank payments, which completed all of its daily settlement routines reliably, though with some difficulty and delays.
To private sector advocates of multilateral clearing, the lesson could not be more obvious.
“Centralized or multilateral clearing facilities offer customers important protections compared with the bilateral counterparty agreements widely used in OTC derivatives markets,” contends Leo Melamed, chairman emeritus and senior policy adviser of the Chicago Mercantile Exchange. It’s a matter of shared interests and enforced trust, he says: “Centralized clearing implies that risks are effectively netted, mutualized and protected by a battery of financial safeguards not available to a party to a bilateral contract.”
Adds Jill Considine, formerly chief executive officer of the New York Clearing House, now CEO of New York-based DTCC, “Through multilateral netting we significantly reduce financial obligations requiring settlement, which in turn reduces systemic risk, lowers settlement costs and enhances the efficient use of capital.”
To be sure, the Fed’s assistance was necessary: Its support to banks and financial markets indirectly helped the independent multilateral systems function as normally as they did. But the experience showed bilateral systems to be dangerously burdensome to a lender of last resort.
That reality hit home two months after the World Trade Center horror, when Enron Corp. collapsed. The Houston-based energy giant ran its own unregulated exchanges and served as counterparty on immense volumes of transactions - a role that can require virtually unlimited credit resources to assure bilateral settlements. The minute that there was doubt about Enron’s ability to provide necessary liquidity, the house of cards came crashing down.
Yet the Fed - which, unlike an Enron, cannot fail and must regulate others’ risk-taking - remains wedded to bilateral settlement. And the Basel-based Bank for International Settlements, the “central bank of central banks,” has supported the Fed’s rationale, which stems from its history as a set of 12 district banks that essentially serve as correspondents for the large banks in their regions.
In their oversight of correspondent banking channels, Fed policymakers have long insisted that because the counterparties in bilateral nettings know each other, they can manage their account activities effectively without exceeding credit limits. Because the netting process results in a relatively small amount of money changing hands, the impact of any single failure is theoretically contained. And if there is a failure, the Fed can target its response.
The Fed believes that this approach lowers systemic risk. It gives commercial banks an incentive to participate in bilateral arrangements by lowering the capital requirements related to such trading activities. The Fed has also lobbied for legislative provisions that would give derivatives, repos and other claims in bilateral netting systems priority in a financial institution bankruptcy.
Fed officials acknowledge that Fed Wire, as an indispensable utility for transactions among correspondent banks, raises risk-management concerns. The number of big banks is shrinking through mergers - the global custody market is an extreme example of such consolidation - making the potential failure points ever more critical. But a high-ranking Fed officer says: “Concentration is driven by the business case. We assume that now the counterparties will be asking tougher questions.”
MULTILATERAL NETTING MAY HAVE EFFICIENCY on its side, but bilateralism has a longer history, rooted in manual methods of clearing and settlement. As recently as 30 years ago, messengers were still wandering around Wall Street carrying envelopes stuffed with cash or stock certificates to pay for and deliver securities that the customers of one broker had bought from or sold to the customers of another broker. At the banks, designated trustees sat at Rube Goldberg-like contraptions that permitted 64 stock certificates at a time to be legally signed to certify the transfer of ownership.
In the 1970s the securities industry, overburdened by a paper glut, began to dismantle those old ways by converting certificates to book entries, standardizing clearing procedures and establishing Depository Trust Co. to coordinate settlement. The system evolved into a prime example of risk-mitigating multilateral netting.
When Depository Trust & Clearing Corp. - now the parent of the merged Depository Trust and National Securities Clearing Corp. - handled a record $722 billion of securities on April 4, 2000, its members had to ante up only $21.7 billion, or 3 percent, to settle their accounts. Explains DTCC’s Considine: “The benefit of the clearing corporation is that firms are able to eliminate counterparty risk. We act as a central counterparty, guaranteeing to each side that their trades will be completed.”
Similarly, before high-speed data communications facilitated electronic funds transfers, banks collecting large checks written to their customers dispatched messengers to make direct presentment to the banks on which the checks were drawn. These “direct sends” ensured quick crediting through correspondent accounts.
