## _wp04192

## Source details

**Canonical URL:** [_wp04192](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04192.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04192.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04192.pdf.json)

---

### I. INTRODUCTION
- Mike Mussa’s observation (circa January or February of 1999) framed a “theorem”: IMF loans, if repaid and provided under adequate safeguards, need not generate moral hazard.
- Core insight: IMF lending creates no moral hazard when it occurs “at no one’s expense” — i.e., it is not a transfer from IMF creditor countries/taxpayers and does not impose costs on borrowing-country taxpayers.
- Two critical conditions:
  - (1) IMF loans must reflect repayment risk (no implicit subsidy).
  - (2) Borrowing countries act as single-agent benevolent governments (no domestic distributional agency problem).
- Paper objective: formalize the “Mussa theorem” in a tight model linking assumptions to results; then analyze when IMF lending produces symptoms commonly associated with moral hazard (higher capital flows, lower risk premia) and when those symptoms are efficient.

### II. A SIMPLE MODEL OF THE IMF — Assumptions and Laissez-Faire equilibrium
- Model environment and notation:
  - Large number of identical emerging market (capital scarce) countries; rest of world = investors from rich countries with wealth W per emerging market country (denoted W).
  - Time: three periods, t=0,1,2.
  - Investment: country invests k in period 0; output in period 2 is ykθ, where θ is stochastic. Investment cost = k + f(k), with f(.) increasing and concave.
  - Information: a signal perfectly informative about θ is observed in period 1; θ distributed with density g(·).
  - Financing: k financed by short-term debt maturing in period 1; debt due = d; creditors atomistic foreign investors with liquidation rights and sequential service constraint.
  - Liquidation value: kλθ with 1λ< (liquidation never efficient).
  - Pledgeability: in period 1 country can pledge fraction α of output to new lenders; 0<α≤1 (α<1 due to limited enforcement or debtor moral hazard).
  - Preferences: agents’ utility summarized as consumption in period 2 only; lenders risk neutral (zero expected return).
  - Assumption A1: emerging market country is run by a benevolent policymaker who chooses k to maximize representative resident welfare.
  - Equilibrium characterized by pair (k*, d*) that maximizes domestic welfare subject to foreign lenders’ participation.
  - Ex ante global welfare per country: GW0 = U0 + V0.

- Laissez-faire (no IMF) equilibrium key definitions and conditions:
  - Rollover crisis: occurs if d > kαθ; with small λ assume either no liquidation (d ≤ kαθ) or complete liquidation (d > kαθ).
  - Threshold θ̄ defined by d = kαθ̄. (Equation (1))
  - Zero-profit condition for lenders (neglecting liquidation payoff): ∫_{θ̄}^{∞} g(θ) dθ = 1/(1+rdk). (Equation (2))
  - Combined equilibrium summarized by equation (3) in text linking θ̄, k, and r.
  - Equilibrium θ̄ depends on α; equilibrium interest rate r_{lf} is decreasing with α.
  - Welfare-maximizing k (k_{lf}) solves max_k ∫_{θ̄}^{∞} [kθ - (k + f(k))] g(θ) dθ. (Equation (4))
  - First-order condition: 1 + ∫_{θ̄}^{∞} θ g(θ) dθ = f'(k_{lf}). (Equation (5))
  - Debtor country welfare under laissez-faire: U0^{lf} = ∫_{θ̄}^{∞} [kθ - (k + f(k))] g(θ) dθ.
  - Lenders’ ex ante welfare: V0 = W. Hence GW0^{lf} = U0^{lf} + W.
  - First-best (equity-like contract) yields higher capital than laissez-faire: k_{fb} satisfies f'(k_{fb}) = E[θ].

### II.C Effects of IMF Lending (mechanics and implications)
- IMF features in the model:
  - IMF can pledge fraction β of output (β generally different from α); assume βα>.
  - Assumption A2: IMF lends at the actuarially fair interest rate (expected profit = 0).
  - With IMF, equilibrium in period 0 is as if private creditors had extraction power β; equations (1)-(2) hold with α replaced by β.
  - Because β>α, liquidation threshold with IMF satisfies θ_{imf} < θ_{lf}; IMF lends to θ ∈ [θ_{imf}, θ_{lf}] — it bails out private creditors for those states.
  - IMF bailouts in model do not involve transfers from rest of world to debtor or private investors: IMF charges actuarially fair rate, so no implicit subsidy.

