The Asian Financial Crisis of 1997: Miracle, Panic, and the Argument That Followed

Eighteen Months

In the middle of 1997, East Asia was the reference case for successful late development. Thailand, Malaysia, Indonesia, and South Korea had compiled decades of high growth, high savings, high investment, rising literacy, and — by the standards of the developing world — modest inflation and disciplined budgets. The World Bank had published The East Asian Miracle in 1993 to explain the record. Capital was flowing in on the strength of it.

By the end of 1998, Thailand, Indonesia, and Korea were in IMF programs; the Indonesian rupiah had lost most of its value; Suharto, after thirty-two years in power, had resigned on 21 May 1998; Malaysia had walked out of the international capital market by imposing controls; and the phrase “Asian model” had gone from an object of study to a term of accusation. Output contracted sharply across the region in 1998. Millions of households that had crossed out of poverty crossed back.

Nothing in the standard crisis playbook of the era anticipated this. These were not countries with runaway budget deficits, hyperinflation, or the fiscal pathologies that had produced the Latin American debt crises of the 1980s. Their governments were, for the most part, in surplus. Whatever went wrong went wrong somewhere the existing models were not looking.

What Actually Broke

The immediate trigger was the Thai baht. Thailand had run a de facto peg to the U.S. dollar, and the peg had done exactly what pegs do when they are credible: it lowered the perceived risk of borrowing in dollars, which encouraged domestic banks and firms to do so, which financed a boom in property and industrial capacity, which produced a current account deficit and an increasingly fragile balance sheet behind it.

When the boom cooled and speculators began testing the peg, the Bank of Thailand defended it — spending reserves, and, importantly, committing reserves it had not yet spent through forward contracts that did not appear in the headline reserve figures. When the true position became clear, the defense was over. On 2 July 1997 Thailand announced a managed float. The baht fell immediately and kept falling.

What made this a regional crisis rather than a Thai one was the structure it exposed elsewhere. Across the region, the same configuration recurred in different proportions:

A currency mismatch. Borrowing was denominated in dollars; revenue was denominated in local currency. A devaluation therefore did not merely change relative prices — it enlarged the domestic-currency value of the debt directly. Firms that were solvent at 25 baht to the dollar were insolvent at 50.

A maturity mismatch. Much of the foreign borrowing was short-term interbank credit, rolled over continuously. Long-lived assets — factories, office towers, power plants — were funded by liabilities that could refuse to renew on thirty days’ notice.

A thin supervisory layer. Financial liberalization had opened the channels for capital to arrive before the prudential machinery existed to ask what it was funding. Korea’s merchant banks and Thailand’s finance companies were, in effect, the crisis’s transmission mechanism.

Jargon note: economists call the combination of borrowing foreign and short a double mismatch. Barry Eichengreen and Ricardo Hausmann later gave the first half of it a memorable and contested name — original sin — the observation that most emerging economies cannot borrow abroad in their own currency, so that any external borrowing carries an exchange-rate risk that cannot be hedged away in aggregate. A devaluation that would be expansionary for a country with domestic-currency debt is contractionary for a country without it.

The relevant vulnerability indicator, in hindsight, was not the budget balance or even the current account. It was the ratio of short-term foreign-currency liabilities to usable reserves — the arithmetic of whether a country can honor its obligations if nobody rolls over. On that measure Korea, which had almost none of the fiscal or inflationary problems normally associated with crisis countries, was among the most exposed economies in the world.

The Programs, and the Fight About Them

The IMF assembled the largest emergency packages in its history to that point, in quick succession: Thailand in August 1997, Indonesia at the end of October, and Korea in early December. Reported headline totals vary by source depending on whether one counts only Fund money or the full assembled package including the World Bank, the Asian Development Bank, and bilateral commitments — roughly $17 billion for Thailand, on the order of $40 billion for Indonesia, and a Korean package announced in the $55–58 billion range, the largest the Fund had ever put together. A useful discipline when reading these figures is to notice how much of a “package” was actually disbursed versus pledged; the announced number was doing part of its work as a signal.

The conditions attached to the money became the most contested element of the entire episode. In outline, the early programs called for higher interest rates to stabilize the currency, fiscal tightening, and rapid financial-sector restructuring including the closure of insolvent institutions.

