## mundell-fleming-lecture

## Source details

**Canonical URL:** [mundell-fleming-lecture](https://www.imf.org/-/media/files/news/seminars/2023/arc/mundell-fleming-lecture.pdf)

## Other formats

- [Markdown version](/-/media/files/news/seminars/2023/arc/mundell-fleming-lecture.pdf.md)
- [Structured JSON version](/-/media/files/news/seminars/2023/arc/mundell-fleming-lecture.pdf.json)

---

### 2.1 The Neoclassical Paradigm meets the Allocation Puzzle

- Neoclassical conventional wisdom: promised benefits of external bond markets
  - Relax the saving=investment constraint by allowing access to global savings to fund investment and speed the transition to the steady state of the neoclassical growth model.
  - Insurance: sovereign debt markets enable risk sharing and can be used to “smooth” consumption or government expenditure relative to a volatile aggregate income process.
  - Acknowledged frictions: (i) limited commitment; (ii) limited state contingency; (iii) rollover risk; and (iv) deadweight costs of default (loss of reputation, declines in trade or output, increases in inequality).

- Neoclassical implication: debt and capital as complements
  - Representative constraint example: B ≤ νK, where debt is collateralized by physical capital and ν>0 is the allowable leverage ratio.
  - Prediction: sovereign borrowing increases physical capital beyond private saving; debt and capital increase together along a transition path.

- The Allocation Puzzle (empirical contradiction)
  - Gourinchas and Jeanne (2013): emerging and developing economies that grew relatively fast exported savings on net, while countries that lagged behind imported capital.
  - Empirical scatter (1970–2004) shows an upward sloping relationship: countries that grew the fastest were net exporters of savings — hard to reconcile with neoclassical logic.

- Alternative paradigm: Growth in the Shadow of Expropriation
  - Replace collateral constraint B ≤ νK with incentive/political constraint WG(B) ≥ WD(B), where WG is the value of the political incumbent as a function of external debt B and WD is the value of deviating/defaulting as a function of physical capital K.
  - Two interpretations:
    - Trigger-strategy view: WG decreases in B; WD increases in K — generating a negative relationship between debt and capital.
    - Expropriation/taxation view: more debt tempts government to tax/expropriate capital, reducing private willingness to invest — “growth in the shadow of expropriation.”
  - Political economy ingredients:
    - Government as sequence of stochastic incumbents biased toward present consumption.
    - Incumbents do not fully weight welfare of owners of capital (foreign residents or non-insiders).
  - Prediction: sovereign borrowing “crowds out” private investment despite deep global savings; debt reduces investment and growth. Transition speed driven by political-economy frictions rather than technology.

- Empirical evidence supporting the alternative (Figures 3–5)
  - Public net foreign assets (public saving) vs growth: positive relationship stronger than for whole economy — countries that borrow more suffer from more severe present bias.
  - Private capital flows: average relationship negative — faster growth attracts private capital flows on net (consistent with neoclassical intuition for private flows).
  - Mean investment/GDP vs change in external public debt/GDP: countries relying more on external debt markets have lower investment rates.
  - Aggregate interpretation: evidence argues against sovereign debt markets increasing investment above domestic savings; data consistent with debt reducing investment.

- Caveat: extending the sample (1970–2021 and 2004–2021)
  - Benchmark sample: 1970–2004 (large run up in debt in emerging/developing economies).
  - Full sample: 1970–2021 — the positive relationship found in the earlier sample is attenuated.
  - Subsample 2004–2021: relationship has switched sign relative to benchmark.
  - Many large debtors in the benchmark sample shifted toward less indebted in the full sample due to debt forgiveness, default/restructuring, and commodity booms.
  - Additional considerations: domestic debt market development with foreign participation not captured by external debt measures; histories of borrowing/default/restructuring may matter.