- Observable consequences (symptoms often labeled “investors’ moral hazard”):
  - capital flows increase: k_{imf} > k_{lf};
  - borrowing cost falls: r_{imf} < r_{lf}.
- Interpretation:
  - Despite these symptoms, no moral hazard stricto sensu in baseline model because no hidden action; increased flows and lower rates reflect efficient insurance.
  - Welfare: IMF increases ex ante welfare of emerging market residents (U0^{imf} > U0^{lf}) and leaves investors’ welfare unchanged (V0^{imf} = V0^{lf} = W), hence GW0^{imf} > GW0^{lf}.

### II.D The Mussa Theorem — Formal statement and welfare implications
- Introducing policy effort e:
  - Policy effort e affects distribution of θ: g(θ,e) with effort cost h(e) to representative resident (h(.) increasing and convex).
  - Country can commit to e at time 0 (no hidden-action debtor moral hazard for this analysis).
  - Under laissez-faire, θ_{lf} is decreasing in e: dθ_{lf}/de < 0; higher effort reduces default risk and interest rates.
  - Optimal effort under laissez-faire e_{lf} solves arg max_e ∫_{θ_{lf}(e)}^{∞} (kθ - (k + f(k))) g(θ,e) dθ - h(e). (Equation (8))
  - With IMF, analogous first-order conditions with θ_{imf}(e) replacing θ_{lf}(e).
- Comparative statics examples (three-state θ: θ_l < θ_m < θ_h):
  - If effort shifts probability from medium to high state, IMF reduces effort: e_{imf} < e_{lf}.
  - If effort shifts probability from low to medium state, IMF may increase effort: e_{imf} > e_{lf}.
- Crucial interpretation:
  - If A1 and A2 hold, any reduction in effort induced by IMF is an efficient response — IMF reduces effort from a second-best over-deterrent level to a first-best level given improved repayment technology β.

- Mussa theorem (text conclusion under A1 and A2):
  - Anticipation of IMF bailouts increases capital flows and reduces borrowing costs for emerging market countries.
  - Anticipation of bailouts may decrease domestic efforts to avoid crisis.
  - IMF does not generate moral hazard stricto sensu.
  - Expectation of IMF bailouts unambiguously increases welfare of recipient countries at no cost to the rest of the world.

### II.E Two doors for moral hazard to return (violations of A1 and A2) and other channels
- Door 1 — Domestic agency problem (violation of A1):
  - If policymaker is not benevolent and diverts a fraction of capital inflows (pandering to private benefit), IMF-induced larger capital inflows can worsen domestic welfare (indirect moral hazard).
  - If capital inflows are excessive under laissez-faire due to malevolent policymaker, IMF allowing more inflows can reduce representative resident welfare.
- Door 2 — IMF subsidy (violation of A2):
  - If IMF lends at an excessively low interest rate (subsidy σ ≥ 0, with imf_d σ ≤ d_{imf}), this constitutes a transfer from IMF financiers to recipient country.
  - Subsidy can make the medium state more attractive relative to the high state, reducing effort below first-best and potentially reducing global welfare if subsidy large enough.
  - Example: three-state model with subsidy σ reduces optimal effort and can lower global welfare.
- Additional door — Hidden-action debtor moral hazard:
  - If effort e is a hidden action (no commitment), IMF lending could aggravate the debtor moral hazard introduced by lack of commitment only if IMF increases ex ante borrowing cost r. Paradox: IMF would have to raise ex ante borrowing cost for it to lower welfare via hidden-action channel.
- Even when IMF-induced moral hazard exists, it may not be “excessive”: distortions can be mild and IMF may remain net-beneficial; but transfer-based lending implies redistribution from IMF financiers to recipients and not a Pareto improvement.

### III. IMPLICATIONS OF THE MUSSA THEOREM — Empirical literature, evidence, and IMF rules
- Empirical critique:
  - Many empirical “tests of moral hazard” look for symptoms (increased capital flows, lower country risk/premia) that arise under effective IMF lending whether or not moral hazard is present.
  - Such tests cannot distinguish whether observed effects are evidence of moral hazard or simply indicate the IMF “doing its job.”
  - Tests finding that IMF lending allows more expansionary fiscal and monetary policies are not by themselves tests of moral hazard.
  - Standard empirical “test for moral hazard” (lower borrowing costs or larger capital inflows following IMF lending) is not meaningful because such effects are predicted by models of crisis lending with or without moral hazard.
- Testing recommendations:
  - Empirical work should focus on the assumptions of the Mussa theorem (A1: benevolent domestic policymaker/commitment; A2: IMF lending at actuarially fair terms) rather than only on observable symptoms.
  - The Mussa theorem provides a precise empirical standard to determine whether IMF lending contains a subsidy; this differs from comparing IMF lending rates to market interest rates faced by crisis countries.
  - Assumption A2 requires only that the IMF rate of charge be high enough to cover the repayment risk faced by the IMF, which may be substantially lower than the risk borne by private lenders.

