Skip to main content
The Mexican Peso Crisis

Currency Crisis Theory: First, Second, and Third Generation Models

Pomegra Learn

Why Do Currencies Collapse? What Theory Tells Us About Mexico 1994

Currency crises have a paradoxical quality: they often arrive with great suddenness after conditions that seemed, at minimum, manageable. Governments maintain exchange rate pegs for years, then lose them in days. Reserves that seemed adequate become clearly insufficient in a matter of weeks. Academic economists have spent the four decades since Mexico's 1976 crisis developing frameworks — called generation models — that attempt to explain this pattern. The three generations of currency crisis theory offer progressively richer explanations: the first explains crises as the inevitable result of inconsistent fundamentals, the second explains why crises can be self-fulfilling even when fundamentals are defensible, and the third explains how balance sheet vulnerabilities can amplify and transmit crises in unexpected ways. Mexico 1994 illuminates all three frameworks simultaneously — which is part of why it remains one of the most studied currency crises in economic history.

Currency crisis: A forced abandonment of a fixed or managed exchange rate peg, typically involving rapid depreciation of 20 percent or more, often accompanied by reserve depletion, emergency interest rate increases, and requests for external financial assistance.

Key Takeaways​

  • First-generation models (Krugman 1979) explain crises as the mechanical result of fiscal deficits financed by money creation eroding the reserve base needed to defend the peg.
  • Second-generation models (Obstfeld 1994) show that crises can be self-fulfilling: if enough investors believe a devaluation is coming, the resulting speculative attack can force the devaluation even when fundamentals alone would not have required it.
  • Third-generation models add balance sheet effects: foreign currency debt creates vulnerabilities that make crises more severe and harder to prevent through conventional policy responses.
  • Mexico 1994 fits all three frameworks: it had genuine fundamental vulnerabilities (current account deficit, reserve depletion) that justified attack, self-fulfilling dynamics (the investor refusal-to-roll problem), and balance sheet amplification (tesobonos doubling in peso cost after devaluation).
  • The distinction between inevitable and self-fulfilling crises matters enormously for policy: a fundamentals-driven crisis requires adjustment; a self-fulfilling crisis may be preventable through credible commitment devices or lender-of-last-resort backstops.
  • The Asian crisis (1997) pushed economists toward third-generation models; the European sovereign debt crisis (2010–12) revived second-generation thinking about self-fulfilling dynamics in advanced economies.

First-Generation Models: The Fundamentals View​

The first-generation framework emerged from Paul Krugman's 1979 paper, formalized and extended by Robert Flood and Peter Garber in 1984. The model's logic is mechanically straightforward.

A government fixes its exchange rate and commits to defending it by selling foreign exchange reserves whenever the market demands. The government also runs a fiscal deficit that it finances — at least partly — through money creation. Money creation means the domestic money supply expands faster than the peg requires. To maintain the peg, the central bank must sterilize this monetary expansion by selling foreign exchange reserves. The reserve stock therefore shrinks continuously.

At some point, rational investors realize the reserves will be fully depleted. They do not wait for that moment to arrive organically. Instead, they collectively buy foreign exchange from the central bank at the current pegged rate — a speculative attack — exhausting the remaining reserves before the "natural" depletion date. The timing of the attack is predictable in the model: it happens when the shadow floating rate (what the exchange rate would be if it floated freely) equals the current pegged rate.

First-generation predictions:

  • Crises are driven by inconsistency between domestic policy (deficit monetization) and external commitment (fixed exchange rate)
  • Speculative attacks are rational and their timing is deterministic given the fundamentals
  • The crisis is inevitable once the fundamental inconsistency is in place; only the timing is uncertain
  • Policy solution: eliminate the fundamental inconsistency (stop the monetary financing, run fiscal surpluses)

Mexico's fit with first-generation theory: Partial. Mexico in 1994 had genuine fundamental vulnerabilities — the current account deficit exceeded 8 percent of GDP, real exchange rate appreciation had eroded competitiveness, and capital flows were potentially reversible. The reserve depletion through 1994 fit the Krugman pattern of a declining reserve buffer. But Mexico's deficit was not primarily financed by money creation; the government was running an approximately balanced budget and not printing money in the classic sense. The tesobono problem was a different kind of inconsistency — a maturity and currency mismatch rather than pure monetization.

Second-Generation Models: Self-Fulfilling Crises​

The first-generation framework could not explain a class of crises that seemed to occur when fundamentals were defensible — where the government had fiscal surpluses, reserves were not obviously inadequate, and monetary policy was not expansionary. The UK and France in the 1992 European Exchange Rate Mechanism crisis were canonical examples: neither country had obviously inconsistent fundamentals, yet both faced devastating speculative attacks.

Maurice Obstfeld's 1994 models introduced the second-generation framework. The key insight is a trade-off in government decision-making.

