## _wp08149

## Source details

**Canonical URL:** [_wp08149](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08149.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08149.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08149.pdf.json)

---

### I. Introduction — purpose and main conclusion
- Research question: What is the optimal amount of international reserves for countries with limited access to foreign capital, rationalized as insurance against large external shocks (natural disasters or terms of trade shocks)?
- Framework: intertemporal optimization for a small open economy that holds costly foreign reserves to smooth import fluctuations under a balance of payments constraint.
- Main finding:
  - The optimal reserves level is very sensitive to parameter calibration.
  - An average Caribbean country only needs 1.8 months of import.
  - An average Sahel country needs just over 4 months.
- Caveat: These estimates are a lower bound because countries may accumulate reserves for objectives other than self-insurance (providing liquidity, limiting exchange rate volatility, maintaining a fixed-peg, pursuing export-led growth via undervalued exchange rate).

### II. Natural disasters: empirical patterns and economic impacts
- Data source: natural disasters 1963-2007 (EMDAT).
- Method: five-year event window centered on a shock; measure behavior of real output growth, real export growth, real import growth, and change in the nominal exchange rate.
- Hurricanes in the Caribbean (averages reported):
  - A major hurricane hits a Caribbean country every 25 years, i.e. with a probability 96.0=ndπ each year.
  - Output growth falls by 3 percentage points.
  - Exports fall by 5 percentage points.
  - Imports decline by an average of only 1 percent.
  - Reserves-to-imports ratio growth declines by an average of 16 percent.
  - Example (Grenada, September 2004): Estimated damage over US$800 million; donor pledges over US$150 million; immediate liquidity available US$12 million.
- Droughts in the Sahel (averages reported):
  - Major drought definition: either 10 percent of the population affected or at least four droughts occurred during the year.
  - An average Sahel country faces one major drought every 12 years, i.e. with a probability 92.0=ndπ each year.
  - Output growth drops marginally by 0.3 percentage points on impact before rebounding the next year.
  - Exports growth drops by 8 percentage points.
  - Imports remain roughly constant.
  - Reserves-to-imports ratio growth rate falls by 17 percentage points.

### III. Model of optimal international reserves — structure and key mechanisms
- Economy: two countries (Home and Foreign); Home consumes home goods cH and foreign goods cF; Home exports fraction δ of its output Yt.
- Key constraints and notation preserved:
  - Balance-of-payments constraint: ttttFttF TrRRcc+−−≤+1*, ,ε
  - Aggregate resource constraint: ()() tttttFtH TrRRrRYcc−−+−=++1, ,11 εε
  - Opportunity cost: holding reserves pays rR (payable in home goods).
- Shocks:
  - Two-state Markov process (normal and disaster states); disaster reduces Y, exports and affects real exchange rate.
  - Normal: Yn and exports fraction cnF = δ.
  - Disaster: Yd = ηY·Yn, notation nY, dY, Yη, Xη, εη with 1<εη.
  - Disaster arrival probability ndπ and duration dnπ (with 1/dnπ as expected duration).
- Aid assumption: Foreign provides an exogenous and constant stream of aid Tr each period.
- Optimality condition (first-order condition, notation preserved):
  - Home chooses R to satisfy: ()'1,'1,'++−=tCtCttC HFF uruEuβ.
- Steady-state normal-state solution (equation preserved as (1)):
  - (1) ⎪⎩⎪⎨⎧+=−−= TrYc rRYYc n F nn H ε δ δ *

B. Approximated closed-form solution for reserves-to-imports ratio
- Under log-utility and when shocks are rare and have little persistence, Home optimally uses all reserves at once when a shock hits. The approximated closed-form solution is preserved:
  - (2) ()()() n X ndnd nd F Y Tr r c R ε δ η η θ θ π β δ ε δ π β β π ε + ⎥⎥⎥⎥⎦⎤ ⎢⎢⎢⎢⎣⎡ − − −−+ − − ≈ 1 1 1 11 1 1 *
  - (Approximation uses the fact that 1<<r.)
- Comparative statics from (2):
  - A higher shock probability (i.e. a higher ndπ), a larger drop in the value of exports (i.e. a higher Xη or εη), or a larger export sector (i.e. a higher δ) raises the optimal reserves-to-import ratio.
  - A higher opportunity cost of holding reserves (i.e. a higher r) lowers the reserves-to-import ratio.
  - A higher level of transfers Tr lowers the optimal reserves-to-import ratio.