- Evidence summarized in the source:
  - Joshi and Zettelmeyer (2004) find that lending to low income countries carried an average annual rate of return, in the 1973-2003 period, of about 400 basis points below the rate the IMF would have received from lending the same amount to industrial countries at comparable maturities; this largely reflects deliberate concessional lending and HIPC debt relief.
  - For upper middle income countries (“emerging markets”) the spread between IMF rates of return and lending rates by industrial countries has been approximately zero.
  - Jeanne and Zettelmeyer (2001) predicted that about 95 percent of the currently open lending cycles to emerging market countries would eventually be repaid in full.
  - The shift to large-scale crisis lending beginning with the 1995 Mexican loan did not appear to constitute a structural break in repayment record for loans made between 1995 and 1999; those loans have been fully repaid or are well on their way to being repaid.
  - Loans to Argentina in late 2000 and 2001 are highlighted as potentially different — made when Argentina already had substantial outstanding obligations to the Fund and in response to fiscal problems that carried the risk of a capital account crisis — raising concern that such loans might carry higher default risk and thus imply a subsidy under current terms.

- IMF lending rules and exceptional access (as summarized):
  - Current IMF rules imply lending from general resources should not contain a transfer element: the IMF is allowed to lend only to countries likely to be in a position to repay in the medium term.
  - New guidelines issued in February of 2003 permit lending in large amounts (exceeding the usual limit of 100 percent of quota in any given year and 300 percent of quota in total) only when:
    - “a member is experiencing exceptional balance of payments pressures on the capital account resulting in a need for Fund,” and
    - “a rigorous and systematic analysis indicates that there is a high probability that the debt will remain sustainable,” and
    - “the member has good prospects of regaining access to private markets within the time Fund resources would be outstanding.”
  - Most instances of large-scale IMF crisis lending since 1995 would have satisfied the 2003 guidelines, but the most recent cases (the only large-scale crisis lending after the guidelines were issued) did not satisfy them, indicating the possibility that such criteria can be violated.

### Policy implications and conclusions
- Sources of IMF-induced moral hazard to avoid:
  - Lending at subsidized terms (violations of assumption A2).
  - “Indirect moral hazard” when IMF support amplifies preexisting distributional or governance problems, effectively redistributing wealth at the expense of the average domestic taxpayer.
- Recommended policy directions from the Mussa theorem:
  - Avoid lending at terms and in circumstances that would violate assumptions A1 and A2.
  - Ensure lending is “temporary” and occurs “under adequate safeguards” as envisaged in the Articles of Agreement.
  - Avoid supporting “gambles for resurrection” — pre-crisis measures that benefit ruling elites by preserving a chance to avoid crisis at enormous taxpayer expense if the gamble fails.
  - Consider linking access to IMF financing to pre-crisis policy quality and institutions (level and structure of outstanding debt, transparency, quality of financial sector regulation, responsiveness to IMF surveillance), while recognizing the time consistency problem this raises.
  - Recognize that what matters for avoiding moral hazard is whether IMF interest rates are actuarially fair in covering the risk faced by the IMF (which may be lower than private creditor risk), not whether IMF charges are lower than private market rates.

- Final conclusions:
  - The Mussa theorem defines conditions under which official crisis lending does not have adverse incentive effects and serves as an intellectual benchmark for analyzing IMF-induced moral hazard.
  - Empirically, whether the Mussa theorem holds is an open question; the theorem nonetheless identifies mechanisms through which moral hazard could arise.
  - The theorem clarifies the meaning of a moral-hazard-inducing “subsidy” in IMF lending: a subsidy exists only if IMF rates do not cover the repayment risk faced by the IMF.
  - The feasibility of incentive-friendly large-scale lending depends on resolving the time consistency problem inherent in conditioning access on pre-crisis policy quality and institutions.

*Source: _wp04192 - References..............................................................................................................*

### References..............................................................................................................

### _wp04192 - References..............................................................................................................