A government defending a currency peg bears costs — typically higher interest rates to attract capital and prevent outflows. Those costs are acceptable when the economy is performing well. They become increasingly unacceptable when the economy is weak, unemployment is high, or political pressure mounts for monetary easing. At some interest rate premium, the government would rationally prefer to abandon the peg rather than continue paying the defense cost.

This creates a circular logic:

  • If investors believe the government will defend the peg, they will not attack, so the defense costs remain low, so the government will defend — consistent equilibrium
  • If investors believe the government will abandon the peg, they will attack, forcing higher defense costs, making abandonment more likely — also a consistent equilibrium

The economy has multiple equilibria. Which one prevails depends on expectations, not solely on fundamentals. A crisis can be triggered by a shift in expectations — caused by a political shock, a change in government, news about reserves, or simply a coordination of investor pessimism — even when fundamentals would, in the absence of the attack, have been sustainable.

Second-generation predictions:

  • Crises can occur in a "zone of vulnerability" where both the peg-defense equilibrium and the peg-abandonment equilibrium are consistent
  • Crises may be unpredictable even given full information about fundamentals
  • Policy solution: either eliminate the vulnerability zone (through structural reforms, credible institutional commitment) or create commitment devices (currency boards, eurozone membership) that raise the cost of abandonment
  • Lender-of-last-resort facilities can also eliminate the multiple-equilibria problem by ensuring defense costs are always manageable

Mexico's fit with second-generation theory: Strong. Mexico was in the vulnerable zone by late 1994. The tesobono rollover problem was precisely a self-fulfilling mechanism: investors would roll if they believed others would roll, but if enough investors refused to roll, the government would default, confirming the refusal. The political shocks (Zapatista, Colosio, Ruiz Massieu assassinations) functioned as coordination devices that shifted the equilibrium from the "investors roll" outcome to the "investors exit" outcome without fundamentally changing the economic data.

The political class's commitment to maintaining the peg through the election year also fits the second-generation political economy story: Mexican authorities were willing to pay extremely high defense costs (depleting $18 billion in reserves from January to November 1994) to maintain the appearance of stability.

Third-Generation Models: Balance Sheet Effects​

The Asian financial crisis of 1997 produced currency collapses that neither first nor second-generation models fully explained. Thailand, Korea, and Indonesia did not have obvious fiscal monetization problems (first-generation) or straightforward political commitment failures (second-generation). Instead, they had accumulated large stocks of private-sector short-term foreign currency debt through their banking systems.

Third-generation models, associated with work by Guillermo Calvo, Paul Krugman (revisiting his own framework), Roberto Chang, and Andrés Velasco, emphasize balance sheet effects:

The balance sheet channel: When a firm or bank has liabilities denominated in foreign currency and assets denominated in domestic currency, a devaluation creates an instant deterioration of the balance sheet. Net worth falls in proportion to the devaluation times the foreign currency liability stock. If this balance sheet effect is large enough, it can trigger insolvency, credit contraction, investment collapse, and economic recession — making the post-crisis environment far worse than the exchange rate change itself would suggest.

The collateral-amplification mechanism: Balance sheet deterioration reduces collateral values, which forces creditors to cut credit lines, which reduces investment, which reduces output, which further deteriorates balance sheets — a contractionary spiral that can persist long after the exchange rate stabilizes.

Fire sale dynamics: A sudden reversal of capital flows forces rapid deleveraging. Assets that were collateral for borrowing must be sold at distressed prices to repay foreign currency loans, depressing asset prices further and generating losses for others in the financial system.

Mexico's fit with third-generation theory: Partial but present. The tesobono structure was precisely a balance sheet problem: government liabilities denominated in dollars against domestic peso revenues. The post-devaluation doubling of the peso-denominated cost of tesobono repayment was exactly the balance sheet deterioration that third-generation models predict. The broader banking system in Mexico also had currency mismatches that contributed to the 1995 banking crisis (covered in a subsequent article). Mexico 1994 therefore contains the seeds of the third-generation mechanism, though the Asian crisis of 1997 made it the canonical illustration.

Synthesizing the Three Frameworks​

The three frameworks are not mutually exclusive. A complete account of Mexico 1994 requires elements of all three:

From first-generation theory: Mexico's structural vulnerabilities — the current account deficit, the real appreciation, the declining reserve buffer — created a background condition of fundamental inconsistency. The exchange rate band was not obviously sustainable given the capital flow dynamics.

From second-generation theory: The actual timing and character of the crisis, including the political-shock-triggered nature of the reserve depletion and the self-fulfilling rollover problem, fits the multiple-equilibria framework. The Zedillo government's mishandled December 1994 devaluation — announcing a 15 percent devaluation that immediately collapsed to 50 percent — eliminated the defense equilibrium by demonstrating that the government's commitment was weaker than believed.

From third-generation theory: The tesobono balance sheet problem amplified the post-devaluation fiscal cost and contributed to the severity of the subsequent recession. The banking system's currency and maturity mismatches created second-round effects that made recovery slower and costlier than the exchange rate adjustment alone would have implied.