- Sovereign debt as a volatility generator (Figures 9–12)
  - Method: compute standard deviations of annual growth in GDP, government expenditure, and private consumption over 1970–2004; correlate these volatilities with change in external public debt over this period.
  - Key empirical relationships:
    - Strong positive relationship between StDev(GDP per capita growth) and change in external debt/GDP.
    - Even stronger positive relationship between volatility in government expenditure and change in external debt/GDP.
    - Ratio StDev(Govt Expenditure Growth)/StDev(GDP Growth) increases with public debt accumulation.
    - Relative volatility of consumption to income (StDev Cons Growth/StDev Y Growth) vs change in public debt/GDP: relationship not nearly as strong; no indication debt reduces relative volatility of private consumption.
  - Interpretation: sovereign borrowing associated with higher volatility of income and higher relative volatility of public and private consumption. Reverse causation acknowledged but long-sample logic suggests policy choice plays a role.

- Taking stock: implications and policy perspectives
  - Empirical summary: sovereign debt generates (or at best is associated with) slower growth and more volatility — opposite to neoclassical promise but consistent with models of debt overhang and political-economy frictions.
  - Two policy/interpretation responses:
    - “Double down” on neoclassical promise: correct market inefficiencies and provide debt/fiscal guidelines to recover benefits of sovereign debt markets.
    - View inefficiencies as second-best: poorly working debt markets may help constrain political-economy frictions; imposing limits on government borrowing could be desirable even if limits stem from correctable frictions.
  - Next step: investigate premises using a canonical sovereign debt model.

### 3.1 A View from the Standard Quantitative Model

- Model setup and key assumptions
  - Based on Chatterjee and Eyigungor (2012) and the “EG-LT” calibration in Aguiar, Amador, and Monteiro (2023).
  - Main ingredients: stochastic endowment, no investment, default that is costly but strategic, and an impatient decision maker relative to an international risk-free interest rate R★.
  - Interpretative notes:
    - Absence of investment biases results in favor of sovereign borrowing given the Allocation Puzzle and debt crowding out.
    - Default costs are crucial but hard to measure; empirical work suggests large deadweight costs of default, with costs falling disproportionately on poorer households.
    - Government impatience used as reduced-form proxy for political-economy distortions.
    - Strategic default is an assumption; alternative creditor-run forced default (self-fulfilling runs) considered later.

- Quantitative implications and predicted moments (ergodic distribution) — Table 1 predicted moments
  - B/Y: 17.5%
  - Default Frequency: 7% per annum
  - Mean r − r★: 8.4 %
  - StDev r − r★: 4.6%
  - σ(ln c)/σ(ln y): 1.11
  - ρ(TB/Y, Y): -0.66
  - Calibration targets and fit: predictions broadly in line with many emerging markets and extreme cases such as Argentina.

- Welfare comparison: private agents versus sovereign access to debt (Figure 13)
  - Research question: do private agents prefer financial autarky (no sovereign international borrowing) to equilibrium with government access to sovereign debt markets?
  - Method:
    - Fix government preference parameterization.
    - Compute equilibrium allocation from zero debt.
    - Value the allocation under alternative private-agent preferences varying subjective discount rate and coefficient of relative risk aversion.
  - Calibration point G (government preferences):
    - World risk-free interest rate: 4% in annual terms
    - Government discount rate: 19%
    - Government coefficient of risk aversion: 2
  - Main result (Figure 13):
    - Parameter space divided by a solid diagonal: northwest region = private agents prefer financial autarky; southeast region = private agents prefer access to debt markets.
    - More risk averse private agents dislike access to debt markets because government borrowing increases consumption volatility via procyclical borrowing and lost output in default.
    - For modest levels commonly used in closed-economy macro (risk aversion 2–5 and discount rates 5–10%), private agents would prefer autarky.
  - Conclusion: in the standard model, access to debt markets by an impatient sovereign can be welfare reducing for private citizens under plausible preference disagreements.