### IV. Calibration and numerical solution — parameters and approach
- Temporal frequency: monthly frequency used for calibration.
- Preference specification: Stone-Geary utility (equation (3) notation preserved).
- Model parameters (11 parameters):
  - the probability of a disaster ndπ and its persistence dnπ,
  - the output loss Yη,
  - the exports loss Xη,
  - the exchange rate depreciation εη,
  - the opportunity cost of holding reserves r,
  - the coefficient of risk aversion ρ,
  - the preference for Foreign goods θ,
  - the size of the export sector δ,
  - the subsistence levels of consumption cH and cF.
- Sensitivity: small changes in key parameters (shock persistence, δ, ρ) can have large consequences on the optimal reserves level.
- Numerical solution: value function iteration used.

### V. Calibration details and main numerical results (monthly frequency)
- Common calibration inputs:
  - Monthly discount factor β = 0.9966.
  - Coefficient of risk aversion ρ = 5.
  - Subsistence levels of consumption fixed to zero unless otherwise noted.
  - Opportunity cost of holding reserves = 0.33 percent per month (or 4 percent a year).
- Selected calibrations and optimal reserves-to-imports ratios (*FcR, months of imports):
  - Caribbean — Hurricanes:
    - Probability πnd = 0.33% (25 years).
    - Persistence πdn = 0.83% (1.25 months).
    - Output loss ηy = 36%.
    - Exports loss ηx = 100% (10% yearly).
    - Terms of trade loss ηε = 0.
    - Exports-to-GDP ratio δ = 0.4.
    - Risk Aversion ρ = 5.
    - Subsistence level of imports Fc = 0%.
    - Transfers Tr = 10%.
    - Optimal Reserves-to-Imports ratio *FcR = 1.60 (months of imports).
  - Caribbean — Terms of Trade (ToT) shocks:
    - Probability πnd = 0.55% (17 years).
    - Persistence πdn = 0.08% (1 year).
    - Output loss ηy = 0.
    - Exports loss ηx = 0.
    - Terms of trade loss ηε = 10%.
    - Exports-to-GDP ratio δ = 0.4.
    - Risk Aversion ρ = 5.
    - Subsistence level of imports Fc = 0%.
    - Transfers Tr = 10%.
    - Optimal Reserves-to-Imports ratio *FcR = < 0.01 (months of imports).
  - Sahel — Droughts:
    - Probability πnd = 0.66% (12 years).
    - Persistence πdn = 0.16% (6 months).
    - Output loss ηy = 1%.
    - Exports loss ηx = 16% (8% yearly).
    - Terms of trade loss ηε = 0.
    - Exports-to-GDP ratio δ = 0.2.
    - Risk Aversion ρ = 5.
    - Subsistence level of imports Fc = 26% (in % of initial GDP).
    - Transfers Tr = 10%.
    - Optimal Reserves-to-Imports ratio *FcR = 2.05 (months of imports).
  - Sahel — Terms of Trade shocks:
    - Probability πnd = 0.88% (10 years).
    - Persistence πdn = 0.08% (1 year).
    - Output loss ηy = 0.5%.
    - Exports loss ηx = 5%.
    - Terms of trade loss ηε = 15%.
    - Exports-to-GDP ratio δ = 0.2.
    - Risk Aversion ρ = 5.
    - Subsistence level of imports Fc = 26% (in % of initial GDP).
    - Transfers Tr = 10%.
    - Optimal Reserves-to-Imports ratio *FcR = 2.46 (months of imports).
- Combined shocks (three-state Markov):
  - Caribbean optimal reserves-to-imports ratio = 1.78 (months of imports).
  - Sahel optimal reserves-to-imports ratio = 4.13 (months of imports).