### I. INTRODUCTION
- Mike Mussa’s observation (circa January or February of 1999) framed a “theorem”: IMF loans, if repaid and provided under adequate safeguards, need not generate moral hazard.
- Core insight: IMF lending creates no moral hazard when it occurs “at no one’s expense” — i.e., it is not a transfer from IMF creditor countries/taxpayers and does not impose costs on borrowing-country taxpayers.
- Two critical conditions (made explicit by authors): (1) IMF loans must reflect repayment risk (no implicit subsidy), and (2) borrowing countries act as single-agent benevolent governments (no domestic distributional agency problem).
- Paper objective: formalize the “Mussa theorem” in a tight model linking assumptions to results; then analyze when IMF lending produces symptoms commonly associated with moral hazard (higher capital flows, lower risk premia) and when those symptoms are efficient.

### II. A SIMPLE MODEL OF THE IMF — Assumptions
- Economy: large number of identical emerging market (capital scarce) countries; rest of world = investors from rich countries with wealth W per emerging market country (denoted W).
- Time: three periods, t=0,1,2.
- Investment: country invests k in period 0; output in period 2 is ykθ, where θ is stochastic. Investment cost = k + f(k), with f(.) increasing and concave.
- Information: a signal perfectly informative about θ is observed in period 1; θ distributed with density g(·).
- Financing: k financed by short-term debt maturing in period 1; debt due = d; creditors atomistic foreign investors with liquidation rights and sequential service constraint.
- Liquidation value: kλθ with 1λ< (liquidation never efficient).
- Pledgeability: in period 1 country can pledge fraction α of output to new lenders; 0<α≤1 (α<1 due to limited enforcement or debtor moral hazard).
- Preferences: agents’ utility summarized as consumption in period 2 only; lenders risk neutral (zero expected return).
- Assumption A1: emerging market country is run by a benevolent policymaker who chooses k to maximize representative resident welfare.
- Equilibrium characterized by pair (k*, d*) that maximizes domestic welfare subject to foreign lenders’ participation.
- Ex ante global welfare per country: GW0 = U0 + V0.

### II.B Laissez-Faire (no IMF)
- Rollover crisis: occurs if d > kαθ; with small λ assume either no liquidation (d ≤ kαθ) or complete liquidation (d > kαθ).
- Define threshold θ̄ as minimum productivity where debt d can be serviced: d = kαθ̄. (Equation (1))
- Zero-profit condition for lenders (neglecting liquidation payoff): ∫_{θ̄}^{∞} g(θ) dθ = 1/(1+rdk). (Equation (2))
- Combined equilibrium condition (substituting d and k): 1 = ∫_{θ̄}^{∞} g(θ) dθ · (1+rdk) ⇒ summarized as equation (3) in text.
- Equilibrium θ̄ depends on α; equilibrium interest rate r_{lf} is decreasing with α.
- Welfare-maximizing k (k_{lf}) solves max_k ∫_{θ̄}^{∞} [kθ - (k + f(k))] g(θ) dθ. (Equation (4)) First-order: 1 + ∫_{θ̄}^{∞} θ g(θ) dθ = f'(k_{lf}). (Equation (5))
- Debtor country welfare under laissez-faire: U0^{lf} = ∫_{θ̄}^{∞} [kθ - (k + f(k))] g(θ) dθ.
- Lenders’ ex ante welfare: V0 = W. Hence GW0^{lf} = U0^{lf} + W.
- First-best (equity-like contract) yields higher capital than laissez-faire: k_{fb} satisfies f'(k_{fb}) = E[θ].

### II.C Effects of IMF Lending
- IMF can pledge fraction β of output (β generally different from α); assume βα>.
- Assumption A2: IMF lends at the actuarially fair interest rate (expected profit = 0).
- With IMF, equilibrium in period 0 is as if private creditors had extraction power β; equations (1)-(2) hold with α replaced by β.
- Because β>α, liquidation threshold with IMF satisfies θ_{imf} < θ_{lf}; IMF lends to θ ∈ [θ_{imf}, θ_{lf}]—it bails out private creditors for those states.
- IMF bailouts in model do not involve transfers from rest of world to debtor or private investors: IMF charges actuarially fair rate, so no implicit subsidy.
- Consequences (symptoms often labeled “investors’ moral hazard”):
  - capital flows increase: k_{imf} > k_{lf};
  - borrowing cost falls: r_{imf} < r_{lf}.
- Despite these symptoms, no moral hazard stricto sensu in baseline model because no hidden action; increased flows and lower rates reflect efficient insurance.
- Welfare: IMF increases ex ante welfare of emerging market residents (U0^{imf} > U0^{lf}) and leaves investors’ welfare unchanged (V0^{imf} = V0^{lf} = W), hence GW0^{imf} > GW0^{lf}.