- The value of a Lender of Last Resort (LoLR) and self-fulfilling runs (section 3.2)
  - Self-fulfilling rollover crisis logic:
    - With sufficient maturing debt, two equilibria possible: good equilibrium where lenders roll over; bad equilibrium where creditors do not participate and government cannot roll over and defaults — a coordination failure independent of fundamentals.
  - LoLR role:
    - By credibly promising to buy bonds in a failed auction, a LoLR can eliminate the bad equilibrium; if credible, runs become non-supportable and LoLR need never actually intervene along equilibrium path.
    - Practical difficulty: distinguishing self-fulfilling runs from fundamental defaults; risk of bailing out fundamentally insolvent debtors.
  - Modeling choices and extremes:
    - Abstract from information problems: LoLR endowed with full information so it can distinguish runs from fundamental default (favors LoLR).
    - Two short-term bond models computed: Rollover Model (with self-fulfilling runs) and LoLR Model.
    - Extreme modeling assumptions to make runs possible:
      - Debt maturity of one period (shortest discrete-time maturity).
      - Government quarterly discount factor set to 0.85 (annual discount rate of 65%).

- Short-term debt model moments — Table 2: With and Without LoLR
  - B/Y:
    - Rollover Model: 7%
    - LoLR Model: 16%
  - Default Frequency:
    - Rollover Model: 1.9% per annum
    - LoLR Model: 1.4% per annum
  - Mean r − r★:
    - Rollover Model: 2.0 %
    - LoLR Model: 1.5%
  - StDev r − r★:
    - Rollover Model: 1.2%
    - LoLR Model: 1.0%
  - σ(ln c)/σ(ln y):
    - Rollover Model: 1.07
    - LoLR Model: 1.20
  - ρ(TB/Y, Y):
    - Rollover Model: -0.19
    - LoLR Model: -0.16
  - Share Defaults from Runs:
    - Rollover Model: 100%
    - LoLR Model: 0%
  - Interpretation:
    - Without LoLR, equilibrium prices incorporate run risk, deterring debt accumulation (lower B/Y).
    - Rollover Model: all defaults driven by self-fulfilling panics.
    - LoLR Model: more borrowing and non-negligible fundamental defaults (moral hazard), though LoLR never pays out in equilibrium.
    - Equilibrium without runs is constrained efficient (planning problem tailored to government preferences); runs can reduce government’s ability to reach its preferred allocation and thus can mitigate welfare losses from preference disagreement.

- Practical policy implications (Section 4)
  - Broad takeaway:
    - From data and simple models, the value of sovereign debt markets to borrowing countries is hard to identify; with modest preference disagreements, access to debt markets can be welfare reducing for private citizens.
    - Even a seemingly costless LoLR with perfect information may not be welfare improving ex ante.
  - Practical considerations:
    - Prohibiting sovereign borrowing is not a standard policy tool for multilateral organizations.
    - Ex post vs ex ante: LoLR intervention beneficial ex post during a run, but credible commitment to refrain from ex post intervention is hard to achieve and needed for ex ante welfare gains.
    - Designing a LoLR requires balancing preventing runs against moral hazard and wasteful bailouts; analysis suggests raising the threshold for interventions may be warranted.
  - Policy analogies and recommendations:
    - Macro-prudential analogy: private efficient borrowing may not be socially efficient; policies that make debt markets less privately efficient (e.g., taxes on borrowing or debt limits) may be socially optimal.
    - No comparable sovereign-debt toolkit exists, but logic suggests erring on side of under-correcting sovereign debt market inefficiencies.
    - Post-default restructurings: ameliorating initial disruption is clearly beneficial; rapid re-entry to global markets may be lower priority relative to immediate stabilization.
  - Overall message for policy analysis:
    - Any cost-benefit calculus for interventions should explicitly consider whether the resulting equilibrium aligns with private-agent welfare or moves equilibrium further from that desired by private agents.