### VI. Results — mechanisms, sensitivity, and simulated dynamics
- Caribbean — Hurricanes (baseline):
  - Baseline socioeconomic shares: Exports = 30 percent of output; Imports = 40 percent of output; Transfers = 10 percent of output.
  - Hurricane calibration: major hurricane every 25 years; exports full stop for 1.25 months to match a total exports loss of 10 percent; output drops by 36 percent; real exchange rate kept constant.
  - Optimal reserves: 1.60 months of imports.
  - Sensitivity findings:
    - Optimal reserves-to-imports ratio increases with shocks’ probability and persistence.
    - Relationship with export sector size is inverse U-shaped.
    - Subsistence level of imports: 10 percent of GDP implies reserves of almost 3 months of import.
    - Optimal reserves increase with degree of risk aversion.
- Sahel — Droughts (baseline):
  - Baseline socioeconomic shares: Exports = 20 percent of output; Imports = 30 percent of output; Transfers = 10 percent of output.
  - Drought calibration: major drought every 12 years; exports drop by 16 percent for 6 months; output largely unaffected annually; real exchange rate kept constant.
  - Subsistence level of imports set at 80 percent of “normal” imports = 26 percent of “normal” GDP.
  - Note: if subsistence level set to zero, optimal reserves-to-import ratio close to zero at 0.08 months.
  - Optimal reserves: ≈ 2.05 months of imports.
  - Mechanism: reserves used progressively; long duration brings consumption close to subsistence, strongly justifying reserves.
  - Sensitivity: subsistence level of imports is a key determinant; larger export/import sector lowers optimal reserves-to-imports ratio.
- Terms of Trade shocks:
  - Empirical calibrations:
    - Caribbean: ten percent decline in terms of trade every 17 years with no significant cost to output or exports growth.
    - Sahel: fifteen percent decline every 10 years with a 5 percent decline in exports growth.
  - Model assumptions for ToT shocks: lasts for a year; no other impact on output and exports except depreciation in ε.
  - Optimal reserves implied:
    - Caribbean: less than 0.01 months of import.
    - Sahel: 2.5 months of imports.
  - Interpretation: Sahel reserves driven by subsistence-consumption effects; Caribbean welfare losses small so minimal reserves needed.
- Combined shocks (simulated dynamics):
  - Caribbean dynamics:
    - After a ToT shock, imports initially kept close to normal; reserves usage slowed to preserve capacity against hurricanes when ToT shock duration exceeds expectation.
    - After a hurricane, exports drop by 50 percentage points but imports decline by only 10 percentage points due to rapid use of reserves.
  - Sahel dynamics:
    - Droughts and ToT shocks milder but frequent; declines in foreign exchange inflows more persistent.
    - Sahel countries do not slow down reserves usage progressively; they can smooth reserves accumulation at a slower pace.
- Key quantitative sensitivities:
  - Optimal reserves very sensitive to parameter calibration.
  - Caribbean optimal reserves in baseline calibration can vary between 0 and 2 months of imports depending on export sector size.
  - Shock persistence of 2 months calls for roughly 3 months of imports in reserves; persistence of 3 months calls for 5 months.
  - Opportunity cost of holding reserves negligible for countries with a small export sector; becomes determinant when exports represent a large share of GDP.
  - Average reserves over time can differ from steady-state optimal level: Sahel steady-state above 4 months, average over 16 years of simulation = 3.2 months.

### VII. Policy implications, recommendations, and scenarios
- Calibration-critical policy advice:
  - The model requires careful country-specific calibration; small changes in key parameters substantially affect optimal reserves.
- Practical benchmarks (calibrated examples):
  - Caribbean average country: ~1.8 months of imports as optimal reserves.
  - Sahel average country: ~4.1 months of imports as optimal reserves.
- Policy levers and considerations:
  - Prioritize accurate measurement of:
    - Return period and persistence of natural disasters and ToT shocks.
    - Exports-to-GDP ratio and composition of export sector.
    - Subsistence consumption levels.
    - Opportunity cost of holding reserves.
  - Account for heterogeneous effects across countries.
  - Recognize historical average reserves may understate optimal steady-state targets; simulate dynamic paths when assessing adequacy.

### VIII. Conclusion and suggested extensions
- Main conclusion:
  - Optimal reserves level as insurance against large external shocks is highly sensitive to parameter calibration; country-specific, carefully calibrated analysis is necessary.
- Suggested extension:
  - Apply framework to evaluate optimal size of Sovereign Wealth Funds for commodity-reliant economies to smooth price fluctuations and provide alternative revenue sources.

