What if LTCM had been allowed to fail in 1998?

On the morning of Wednesday, September 23, 1998, William McDonough, president of the New York Federal Reserve Bank, summoned the chief executives of fourteen major banks to the tenth floor of 33 Liberty Street in Manhattan. Long-Term Capital Management, the hedge fund founded by John Meriwether with Nobel laureates Myron Scholes and Robert Merton on the investment committee, had lost roughly 4.6 billion dollars over the prior four months. About 90 percent of its capital was gone. The fund still had over 100 billion dollars of positions on its books and notional derivatives exposure of roughly 1.25 trillion. Forced unwinding of those positions would, in McDonough's view, take at least one major bank with it.

The CEOs were not asked. They were told that the New York Fed had concluded a private rescue was preferable to bankruptcy and that the room would not adjourn until a deal was struck. By late afternoon, fourteen institutions had agreed to put up 3.625 billion dollars to recapitalize LTCM in exchange for 90 percent of its equity. Bear Stearns, which had cleared LTCM's trades but had no equity exposure, refused to participate. The other thirteen, plus Société Générale and Paribas joining slightly later, signed.

No Federal Reserve money went directly to LTCM. The Fed provided the room, the convening authority, and an implicit signal that it considered the alternative unacceptable. The positions were wound down through 1999 in orderly fashion. The consortium recovered most of its capital. In the contemporary press it was treated as a clean save. In retrospect, it set a precedent that ten years later proved much harder to manage.

The decision made

The Fed's intervention extended a doctrine that had previously applied only to insured commercial banks. LTCM was a private hedge fund with fewer than 200 investors, all of them sophisticated, none of them protected by deposit insurance. The argument for rescue rested on the systemic damage forced selling could cause to counterparties. That argument, once accepted in 1998, was available to every subsequent large leveraged investor whose unwinding might damage the banking system. Bear Stearns invoked something close to it in March 2008. Lehman Brothers tried in September 2008 and failed. AIG succeeded. The intellectual line runs back to McDonough's room.

The other consequence was that LTCM's specific risk model, which had spectacularly underestimated the correlation of relative-value spreads in a panic, was treated as an idiosyncratic failure rather than a systemic problem. The next decade of structured finance, including the mortgage-backed securities and synthetic CDOs that powered the 2008 crisis, used very similar assumptions about the independence of underlying risks. The lesson the survivors took from 1998 was that you could be wrong at very large scale and the system would catch you. They built on that.

The path not taken

Suppose McDonough had concluded that the systemic risk was overstated, that the 14 banks could absorb their direct losses, and that LTCM should file for bankruptcy. The fund's positions go into court-supervised liquidation under a special master. Spreads widen further for several weeks as the unwind happens, but most of LTCM's bets, when held to maturity, were money-good. The total loss to counterparties is probably in the 5 to 10 billion dollar range, distributed across the largest banks in the world.

One or two midsize Wall Street firms with the heaviest exposure probably do fail. Salomon Smith Barney, recently merged into Citigroup, was a likely candidate. Lehman Brothers, which had its own difficult fall in 1998, would have been stress-tested. The losses would have been visible, ugly, and widely covered. The October 1998 emergency Fed rate cut, which actually happened in response to LTCM and the Russia crisis, would have been larger and probably announced without the convening fiction.

The cascade

A failed LTCM in late 1998 produces several cascades the actual rescue prevented. First, the political appetite for hedge fund regulation becomes real. Brooksley Born, then chair of the Commodity Futures Trading Commission, had spent 1997 and 1998 arguing that over-the-counter derivatives needed federal oversight. She was overruled by Robert Rubin, Larry Summers, and Alan Greenspan in May 1998 in a now-notorious meeting. With LTCM in bankruptcy, her case becomes politically unkillable. Some version of derivatives clearing and reporting moves through Congress in 1999 or 2000, ten years before Dodd-Frank.

Second, the surviving large banks recalculate their counterparty risk and trim leverage to single-counterparty exposures. The growth of prime brokerage businesses through the 2000s is slower and more capital-intensive. Goldman Sachs, Morgan Stanley, and JPMorgan, all of whom had to write down significant LTCM-related exposure, build internal limits that the actual 2000s did not produce. The shadow banking system that grew up in the gap between regulated banks and unregulated funds is smaller. The 2008 crisis still happens, because the underlying mortgage credit dynamics are independent of LTCM, but it happens with more capital cushion and clearer rules about who eats the losses.

Third, the implicit guarantee that became explicit in October 2008 is harder to claim. A 1998 precedent of letting a large leveraged firm fail provides political cover for letting Bear Stearns or Lehman work through Chapter 11 in 2008. The actual response in 2008 was hampered by Treasury and Fed officials who genuinely believed that no Wall Street firm could be allowed to fail without taking the system with it. That belief had hardened in part because LTCM had shown what they thought a non-rescue would look like, while hiding what an actual non-rescue might have shown.

The modern echo

The 2021 Archegos collapse, the 2022 UK gilt crisis, and the March 2023 Silicon Valley Bank failure each turned on the same question McDonough faced in 1998. Is the systemic damage from a forced unwind larger than the moral hazard cost of preventing it? In each case the answer was that the firm in question would be wound down or rescued, and the precedent for that answer was set on a Wednesday afternoon in lower Manhattan in September 1998, when fourteen bank CEOs decided they would rather pay the bill than find out.