*Source: mundell-fleming-lecture*

### 2.1  The Neoclassical Paradigm meets the Allocation Puzzle

### 2.1  The Neoclassical Paradigm meets the Allocation Puzzle

### Neoclassical conventional wisdom: promised benefits of external bond markets
- Two main benefits posited:
  - Relax the saving=investment constraint of a closed economy by allowing access to global savings to fund investment and speed the transition to the steady state of the neoclassical growth model.
  - Insurance: sovereign debt markets enable risk sharing and can be used to “smooth” consumption or government expenditure relative to a volatile aggregate income process.
- Acknowledged frictions in international debt markets: (i) limited commitment; (ii) limited state contingency; (iii) rollover risk; and (iv) deadweight costs of default (loss of reputation, declines in trade or output, increases in inequality).
- Many papers seek to mitigate these frictions to deliver the promised benefits; the author examines this premise skeptically.

### Neoclassical implication: debt and capital as complements
- Representative constraint example: 퐵≤휈퐾 (Cohen and Sachs (1986); Barro, Mankiw, and Sala-I-Martin (1995)), where debt is collateralized by physical capital and 휈>0 is the allowable leverage ratio.
- Prediction: sovereign borrowing increases physical capital beyond private saving; debt and capital increase together along a transition path, with transition speed dictated by technological constraints and bounded below by the (counter-factually fast) closed economy transition rate.
- No distinction between public and private debt in the neoclassical allocation (Ricardian representative agent).

### The Allocation Puzzle (empirical contradiction)
- Gourinchas and Jeanne (2013) documented that emerging and developing economies that grew relatively fast exported savings on net, while countries that lagged behind imported capital — termed the “Allocation Puzzle.”
- Empirical replication (Figure 2): each point is a country over 1970-2004.
  - Horizontal axis: difference in net foreign assets as a ratio to GDP over 1970-2004, expressed in annualized changes (Net foreign assets from Lane and Milesi-Ferretti (2018)).
  - Vertical axis: growth in GDP per capita over 1970-2004, annualized changes and relative to the growth rate of the US during this time frame.
  - The scatter depicts an upward sloping relationship: countries that grew the fastest were net exporters of savings.
- This pattern is difficult to square with the standard neoclassical logic.

### Alternative paradigm: Growth in the Shadow of Expropriation
- Replace collateral constraint 퐵≤휈퐾 with incentive/political constraint 푊_G(퐵)≥푊_D(퐵), where:
  - 푊_G is the value of the political incumbent as a function of the stock of external debt 퐵.
  - 푊_D is the value of “deviating” or defaulting on debt promises, as a function of the physical capital stock 퐾.
- Two interpretations:
  - Trigger-strategy view: repayment is credible if 푊_G > 푊_D; 푊_G decreases in 퐵 (more debt reduces resources available for incumbent consumption); 푊_D increases in 퐾 (more domestic capital reduces punishment from losing access to credit), generating a negative relationship between debt and capital.
  - Expropriation/taxation view: more debt tempts the government to tax or expropriate capital, reducing private agents’ willingness to invest — “growth in the shadow of expropriation.”
- Political economy ingredients:
  - Government as sequence of stochastic political incumbents biased toward present consumption (spending while in power).
  - Incumbents do not fully weight the welfare of owners of capital (foreign residents or non-insiders).
- Prediction: sovereign borrowing “crowds out” private investment despite a deep pool of global savings; debt reduces investment and growth. Speed of transition driven by severity of political economy frictions rather than technology.

### Empirical evidence supporting the alternative (Figures 3–5)
- Figure 3 (analogous to Figure 2 but using public net foreign assets): horizontal axis is public net foreign assets (foreign reserves minus external public debt), over 1970-2004.
  - The positive relationship between public saving and growth is even stronger than for the economy as a whole.
  - Interpretation: countries that borrow more suffer from more severe present bias (Aguiar and Amador (2011)).
- Figure 4: private capital flows (total net flows minus public) — plots growth against change in private net foreign assets.
  - The average relationship is negative: faster growth attracts private capital flows on net (consistent with neoclassical intuition for private flows).
- Figure 5: mean investment/GDP (1970-2004) against change in external public debt/GDP.
  - Countries that rely more on external debt markets have lower investment rates.
- Aggregate interpretation: evidence argues against the belief that sovereign debt markets increase investment above domestic savings; rather, data consistent with debt reducing investment.