*Italic: Source — _wp08149 - Section II on the impact of disasters and terms of trade shocks, I can calibrate the parameters (PDF chapter/section).*

### Appendix 1.  List of Countries in Each Group ........................................................................21

### Appendix 1.  List of Countries in Each Group

### I. Introduction — purpose and main conclusion
- Research question: What is the optimal amount of international reserves for countries with limited access to foreign capital, rationalized as insurance against large external shocks (natural disasters or terms of trade shocks)?
- Framework: intertemporal optimization for a small open economy that holds costly foreign reserves to smooth import fluctuations under a balance of payments constraint.
- Main finding:
  - The optimal reserves level is very sensitive to parameter calibration.
  - An average Caribbean country only needs 1.8 months of import.
  - An average Sahel country needs just over 4 months.
- Caveat: These estimates are a lower bound because countries may accumulate reserves for objectives other than self-insurance (providing liquidity, limiting exchange rate volatility, maintaining a fixed-peg, pursuing export-led growth via undervalued exchange rate).

### II. Natural disasters: empirical patterns and economic impacts
- Data source: natural disasters 1963-2007 (EMDAT).
- Method: five-year event window centered on a shock; measure behavior of real output growth, real export growth, real import growth, and change in the nominal exchange rate.

A. Hurricanes in the Caribbean
- Example (Grenada, September 2004):
  - Estimated damage: over US$800 million—or twice Grenada’s GDP.
  - Donor pledges: over US$150 million; immediate liquidity available: only US$12 million.
- Frequency and impacts (averages reported):
  - A major hurricane hits a Caribbean country every 25 years, i.e. with a probability 96.0=ndπ each year.
  - Output growth falls by 3 percentage points.
  - Exports fall by 5 percentage points.
  - Imports decline by an average of only 1 percent.
  - Reserves-to-imports ratio growth declines by an average of 16 percent.

B. Droughts in the Sahel
- Classification: droughts defined as “major” when either 10 percent of the population is affected or when at least four droughts occurred during the year.
- Frequency and impacts (averages reported):
  - An average Sahel country faces one major drought every 12 years, i.e. with a probability 92.0=ndπ each year.
  - Output growth drops only marginally by 0.3 percentage points on impact before rebounding the next year.
  - Exports growth drops by 8 percentage points.
  - Imports remain roughly constant.
  - Reserves-to-imports ratio growth rate falls by 17 percentage points.

### III. Model of optimal international reserves — structure and key mechanisms
- Economy: two countries (Home and Foreign); Home consumes home goods cH and foreign goods cF; Home exports fraction δ of its output Yt.
- Balance-of-payments constraint (as presented in text):
  - At each date t: ttttFttF TrRRcc+−−≤+1*, ,ε  (notation preserved as in source)
  - Aggregate resource constraint: ()() tttttFtH TrRRrRYcc−−+−=++1, ,11 εε  (notation preserved)
- Opportunity cost: holding reserves pays rR (payable in home goods).
- Shocks: two-state Markov process (normal and disaster states); disaster reduces Y, exports and affects real exchange rate. Notation preserved:
  - In normal: Yn and exports fraction cnF = δ.
  - In disaster: Yd = ηY·Yn, Xη, εη etc. (using source notation: nY, dY, Yη, Xη, εη with 1<εη).
  - Disaster arrival probability and duration: ndπ and dnπ (with 1/dnπ as expected duration).
- Aid assumption: Foreign provides an exogenous and constant stream of aid Tr each period; immediate post-disaster foreign-exchange losses must be covered by reserves.
- Optimality condition (first-order condition, notation preserved):
  - Home chooses R to satisfy: ()'1,'1,'++−=tCtCttC HFF uruEuβ.