Policy Implications of the Theory Distinctions​

The generation distinction has direct policy relevance.

If the crisis is primarily first-generation: The prescription is fundamental adjustment — fiscal consolidation, exchange rate realignment, monetary discipline. External financing buys time for adjustment but does not substitute for it.

If the crisis is primarily second-generation: The prescription may be different. Structural adjustment may not be necessary; instead, a credible commitment device (an IMF backstop facility, a currency board arrangement, eurozone membership) might eliminate the vulnerability zone entirely by ensuring that defense costs can always be met. The IMF's Contingent Credit Line and Flexible Credit Line facilities, developed partly in response to Mexico and Asia, reflect second-generation thinking.

If the crisis is primarily third-generation: The prescription must include balance sheet restructuring — debt restructuring, bank recapitalization, forced deleveraging — in addition to exchange rate and macro adjustments. External financing must be large enough to prevent the fire-sale amplification.

Mexico 1994 ultimately required all three: structural reform (fiscal, current account), IMF backstop to restore rollover confidence, and bank recapitalization to address balance sheet damage in the financial system.

The Models in Later Crises​

Currency crisis theory did not stop with three generations. Each subsequent major crisis added refinements:

Asian crisis 1997: Clarified third-generation mechanisms, added the role of private-sector (not just government) balance sheets, emphasized the current account composition issue (short-term bank borrowing versus FDI).

Russia 1998: Combined first-generation fiscal monetization with second-generation confidence collapse and third-generation banking sector balance sheet fragility.

Argentina 2001: Illustrated how currency board commitment devices (intended to solve second-generation problems) can create their own rigidities that make eventual adjustment even more painful.

Eurozone 2010–12: Revived second-generation thinking in an advanced-economy context. Countries like Greece and Italy faced self-fulfilling sovereign spread dynamics; the ECB's "whatever it takes" commitment in 2012 functioned as a second-generation commitment device that eliminated the panic equilibrium.

Common Mistakes in Applying Currency Crisis Theory​

Treating crises as obviously inevitable in retrospect. After a crisis occurs, analysts frequently find the "fundamental inconsistencies" that made it inevitable. This hindsight bias understates how many countries with similar fundamentals did not experience crises — and therefore overstates how predictable any particular crisis was.

Ignoring the multiple equilibria implication. If second-generation models are correct, crisis prevention and crisis management are different problems. Prevention may require credible institutional commitment rather than structural adjustment; management may require restoring confidence rather than imposing austerity.

Treating the generation models as sequential improvements rather than complementary frameworks. The academic literature sometimes presents each new generation as superseding the previous one. In practice, real crises involve elements of all three frameworks simultaneously, as Mexico 1994 demonstrates.

Frequently Asked Questions​

Which generation model best describes Mexico 1994? All three are partially applicable. The first-generation framework explains the background vulnerability; the second-generation framework explains the dynamics and timing; the third-generation framework explains the severity and persistence of the post-crisis recession. Most economists treating Mexico as a pure first or second-generation case are simplifying.

Could the crisis have been prevented with better information disclosure? Probably not entirely, but information mattered. The Zedillo administration's failure to disclose the full extent of reserve depletion and tesobono obligations maintained artificial confidence through late 1994. Better disclosure might have forced earlier adjustment — painful but more manageable than the catastrophic December collapse.

Did the peso crisis change how economists think about capital controls? Significantly. Before 1994, the Washington Consensus consensus strongly favored capital account liberalization. The speed with which foreign capital reversed in Mexico, the Asian crisis, and Russia caused a reassessment. The IMF formally reversed its opposition to temporary, targeted capital flow management measures in a 2012 institutional view paper.

What is the Greenspan-Guidotti rule and how does it relate to these models? The rule — that reserves should equal at least one year of short-term external debt obligations — emerged directly from the Mexico and Asia experiences. It operationalizes the first-generation insight that reserve adequacy relative to near-term obligations determines vulnerability, while also addressing the second-generation concern about rollover confidence.

Summary​

Currency crisis theory has evolved through three generations, each adding explanatory power. First-generation models explain crises as the mechanical result of inconsistent fundamentals — particularly fiscal monetization eroding reserves. Second-generation models show that crises can be self-fulfilling: if investors coordinate on pessimism, their collective exit can force a devaluation that would not have been necessary absent the attack. Third-generation models add balance sheet effects — the amplification of crises through currency mismatches in private and public sector debt. Mexico 1994 contains elements of all three: genuine fundamental vulnerabilities created the conditions, self-fulfilling rollover dynamics determined the timing, and balance sheet amplification (through tesobonos and banking sector mismatches) made the aftermath more severe. Understanding which framework applies — and recognizing that multiple frameworks often apply simultaneously — is essential both for analyzing historical crises and for assessing contemporary currency risks.

Next​

→ The December Devaluation