### Caveat: extending the sample (1970-2021 and 2004-2021)
- Figures 6–8 expand the sample:
  - Benchmark sample: 1970-2004 (large run up in debt in emerging/developing economies).
  - Full sample: 1970-2021 — the positive relationship found in the earlier sample is attenuated in the full sample.
  - Subsample 2004-2021: relationship has switched sign relative to benchmark, weakening the full-sample relationship.
- Figure 8: blue circles = 1970-2004; red diamonds = 1970-2021.
  - Many large debtors in the benchmark sample have shifted toward less indebted in the full sample due to debt forgiveness (e.g., Democratic Republic of Congo), default/restructuring (e.g., Argentina), and the commodity boom and associated GDP growth in the 2000s (e.g., Argentina).
- Additional considerations:
  - Development of domestic debt markets with increased foreign participation is not captured by external debt measures from the WDI.
  - The simple mechanism B↑ ⇒ K↓ visible in the benchmark sample requires care: the level of external debt may not be a sufficient state variable for expropriation risk; histories (borrowing/default/restructuring) may matter (see Fourakis (2023) for reputational dynamics).

### Sovereign debt as a volatility generator (Figures 9–12)
- Method: compute standard deviations of annual growth in GDP, government expenditure, and private consumption over 1970-2004; correlate these volatility measures with change in external public debt over this period.
- Figure 9: standard deviation of annual growth in GDP per capita vs change in external debt/GDP — strong positive relationship between volatility and debt.
- Figure 10: volatility in government expenditure vs change in external debt/GDP — even stronger positive relationship.
- Figure 11: ratio StDev(Growth in Government Expenditure)/StDev(Growth in GDP) vs change in public debt/GDP — more borrowing associated with an increase in government expenditure volatility above and beyond income volatility.
- Figure 12: relative volatility of consumption to income (StDev Cons Growth/StDev Y Growth) vs change in public debt/GDP — relationship not nearly as strong as for government consumption; no indication that debt reduces relative volatility of private consumption.
- Interpretation:
  - Sovereign borrowing is associated with higher volatility of income and higher relative volatility of public and private consumption.
  - Reverse causation concern (volatility driving borrowing) is acknowledged but counter-argument: over long samples, repeated large shocks constitute an underlying stochastic process; theory predicts governments facing volatile processes should accumulate buffer stock saving (Aiyagari et al., 2002). Failure to do so implies resulting expenditure volatility is at least partly a fiscal policy choice, not solely bad luck.

### Taking stock: implications and policy perspectives
- Empirical summary: sovereign debt generates (or, at best, is associated with) slower growth and more volatility — opposite to neoclassical conventional wisdom but consistent with models of debt overhang exacerbated by political economy frictions.
- Two policy/interpretation responses:
  - “Double down” on the neoclassical promise: correct inefficiencies in debt markets and provide debt/fiscal guidelines to governments to recover the original promise of sovereign debt markets.
  - View inefficiencies as a second-best positive outcome: poorly working debt markets may help constrain political economy frictions; more limits on government borrowing could be desirable even if limits stem from correctable market frictions.
- The author will further investigate these premises using a canonical sovereign debt model in the next section.

*Source: mundell-fleming-lecture - 2.1  The Neoclassical Paradigm meets the Allocation Puzzle*

### 3.1  A View from the Standard Quantitative Model

### 3.1  A View from the Standard Quantitative Model

### Model setup and key assumptions
- Builds on Chatterjee and Eyigungor (2012) and the “EG-LT” calibration in Aguiar, Amador, and Monteiro (2023).
- Main ingredients:
  - stochastic endowment
  - no investment
  - default that is costly but strategic
  - an impatient decision maker relative to an international risk-free interest rate 푅★
- Interpretative notes in the text:
  - Absence of investment biases results in favor of sovereign borrowing given the Allocation Puzzle and debt crowding out.
  - Default costs are crucial but hard to measure; empirical identification strategies (Hébert and Schreger (2017); Farah-Yacoub et al. (2022)) suggest large deadweight costs of default, with Farah-Yacoub et al. (2022) arguing costs fall disproportionately on poorer households.
  - Government impatience is used as a reduced-form proxy for political-economy distortions.
  - Strategic default is an open assumption; an alternative model with creditor-run forced default is discussed later (self-fulfilling runs).