- Steady-state normal-state solution (equation label preserved as (1)):
  - (1) ⎪⎩⎪⎨⎧+=−−= TrYc rRYYc n F nn H ε δ δ *

B. Approximated closed-form solution for reserves-to-imports ratio
- Under log-utility (u = ln cH + (1−θ) ln cF) and when shocks are rare and have little persistence, Home optimally uses all reserves at once when a shock hits. The approximated closed-form solution (notation and structure preserved) is:
  - (2) ()()() n X ndnd nd F Y Tr r c R ε δ η η θ θ π β δ ε δ π β β π ε + ⎥⎥⎥⎥⎦⎤ ⎢⎢⎢⎢⎣⎡ − − −−+ − − ≈ 1 1 1 11 1 1 *
  - (Approximation uses the fact that 1<<r.)
- Comparative statics drawn from (2) (intuitive conclusions preserved exactly):
  - A higher shock probability (i.e. a higher ndπ), a larger drop in the value of exports (i.e. a higher Xη or εη), or a larger export sector (i.e. a higher δ) raises the optimal reserves-to-import ratio.
  - A higher opportunity cost of holding reserves (i.e. a higher r) lowers the reserves-to-import ratio.
  - A higher level of transfers Tr (official loans or grants, and private remittances) in steady-state lowers the optimal reserves-to-import ratio.

### IV. Calibration and numerical solution — parameters and approach
- Temporal frequency: monthly frequency is used for calibration to capture urgency and immediate import shortages in the weeks/months after disasters.
- Preference specification for calibration:
  - Stone-Geary specification: utility as in equation (3) (notation preserved):
    - (3) ()()()() ρ θ ρ θ ρρ − −− −+ − −− = −− 1 1 1 1 1 ,1,1,, FtFHtH tFtH cccc cc u
  - ρ is relative risk aversion; θ is preference for home goods; cH and cF are subsistence consumption levels.
- Model parameters: the optimal level of reserves depends on 11 parameters (explicitly listed in the source):
  - the probability of a disaster ndπ and its persistence dnπ,
  - the output loss Yη,
  - the exports loss Xη,
  - the exchange rate depreciation εη,
  - the opportunity cost of holding reserves r,
  - the coefficient of risk aversion ρ,
  - the preference for Foreign goods θ,
  - the size of the export sector δ,
  - the subsistence levels of consumption cH and cF.
- Sensitivity: small changes in key parameters (shock persistence, size of export sector δ, degree of risk aversion ρ) can have large consequences on the optimal reserves level.

_Appendix 1.  List of Countries in Each Group — _wp08149 - Appendix 1.  List of Countries in Each Group_ (source PDF content)_.

### Section II on the impact of disasters and terms of trade shocks, I can calibrate the parameters

### Section II on the impact of disasters and terms of trade shocks, I can calibrate the parameters

### Calibration and model setup
- Monthly discount factor β = 0.9966.
- Coefficient of risk aversion ρ = 5.
- Subsistence levels of consumption fixed to zero unless otherwise noted.
- Opportunity cost of holding reserves = 0.33 percent per month (or 4 percent a year).
- Value function iteration used to solve the model numerically.
- Specification uses Stone-Geary utility; relative risk aversion for foreign goods equals FF/(F c − c) ρ (as described in text).

- Table 1 (selected calibration parameters, monthly frequency):
  - Caribbean — Hurricanes:
    - Probability of disaster (Return period) πnd = 0.33% (25 years).
    - Persistence (Duration) πdn = 0.83% (1.25 months).
    - Output loss ηy = 36%.
    - Exports loss ηx = 100% (10% yearly).
    - Terms of trade loss ηε = 0.
    - Exports-to-GDP ratio δ = 0.4.
    - Risk Aversion ρ = 5.
    - Subsistence level of imports Fc = 0%.
    - Transfers Tr = 10%.
    - Optimal Reserves-to-Imports ratio *FcR = 1.60 (months of imports).
  - Caribbean — Terms of Trade (ToT) shocks:
    - Probability πnd = 0.55% (17 years).
    - Persistence πdn = 0.08% (1 year).
    - Output loss ηy = 0.
    - Exports loss ηx = 0.
    - Terms of trade loss ηε = 10%.
    - Exports-to-GDP ratio δ = 0.4.
    - Risk Aversion ρ = 5.
    - Subsistence level of imports Fc = 0%.
    - Transfers Tr = 10%.
    - Optimal Reserves-to-Imports ratio *FcR = < 0.01 (months of imports).
  - Sahel — Droughts:
    - Probability πnd = 0.66% (12 years).
    - Persistence πdn = 0.16% (6 months).
    - Output loss ηy = 1%.
    - Exports loss ηx = 16% (8% yearly).
    - Terms of trade loss ηε = 0.
    - Exports-to-GDP ratio δ = 0.2.
    - Risk Aversion ρ = 5.
    - Subsistence level of imports Fc = 26% (in % of initial GDP).
    - Transfers Tr = 10%.
    - Optimal Reserves-to-Imports ratio *FcR = 2.05 (months of imports).
  - Sahel — Terms of Trade shocks:
    - Probability πnd = 0.88% (10 years).
    - Persistence πdn = 0.08% (1 year).
    - Output loss ηy = 0.5%.
    - Exports loss ηx = 5%.
    - Terms of trade loss ηε = 15%.
    - Exports-to-GDP ratio δ = 0.2.
    - Risk Aversion ρ = 5.
    - Subsistence level of imports Fc = 26% (in % of initial GDP).
    - Transfers Tr = 10%.
    - Optimal Reserves-to-Imports ratio *FcR = 2.46 (months of imports).