But that was nothing compared to the speed of Fed Wire. Although the Federal Reserve began operating a rudimentary wire via telegraph early in the last century, electronic transfers truly took off when higher-speed computing and communications became available after World War II. And in 1970 the New York Clearing House launched Chips - the Clearing House Interbank Payments System - as an international, private sector alternative to Fed Wire. Chips set itself apart by championing multilateral settlement, whereas Fed Wire never abandoned bilateral. But in its early years, Chips did not have central guarantees in place to keep the system from collapsing in the type of financial crisis that until 1974 was as unforeseen as the terrorist attacks in 2001.
At the close of business on June 26, 1974, German authorities declared Bankhaus Herstatt insolvent. The American banking day was just getting started; Chase Manhattan Bank, Herstatt’s U.S. correspondent, took the entire $50 million in Herstatt’s account to cover a foreign exchange liability to Chase, stiffing other creditors that had claims on the German bank. It was bilateral netting run wild. Chips ground to a halt, and the inability to move money threatened to provoke a chain reaction of defaults. The network regained stability only when the Fed gave assurances that it would stand by Chips and would not permit settlement to fail.
The lesson: Absent the mediating influence of a strong central administration with the power to enforce group discipline, an institution will respond to a crisis the way Chase did - by first looking after its own interests.
Chips has long since taken precautions designed to absorb systemic shocks. Twenty years ago it instituted same-day settlement; until then payments weren’t final until the next day, which left banks vulnerable to failures that might occur between posting and settlement. In the 1990s Chips required participating banks to post collateral in the form of Treasury bills worth at least 1.5 percent of the maximum exposure the bank would have to the clearinghouse; this guaranteed that clearing could proceed even if the largest of the 58 participating banks failed to cover its debits.
A year ago Chips implemented real-time multilateral netting, eliminating intraday risk, with relatively little balancing to be done at the daily 6:30 p.m. close.
At Chips and DTCC, small amounts of liquidity go a long way. On an average day $1.2 trillion worth of payments, averaging $5 million each, are made through Chips, but because of multilateral netting, banks deposit only about $2.3 billion, or 0.2 percent, to settle their accounts. By comparison, on its best days bilateral Fed Wire achieves 5 percent settlement efficiency.
Computerization makes Chips’ efficiency and risk reduction possible. Chief operating officer John Mohr credits software written on Unisys computers by Robert Cotton, a mathematician and 20-year New York Clearing House veteran who adapted approaches originally developed in the 1980s to help medical schools vet applications for admission. “He taught our system to think complexly,” says Mohr. (On September 11 Mohr was attending a conference in Scottsdale, Arizona; he ran up nearly $1,000 in phone charges coordinating personnel movements between operations centers and holding conference calls with senior bank executives.)
The technology enables Chips to handle payments that pour in at a rate of ten to 15 a second, with some 300 waiting in a queue for processing at any given moment. But even though financial exposure is minimized, multilateral netting does not give banks license to transact willy-nilly. The system never extends credit, never permits a payment that exceeds the sum that the participating bank has in the system and never allows any participant to accumulate excess credits.
If Bank A orders a payment without sufficient funds in its account, the computer will locate another bank in the system that has ordered a payment to Bank A, executing that payment first to enable Bank A’s request to go forward. If the computer finds that a payment to Bank B will result in excess credit, it will search out a bank or banks to which Bank B has ordered a payment and move that payment to the head of the line to make sure liquidity will not sit idle in the system. The program can chase as many as 17 potential links to assure that no bank makes payments beyond its balance or accumulates payments beyond its preset credit limit.
As a result, says Mohr, “the clearinghouse offers liquidity-efficient risk reduction.”
IN THE SECURITIES WORLD, TRANSACTION settlements have been accelerated immensely by two innovations: the immobilization of stock certificates at DTCC and the creation of a nominee - Cede & Co. - that represents literally everybody. Holding securities with a market valuation of $23 trillion, Cede serves as nominee not for the beneficial owners of the securities but for the brokerage houses and institutions that are members of DTCC. The records of who buys, who sells and who owns what are kept by the brokerage houses, institutions and custodian banks.
The U.S. stock market settlement timetable of T+3 - three business days after the trade - applies whether the trade is accomplished at an exchange, over the counter, on an electronic communications network or internally at a broker-dealer. (If a trade is between two customers of the same broker or between the broker and his own customer, DTCC never hears about it; the broker internalizes, or transfers on its own books, the ownership of the claim against Cede & Co.) During and directly after the trading day, the clearing firms compare the records of what every broker-dealer reports was bought or sold; “compared” trades go to the computers of National Securities Clearing Corp., which guarantees both sides. NSCC stands in for both the seller who will deliver the securities to the buyer and the buyer who will pay the seller for them.