### Quantitative implications and predicted moments (Table 1)
- Model generates procyclical bond prices and procyclical borrowing due to persistent endowment shocks and government impatience, producing excess consumption volatility observed in data.
- Table 1: Predicted Moments (ergodic distribution)
  - 퐵/푌: 17.5%
  - Default Frequency: 7% per annum
  - Mean 푟 − 푟★: 8.4 %
  - StDev 푟 − 푟★: 4.6%
  - 휎(ln푐)/휎(ln푦): 1.11
  - 휌(푇퐵/푌, 푌): -0.66
- Calibration targets and fit:
  - Model’s predictions are broadly in line with many emerging markets and extreme cases such as Argentina (target of calibration).

### Welfare comparison: private agents versus sovereign access to debt (Figure 13)
- Research question: Do private agents prefer financial autarky (no sovereign international borrowing) to equilibrium with government access to sovereign debt markets?
- Method:
  - Fix government preference parameterization.
  - Compute equilibrium allocation from zero debt.
  - Value the allocation under alternative private-agent preferences varying subjective discount rate and coefficient of relative risk aversion.
  - Compare value under equilibrium with debt to financial autarky.
- Preference dimensions and interpretations:
  - Discount rate heterogeneity: private households may be more patient than incumbent politicians.
  - Risk aversion heterogeneity: private agents near subsistence may be more risk averse than political incumbents; incumbents may underweight downside risk when borrowing to increase re-election odds.
- Calibration point G (government preferences):
  - World risk-free interest rate: 4% in annual terms
  - Government discount rate: 19%
  - Government coefficient of risk aversion: 2
- Main result (Figure 13):
  - Parameter space divided by a solid diagonal: northwest region = private agents prefer financial autarky; southeast region = private agents prefer access to debt markets.
  - More risk averse private agents dislike access to debt markets because government borrowing increases consumption volatility via procyclical borrowing and lost output in default.
  - Examples:
    - Point A: private agent’s preferences close enough to G; prefers government access to debt markets.
    - Point B: higher risk aversion and reduced discounting relative to A; prefers financial autarky.
  - For modest levels commonly used in closed-economy macro (risk aversion 2-5 and discount rates 5-10%), private agents would prefer autarky.
- Conclusion:
  - In the standard model, access to debt markets by an impatient sovereign can be welfare reducing for private citizens under plausible preference disagreements.

### The value of a Lender of Last Resort and self-fulfilling runs (section 3.2)
- Motivation:
  - Compare extreme of permanent autarky and equilibrium with access to debt; then ask whether improving debt-market efficiency (via a LoLR) is welfare enhancing.
- Self-fulfilling rollover crisis logic (Cole and Kehoe (2000) style):
  - With sufficient maturing debt, two possible equilibria for a government:
    - Good equilibrium: lenders buy new bonds; proceeds used to pay maturing bonds.
    - Bad equilibrium: creditors do not participate in auction; government cannot roll over and defaults, rationalizing non-participation; a coordination failure independent of fundamentals.
- Lender of Last Resort (LoLR) role:
  - By credibly promising to buy bonds in a failed auction, a LoLR can eliminate the bad equilibrium: runs become non-supportable and LoLR need never actually intervene along equilibrium path if promise is credible.
  - Practical difficulty: distinguishing self-fulfilling runs from fundamental defaults; potential for bailing out fundamentally insolvent debtors.
- Modeling choices and assumptions:
  - Abstract from information problems: model LoLR endowed with full information so it can distinguish runs from fundamental default (stacks deck in favor of LoLR).
  - Two versions of short-term bond model computed: one with self-fulfilling runs (Rollover Model) and one with LoLR (LoLR Model).
  - Extreme assumptions flagged:
    - Debt maturity of one period (shortest discrete-time maturity) maximizes rollover crisis risk.
    - Government quarterly discount factor set to 0.85 (annual discount rate of 65%) to allow borrowing into region where run is possible.
  - These extremes needed for runs to be possible in calibration; authors note real-world analogs in lumpy repayment schedules and willingness to risk a run.