### II.D The Mussa Theorem (formal statement and welfare implications)
- Introduce policy effort e affecting distribution of θ: g(θ,e) with effort cost h(e) to representative resident (h(.) increasing and convex).
- Assumption: country can commit to e at time 0 (no hidden-action debtor moral hazard for this analysis).
- Under laissez-faire, threshold θ_{lf} is decreasing in e: dθ_{lf}/de < 0; higher effort reduces default risk and interest rates.
- Optimal effort under laissez-faire e_{lf} solves arg max_e ∫_{θ_{lf}(e)}^{∞} (kθ - (k + f(k))) g(θ,e) dθ - h(e). (Equation (8)) Notation: e_{lf} independent of k.
- With IMF, analogous first-order conditions with θ_{imf}(e) replacing θ_{lf}(e).
- Examples with θ taking three values (θ_l < θ_m < θ_h) illustrate ambiguity:
  - If effort shifts probability from medium to high state, IMF reduces effort: e_{imf} < e_{lf}.
  - If effort shifts probability from low to medium state, IMF may increase effort: e_{imf} > e_{lf}.
- Crucial interpretation: if A1 and A2 hold, any reduction in effort induced by IMF is an efficient response—IMF reduces effort from a second-best over-deterrent level to a first-best level given improved repayment technology β.
- Mussa theorem (text): Under A1 and A2:
  - Anticipation of IMF bailouts increases capital flows and reduces borrowing costs for emerging market countries.
  - Anticipation of bailouts may decrease domestic efforts to avoid crisis.
  - IMF does not generate moral hazard stricto sensu.
  - Expectation of IMF bailouts unambiguously increases welfare of recipient countries at no cost to the rest of the world.

### II.E Two doors for moral hazard to return (violations of A1 and A2)
- Door 1 — Domestic agency problem (violation of A1):
  - If policymaker is not benevolent and diverts a fraction of capital inflows (pandering to private benefit), IMF-induced larger capital inflows can worsen domestic welfare. IMF can thus exacerbate domestic expropriation (“indirect moral hazard”).
  - If capital inflows are excessive under laissez-faire due to malevolent policymaker, IMF allowing more inflows can reduce representative resident welfare.
- Door 2 — IMF subsidy (violation of A2):
  - If IMF lends at an excessively low interest rate (subsidy σ ≥ 0, with imf_d σ ≤ d_{imf}), this constitutes a transfer from IMF financiers to recipient country.
  - Subsidy makes the medium state more attractive relative to the high state, reducing effort below first-best and potentially reducing global welfare if subsidy large enough.
  - Example: three-state model with subsidy σ reduces optimal effort and can lower global welfare.
- Additional door — Hidden-action debtor moral hazard:
  - If effort e is a hidden action (no commitment), IMF lending could aggravate the debtor moral hazard introduced by lack of commitment only if IMF increases ex ante borrowing cost r. Paradoxically, IMF would have to raise ex ante borrowing cost for it to lower welfare via hidden-action channel.
- Even when IMF-induced moral hazard exists, it may not be “excessive”: distortions can be mild and IMF may remain net-beneficial; but transfer-based lending implies redistribution from IMF financiers to recipients and not a Pareto improvement.

### III. IMPLICATIONS OF THE MUSSA THEOREM — Empirical literature
- Key insight: many empirical “tests of moral hazard” look for symptoms (increased capital flows, lower country risk/premia) that arise under effective IMF lending whether or not moral hazard is present.
- Therefore, such empirical tests miss the point; testing should focus on the assumptions of the Mussa theorem (A1: benevolent domestic policymaker/commitment; A2: IMF lending at actuarially fair terms) rather than only on the observable symptoms.

*Source: _wp04192 - References..............................................................................................................*

### references). Without exception, the tests performed in this literature are incapable of

### _wp04192 - references). Without exception, the tests performed in this literature are incapable of

### Empirical literature critique
- Existing empirical tests that link IMF involvement to changes in market variables cannot distinguish whether observed effects are evidence of moral hazard or simply indicate the IMF “doing its job.”
- Papers that find IMF lending allows countries to run more expansionary fiscal and monetary policies are not by themselves tests of moral hazard because IMF lending is intended to permit such policies when countries face balance of payments difficulties.
- The standard empirical “test for moral hazard” (lower borrowing costs or larger capital inflows following IMF lending) is not meaningful because such effects are predicted by models of crisis lending with or without moral hazard.