The case for this package was not stupid. High rates were meant to make holding the local currency attractive enough to stop the run; closing insolvent banks was meant to stop the bleeding and re-establish that lending decisions had consequences; fiscal restraint was meant to reassure creditors that the government would not monetize the cost of the bailout.

The case against became, over the following years, close to a consensus on several points:

The fiscal condition was misdiagnosed. These were not countries in fiscal crisis. Contracting the budget in the middle of a collapse in private demand deepened the recession, and the Fund itself relaxed the fiscal targets during 1998 as this became evident.

High interest rates cut both ways. In an economy where the binding problem is corporate insolvency rather than currency speculation, very high rates can accelerate the bankruptcies that are frightening creditors in the first place, weakening rather than strengthening the currency. Whether rates helped or hurt in each specific country remains genuinely disputed; that it was a trade-off rather than a free lever is not.

The bank closures were handled badly. The abrupt closure of a set of Indonesian banks in late 1997 without a clear, credible deposit guarantee is widely blamed — including in the Fund’s own later self-assessments — for triggering runs on institutions that had been solvent, converting a currency crisis into a banking panic.

Jargon note: conditionality is the set of policy commitments a borrowing government makes in exchange for IMF financing. The rationale is that the Fund lends into situations where lending without policy change would simply fund the same behavior again — a moral hazard argument. The counter-argument is not that conditionality is illegitimate but that the content of it in 1997–98 extended well past what the balance-of-payments problem required, into corporate governance, trade policy, and domestic industrial arrangements that had little to do with the crisis and a great deal to do with a broader agenda. Readers should compare our reconstruction of the Washington Consensus, which suffers from exactly the same tendency to be described by its critics rather than its authors.

Two Theories of the Crisis

Behind the policy dispute sat a genuine theoretical disagreement, and it is the reason this episode still appears on graduate reading lists.

The fundamentals view. On this reading, the crisis was earned. East Asian growth had been driven by extraordinary rates of factor accumulation rather than by productivity gains — the argument Paul Krugman had popularized in “The Myth of Asia’s Miracle” (Foreign Affairs, 1994), drawing on Alwyn Young’s growth accounting. Add to that a pattern of implicitly guaranteed lending, in which politically connected borrowers took risks knowing that the state stood behind the banks, and you get over-investment in low-return projects funded by debt that was never priced for its true risk. The crisis was the moment the accumulated bad investment was recognized. In this account, the underlying disease was moral hazard and the crisis was the cure.

The panic view. On this reading, the crisis was a run. Steven Radelet and Jeffrey Sachs argued in the Brookings Papers on Economic Activity (1998) that while real weaknesses existed, they were nowhere near sufficient to explain the scale of the collapse. What happened was a self-fulfilling creditor panic: each foreign lender, seeing others refuse to roll over short-term credit, was individually rational to refuse as well, and the collective refusal made the borrowers insolvent. This is the international version of a classic bank run — solvent-but-illiquid institutions destroyed by the coordination failure among their creditors — and Roberto Chang and Andrés Velasco formalized it in exactly those terms.

The distinction is not academic. If the crisis was fundamentals, the correct response is restructuring and discipline, and a rescue merely postpones the reckoning. If the crisis was panic, the correct response is a lender of last resort large enough and fast enough to stop the run, and imposing austerity on a panicking economy is like fighting a bank run by publicly questioning the bank’s solvency.

The defensible synthesis — and it is roughly where the literature landed — is that the fundamentals determined which countries were vulnerable and the panic determined how far the collapse went. Weak balance sheets loaded the gun; the coordination failure among short-term creditors fired it. That synthesis has an uncomfortable implication for anyone who wants a clean story: a country can do most things right and still be destroyed by a run, if its liability structure makes a run possible.

Malaysia, and the Question Nobody Wanted Asked

On 1 September 1998, Malaysia did the thing that was not supposed to be done. It imposed selective controls on capital outflows, pegged the ringgit at 3.80 to the dollar, and locked up portfolio investment for twelve months. Mahathir Mohamad accompanied the measures with rhetoric about speculators that made the policy easy to dismiss as demagoguery.