### Short-term debt model moments (Table 2)
- Table 2: Moments of Short-Term Debt Model: With and Without LoLR
  - 퐵/푌:
    - Rollover Model: 7%
    - LoLR Model: 16%
  - Default Frequency:
    - Rollover Model: 1.9% per annum
    - LoLR Model: 1.4% per annum
  - Mean 푟 − 푟★:
    - Rollover Model: 2.0 %
    - LoLR Model: 1.5%
  - StDev 푟 − 푟★:
    - Rollover Model: 1.2%
    - LoLR Model: 1.0%
  - 휎(ln푐)/휎(ln푦):
    - Rollover Model: 1.07
    - LoLR Model: 1.20
  - 휌(푇퐵/푌, 푌):
    - Rollover Model: -0.19
    - LoLR Model: -0.16
  - Share Defaults from Runs:
    - Rollover Model: 100%
    - LoLR Model: 0%
- Interpretation:
  - Model without LoLR has lower average debt levels because equilibrium prices incorporate run risk, deterring debt accumulation.
  - In rollover model, all defaults are due to self-fulfilling panics.
  - LoLR model features more borrowing and a non-negligible number of defaults (fundamental defaults), reflecting potential moral hazard though LoLR never pays out in equilibrium.
  - Equilibrium without runs is constrained efficient (solves a planning problem maximizing joint surplus of government and lender subject to limited commitment and one-period bond restriction).
  - Constrained efficiency implies allocation tailored to government preferences; private-agent welfare may be lower when private preferences differ from government’s.
  - Presence of runs (and the constraints they impose) can mitigate welfare losses from disagreement by limiting government’s ability to reach its preferred allocation.

### Practical policy implications (Section 4)
- Broad takeaway:
  - From data and simple models, the value of sovereign debt markets to borrowing countries is hard to identify; with modest preference disagreements, access to debt markets can be welfare reducing for private citizens.
  - Even a seemingly costless LoLR with perfect information may not be welfare improving ex ante.
- Practical considerations:
  - Prohibiting sovereign borrowing is not within standard policy toolkits of multi-national organizations.
  - Ex post vs ex ante: in the midst of a run, LoLR intervention is clearly beneficial ex post, but credible commitment to refrain from ex post intervention is hard to achieve and needed for ex ante welfare gains.
  - Designing a LoLR facility requires balancing preventing runs against moral hazard and wasteful bailouts; analysis suggests raising the threshold for interventions may be warranted.
- Policy analogies and recommendations:
  - Macro-prudential (MacroPru) analogy: private efficient borrowing may not be socially efficient; policy that makes debt markets less privately efficient (e.g., taxes on borrowing or debt limits) may be socially optimal.
  - No comparable sovereign-debt toolkit exists, but logic suggests erring on side of under-correcting sovereign debt market inefficiencies.
  - Post-default restructurings: ameliorating initial economic disruption is clearly beneficial; re-entry to global markets and restructuring may be lower priority or deprioritized relative to immediate stabilization.
- Overall message for policy analysis:
  - Any cost-benefit calculus for interventions should explicitly consider whether the resulting equilibrium aligns with private-agent welfare or moves equilibrium further from that desired by private agents.

*Source: mundell-fleming-lecture - 3.1  A View from the Standard Quantitative Model (PDF chapter).*

---


_Source: https://www.imf.org/-/media/files/news/seminars/2023/arc/mundell-fleming-lecture.pdf_