Transcript of an IMF Economic Forum: Contractionary Currency Crashes in Developing Countries -- The Mundell-Fleming Lecture, IMF Annual Research Conference (Jeffrey A. Frankel, November 5, 2004)

IMF News, November 5, 2004

Source details

Canonical URL: Transcript of an IMF Economic Forum: Contractionary Currency Crashes in Developing Countries -- The Mundell-Fleming Lecture, IMF Annual Research Conference (Jeffrey A. Frankel, November 5, 2004)

Other formats

Bibliographic details

Published: November 5, 2004

Overview

Lecture theme: political costs of devaluations; why devaluations are often contractionary in emerging markets; what countries can do to reduce vulnerability.
Three related papers/points framed around (1) political costs, (2) contractionary effects, (3) reducing vulnerability (short run and long run).

Political costs of devaluation

Historical benchmark: Richard Cooper (1972) statistic — leaders lose their job 30 percent of the time within the year following a devaluation; control group 14 percent.
Updated methodology: threshold for a “currency crash” raised to 25 percent; requires acceleration relative to preceding year and new crash if none in prior three years.
Key empirical findings:
- Leader turnover:
- Within six months of a devaluation: 23 percent lose office versus 12 percent in normal periods.
- Within 12 months of a devaluation: 29 percent versus 20 percent in normal periods (a 45-percent increase; not quite statistically significant at conventional levels).
- Finance ministers and central bank governors:
- Normal times: 35.8-percent probability of losing job.
- Within 12 months of a devaluation: 58.3 percent.
- Broken-promise effect:
- If leaders/ministers/governors publicly promised not to devalue in the preceding month: executives lost office half the time within six months; two-thirds within 12 months.
- If no such promise: about one-fifth within six months; one-third within 12 months.
Conditioning on IMF programs (±3 months): IMF program presence did not materially change the elevated post-devaluation turnover rates — among countries with IMF programs the leader lost office 21 percent of the time, similar to full sample.

Why devaluations are often contractionary (mechanisms discussed)

Candidate mechanisms (many require rapid passthrough):
- Imported inputs: devaluation raises import-input prices (e.g., oil) → adverse supply shock.
- Consumer-price & wage pass-through: CPI rise combined with real wage rigidity or indexed wages → adverse effects on non-traded sectors.
- Sectoral redistribution: relative price changes alter income distribution; if winners have lower marginal propensity to consume, aggregate demand may fall.
- Real balances effect: general price-level rise reduces real money supply → contractionary.
Balance-sheet effects (emphasized):
- Currency mismatch: firms/banks borrow in foreign currency while revenues are in local currency → devaluation sharply raises local-currency cost of servicing foreign debt.
- Empirical association: magnitude of foreign-denominated debt × magnitude of depreciation correlates strongly with output losses (graphical evidence noted for cases such as Indonesia ’97, Bulgaria ’96).
- Terminology clarified: “original sin” (Hausmann/Eichengreen) denotes structural inability to borrow in local currency; currency mismatch and balance-sheet vulnerability arise from foreign-currency exposure.

Passthrough evidence (Shang-Jin Wei, David Parsley, Jeffrey Frankel work)

Distinguish import-price passthrough at dock vs. retail vs. CPI; avoid composition bias by using narrowly defined products across many countries (e.g., branded cognac, Marlboro carton, Time magazine).
Empirical results (sample starting in 1990):
- Passthrough to import prices at the dock: 80 percent.
- Passthrough to identical imported goods at retail: ~50 percent (described as “just over 40 percent” in regressions).
- Local competitors and CPI: lower passthrough than dock/import-price measures.
- Cross-country pattern: higher passthrough historically for less developed countries; at start of sample passthrough ~0.8 for developing countries versus 0.3 for rich countries.
- Time trends: strong downward trends in passthrough over the 1990s — estimated declines of 0.25 per year for rich countries and an additional 0.5 per year for developing countries (sum = 0.75 per year in raw specification); historic average passthrough during the sample ~0.5.
- Many usual determinants (income per capita, tariffs, distance, wages, inflation) explain part of differences but do not fully explain the large downward trend, especially for poorer countries.

Short-run vulnerability: the “procrastination phase” (sudden stop dynamics)

Empirical characterization:
- Period between peak of reserves (inflows cease) and full crash (25-percent depreciation) typically lasts six to 13 months.
- During this phase countries frequently:
- Publicly assert “no devaluation.”
- Run down official reserves.
- Shift debt maturity from long term to short term.
- Shift currency composition toward foreign currency (example: Mexico ’93–’94).
Mexico ’94 specifics:
- Rapid shift toward dollar-linked debt (tesobonos); by late 1994 tesobonos relative measure reached 70 percent (tesobonos/(tesobonos + cetes) measure).
- Maturity shifted from long-term toward short-term in the two years prior to the crash.
Common feature and policy implication:
- These procrastination strategies can buy time and sometimes succeed if external conditions improve, but they also worsen balance-sheet vulnerability and increase the chance that any eventual devaluation will be contractionary.
- Lesson: after inflows cease, adjust to the new external balance promptly rather than gambling for resurrection, to preserve balance sheets and the ability to use conventional expenditure-switching/expenditure-reducing policy mixes.

Long-run structural mitigation: trade openness and crises

Question: does greater openness to trade raise or reduce vulnerability to sudden stops and currency crashes?
Arguments in favor of openness reducing vulnerability:
- Higher trade/GDP implies smaller required real adjustment to restore external balance for a given loss of inflows.
- Bilateral trade ties can act as “hostages,” lowering creditor fears of default (Eaton & Gersovitz; Andy Rose evidence).
Empirical approach (Cavallo & Frankel):
- Use gravity-model instrument for trade openness (trade/GDP) to address endogeneity of trade.
- Results:
- Greater trade openness is associated with lower likelihood of sudden stops and currency crashes; effect strengthens with gravity IV.
- Quantitative statement: an increase in trade openness of 10 percentage points decreases the likelihood of a sudden stop by approximately 32 percent (example scale: difference between Argentina and Australia).
- Evidence also suggests openness reduces output costs associated with currency crashes.

Takeaways (three main conclusions)

Political costs: Updating Cooper — leaders, finance ministers, and central bank governors are substantially more likely to lose office after a devaluation; effects are larger when prior public assurances against devaluation exist but elevated turnover occurs even without prior promises.
Passthrough and contraction channels: passthrough coefficients declined sharply over the 1990s (the lecture cites “over the last 12 years”); thus contractionary effects are less likely to operate primarily through rapid CPI/import-price passthrough and more likely through currency mismatch and real-balance/balance-sheet effects.
Policy implications to avoid contractionary currency crashes:
- Short run: avoid weakening balance sheets during the procrastination phase—adjust promptly after inflows cease rather than gambling for resurrection.
- Long run: structurally increase trade openness to reduce vulnerability to sudden stops and currency crashes.

Transcript of The Mundell-Fleming Lecture delivered by Jeffrey A. Frankel, Washington, D.C., November 5, 2004 — IMF External Relations Department (Public Affairs, Media Relations).