The economics were harder to dismiss. Malaysia recovered on a timetable broadly comparable to its neighbors without an IMF program, and subsequent evaluations — including work by economists who had been skeptical in advance — found that the controls did not produce the collapse in investor confidence and the long exile from capital markets that critics predicted. The honest verdict in the empirical literature is not that controls were triumphantly vindicated; it is closer to “the controls bought policy room at a cost lower than the profession expected, in a country with the administrative capacity to enforce them and a government willing to bear the reputational hit.”

That is a modest conclusion with immodest consequences, because it undermined a strong prior. Through the 1990s, the direction of travel in international financial policy was toward full capital-account liberalization, up to and including a proposed amendment to the IMF’s Articles of Agreement to make it a Fund objective. The crisis stopped that project. Jagdish Bhagwati, no protectionist, published “The Capital Myth” (Foreign Affairs, 1998) arguing that the case for free trade in goods does not transfer to free trade in short-term capital, because capital flows are subject to panic and manias in a way that widget shipments are not. Dani Rodrik asked, in the same period, who exactly needed capital-account convertibility, and found the empirical case for it thinner than its advocates assumed — a line of argument developed further in our profile of Rodrik. By 2012 the IMF had published an institutional view acknowledging that capital-flow management measures can be a legitimate part of the policy toolkit in defined circumstances. That is an institutional reversal of a kind that rarely happens in fewer than fifteen years.

The Long Aftermath: Self-Insurance

The most consequential legacy of 1997 is something no summit agreed to.

Emerging economies drew the obvious lesson from watching Korea negotiate with the Fund: never again be in a position where you have to ask. The response was reserve accumulation on a scale without precedent — running current-account surpluses, holding down exchange rates, and stockpiling dollar assets as self-insurance against the next sudden stop. Regional arrangements followed, notably the Chiang Mai Initiative launched in 2000, a network of currency swaps designed to make the IMF the second call rather than the first.

The macroeconomic consequences of that decision were global. Ben Bernanke’s 2005 “global saving glut” hypothesis attributed part of the low long-term interest rates of the 2000s — and by extension part of the conditions that produced the American housing boom — to precisely this post-crisis surge of saving flowing from developing Asia into U.S. assets. Whether or not one accepts the full causal chain, the structural point stands: a crisis in Bangkok in 1997 helped set the price of a mortgage in Nevada in 2005.

Jargon note: a sudden stop is an abrupt reversal in net capital inflows, named by Guillermo Calvo. The phrase deliberately borrows a proverb — bankers lend you an umbrella and want it back when it rains. The analytical content is that the reversal is often driven by conditions in creditor markets rather than by anything the borrowing country did, which is why “just don’t do anything to alarm investors” is inadequate policy advice.

What It Changed in Economics

Three durable shifts came out of the episode.

Balance sheets entered macro. Before 1997, open-economy macroeconomics was largely about flows — trade balances, current accounts, real exchange rates. After 1997, the currency and maturity composition of stocks became central, because the crisis demonstrated that a devaluation’s effect depends entirely on who holds what denominated in what. This is the international counterpart of the balance-sheet emphasis in Minsky’s account of financial fragility and in Fisher’s debt-deflation.

Sequencing became respectable. The idea that liberalization has an order — trade before capital account, prudential regulation before financial opening, institutions before markets — had existed for years, but 1997 gave it evidence. It also connects backward to the older development literature on late industrialization and to the state-capacity question, since sequencing advice is useless to a government that lacks the administrative machinery to execute it.

Self-fulfilling equilibria stopped being exotic. Multiple-equilibrium models — where the same fundamentals support both a good outcome and a run, and expectations select between them — moved from a theoretical curiosity to a standard tool for thinking about sovereign debt, currency pegs, and bank funding. A decade later, the euro-area crisis would be argued in almost exactly this vocabulary.

Connection to the Broader Reckonomics Graph

The 1997 crisis is the hinge between the postwar international architecture and the contemporary one. Read it after Bretton Woods to see what replaced the system of managed exchange rates and capital controls that the 1944 conference designed. Read it alongside the Washington Consensus for the policy program it damaged, and alongside the Prebisch–Singer hypothesis for an older tradition of asking whether the world economy treats peripheral countries symmetrically. For the domestic-financial analogue of the same dynamics, see Minsky and the post-crisis era.