At this point the resources of the entire securities industry stand behind the trade - neither buyer nor seller need worry about the credit status of the other. September 10 was a normal trading day in the stock market, and three days later the trades settled with no major problems.
In contrast, the government securities market, which is supposed to settle overnight on a bilateral basis, was in shambles. Governments traders were cut off from the information resources of Cantor Fitzgerald and to a significant degree from their own banks, which could not calculate positions until they knew what other banks were doing to fulfill their obligations.
The Federal Reserve saved this market not only by its extensions of credit but also through an expanded bond-substitution program to support banks’ repurchase agreements. Nevertheless, settlement times had to be extended to five days - longer in some cases - simply because banks had difficulty delivering securities to each other.
It’s not generally understood how the banks accumulated the net debit positions that required the Fed’s $45 billion infusion on September 11. Fed Wire payments are instantaneous and considered final and irrevocable. In theory, banks making wire payments have sufficient funds in their Fed reserve accounts. In practice, banks no longer keep significant amounts of money at the Fed: They can count cash in ATMs as part of their reserves, and they can do repos with big customers to reduce the deposits against which reserves are calculated. In the end the larger banks’ accounts at the Fed are much less than what they need as transaction balances for the next day’s business.
To make up the difference and in effect to maintain its centrality in the payment system, the Fed extends daylight overdrafts in great quantity - tens of billions of dollars a day. This enables banks to make payments on Fed Wire. At the end of the day, 6:30 p.m., banks are expected to have acquired the funds to cover the overdrafts and reestablish their reserve positions, either from payments made to them or by borrowing Fed funds. On September 11 and 12, that money was not available; the system could not close until the clock approached midnight, when the Fed ponied up the necessary $45 billion. Later the Fed returned to its normal open market operations to make sure banks could settle their accounts without spikes in the Fed funds rate.
The Fed shows no signs of questioning the limits or costs of bilateral settlement. The central bank’s abiding faith in bilateralism extends to the continuous linked settlement system for the foreign exchange market that major banks and IBM Corp. have been building in London since 1993 (at a cost of $500 million) and hope to put into widespread service in 2003, six years behind schedule. The Federal Reserve Bank of New York will be the system’s lead regulator; given that it will operate like Fed Wire, as a bilateral mechanism, it is sure to require considerable extensions of credit.
A corollary problem of bilateral netting and internalized trading is information asymmetry - an advantage that accrues to the counterparties. When Merrill Lynch & Co., for example, crosses a customer trade inside its shop, it has more information about market behavior than the firms that lack knowledge of Merrill’s order flow. (This is why Nasdaq dealers are willing to pay brokers for orders, and why Enron fought in Congress - with unfortunate success, as it turned out - for a law that allowed it to operate its EnronOnline exchanges without supervision.) Banks that settle with each other directly share information others don’t see. But reduced transparency makes markets more susceptible to panic and crisis - an externality that regulators are supposed to account for when writing their rules for minimum capital and its allocation.
To say that market forces drive firms to bilateral netting, when the regulators actually encourage such behavior by their capital allocation rules, is disingenuous. The truth is that central banks remain highly secretive institutions that draw no small part of their authority from the general belief that they know things the markets don’t.
As the experience of September 11 suggests, and as Enron has spectacularly demonstrated, over-the-counter bilateral payments, financial instruments and derivatives contracts carry greater risks than exchange-traded, clearinghouse-settled payments, securities, derivatives, commodities and foreign exchange contracts. Large-scale clearing and settling through correspondent banking and other bilateral arrangements risks the creation of many more EnronOnlines and further major interventions by the lender of last resort.
The question now for the Fed and the BIS is whether they have the wisdom - or perhaps just the common sense - to reverse their positions and use their capital allocation rules to promote multilateral arrangements.
Whether in the Chicago options pits or in large interbank transfers of money, multilateralism works. As the Chicago Merc’s Melamed says: “Centralized clearing facilities do not accept credit. They require the payment and collection of daily mark-to-market variations and the imposition of margin requirements reflecting potential risk exposures.”
The Fed and the BIS aren’t convinced.