### Testing moral hazard via the Mussa theorem
- The Mussa theorem provides a precise empirical standard to determine whether IMF lending contains a subsidy, which differs from the standard of comparing IMF lending rates to market interest rates faced by crisis countries.
- Assumption A2 (avoiding moral hazard) requires only that the IMF rate of charge be high enough to cover the repayment risk faced by the IMF, which may be substantially lower than the risk borne by private lenders.
- An alternative empirical approach is to test directly for the validity of the Mussa theorem’s conditions rather than testing for reduced private borrowing costs.

### Evidence on IMF rates of return and repayment
- Joshi and Zettelmeyer (2004) find that lending to low income countries carried an average annual rate of return, in the 1973-2003 period, of about 400 basis points below the rate the IMF would have received from lending the same amount to industrial countries at comparable maturities; this largely reflects deliberate concessional lending and HIPC debt relief.
- For upper middle income countries (“emerging markets”) the spread between IMF rates of return and lending rates by industrial countries has been approximately zero.
- Jeanne and Zettelmeyer (2001) used a duration model and maximum likelihood estimation to compare current IMF “lending cycles” with past repaid cycles and predicted that about 95 percent of the currently open lending cycles to emerging market countries would eventually be repaid in full.
- The shift to large-scale crisis lending beginning with the 1995 Mexican loan did not appear to constitute a structural break in repayment record for loans made between 1995 and 1999; those loans have been fully repaid or are well on their way to being repaid.
- Loans to Argentina in late 2000 and 2001 are highlighted as potentially different—made when Argentina already had substantial outstanding obligations to the Fund and in response to fiscal problems that carried the risk of a capital account crisis—raising concern that such loans might carry higher default risk and thus imply a subsidy under current terms.

### IMF lending rules and exceptional access
- Current IMF rules imply lending from general resources should not contain a transfer element: the IMF is allowed to lend only to countries likely to be in a position to repay in the medium term.
- New guidelines issued in February of 2003 permit lending in large amounts (exceeding the usual limit of 100 percent of quota in any given year and 300 percent of quota in total) only when:
  - “a member is experiencing exceptional balance of payments pressures on the capital account resulting in a need for Fund,” and
  - “a rigorous and systematic analysis indicates that there is a high probability that the debt will remain sustainable,” and
  - “the member has good prospects of regaining access to private markets within the time Fund resources would be outstanding.”
- Most instances of large-scale IMF crisis lending since 1995 would have satisfied the 2003 guidelines, but the most recent cases (the only large-scale crisis lending after the guidelines were issued) did not satisfy them, indicating the possibility that such criteria can be violated.

### Sources of IMF-induced moral hazard and policy implications
- Two main sources of IMF-induced moral hazard:
  - Lending at subsidized terms (violations of assumption A2).
  - “Indirect moral hazard” when IMF support amplifies preexisting distributional or governance problems, effectively redistributing wealth at the expense of the average domestic taxpayer.
- Policy implications from the Mussa theorem:
  - Avoid lending at terms and in circumstances that would violate assumptions A1 and A2.
  - Ensure lending is “temporary” and occurs “under adequate safeguards” as envisaged in the Articles of Agreement.
  - Avoid supporting “gambles for resurrection” — pre-crisis measures that benefit ruling elites by preserving a chance to avoid crisis at enormous taxpayer expense if the gamble fails.
  - Consider linking access to IMF financing to pre-crisis policy quality and institutions (level and structure of outstanding debt, transparency, quality of financial sector regulation, responsiveness to IMF surveillance), while recognizing the time consistency problem this raises.
  - Recognize that what matters for avoiding moral hazard is whether IMF interest rates are actuarially fair in covering the risk faced by the IMF (which may be lower than private creditor risk), not whether IMF charges are lower than private market rates.

### Conclusions
- The Mussa theorem defines conditions under which official crisis lending does not have adverse incentive effects and serves as an intellectual benchmark for analyzing IMF-induced moral hazard.
- Empirically, whether the Mussa theorem holds is an open question; the theorem nonetheless identifies mechanisms through which moral hazard could arise.
- The theorem clarifies the meaning of a moral-hazard-inducing “subsidy” in IMF lending: a subsidy exists only if IMF rates do not cover the repayment risk faced by the IMF.
- The feasibility of incentive-friendly large-scale lending depends on resolving the time consistency problem inherent in conditioning access on pre-crisis policy quality and institutions.

*Source: _wp04192 - references). Without exception, the tests performed in this literature are incapable of*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04192.pdf_