- Combined shocks (shocks combined):
  - Caribbean optimal reserves-to-imports ratio = 1.78 (months of imports).
  - Sahel optimal reserves-to-imports ratio = 4.13 (months of imports).

### A. Self-insurance against natural disasters — Caribbean (hurricanes)
- Baseline socioeconomic shares:
  - Exports = 30 percent of output.
  - Imports = 40 percent of output.
  - Transfers = 10 percent of output.
- Disaster calibration:
  - Major hurricane hits every 25 years.
  - Hurricane assumed to bring exports to a full stop for an estimated average of one month and a quarter (1.25 months) to match a total exports loss of 10 percent identified in Section II.
  - Output drops by 36 percent during the disaster.
  - Real exchange rate kept constant during disasters (fixed exchange-rate regimes prevalent).
- Optimal reserves:
  - Optimal level of reserves covers 1.60 months of imports.
  - Observations from sensitivity analysis (Figure 5):
    - Optimal reserves-to-imports ratio increases with shocks’ probability (return period) and shock persistence.
    - Relationship between size of export sector and optimal reserves is inverse U-shaped:
      - Initially, larger export sector raises optimal reserves-to-imports ratio (easier accumulation due to higher foreign-exchange inflows).
      - Beyond a threshold, opportunity cost of holding reserves becomes non-negligible and reduces optimality.
    - Optimal reserves increase with subsistence level of imports: subsistence level of 10 percent of GDP implies reserves of almost 3 months of import.
    - Optimal reserves increase with degree of risk aversion.

### A. Self-insurance against natural disasters — Sahel (droughts)
- Baseline socioeconomic shares:
  - Exports = 20 percent of output.
  - Imports = 30 percent of output.
  - Transfers = 10 percent of output.
- Disaster calibration:
  - Major drought every 12 years.
  - Droughts cause little physical damage but are more persistent; annual output growth assumed unaffected while exports drop by 16 percent for 6 months (consistent with an 8 percent annual decline documented in Section II).
  - Most Sahel countries in CFA Franc zone; real exchange rate kept constant during disasters.
  - Subsistence level of imports set at 80 percent of “normal” imports = 26 percent of “normal” GDP (calibrated to 2008 riots triggered by 50 percent food price increase and food basket representing roughly 40 percent of imports).
  - Note: if subsistence level of imports set to zero, optimal reserves-to-import ratio close to zero at 0.08 months.
- Optimal reserves:
  - Optimal reserves-to-import ratio ≈ 2.05 months of imports.
  - Mechanism: reserves used progressively to minimize decline in imports over drought duration; long duration brings consumption close to subsistence (famine risk), which strongly justifies holding reserves.
  - Sensitivity (Figure 6):
    - Key determinant is subsistence level of imports: as subsistence approaches import levels during droughts, optimal reserves increase sharply.
    - Larger export/import sector lowers optimal reserves-to-imports ratio as consumption moves away from subsistence levels and droughts become less costly.

### B. Self-insurance against terms of trade shocks
- Empirical calibrations:
  - Average Caribbean country: ten percent decline in terms of trade every 17 years with no significant cost to output or exports growth.
  - Average Sahel country: fifteen percent decline every 10 years with a 5 percent decline in exports growth.
- Model assumptions for ToT shocks:
  - Terms of trade shock lasts for a year.
  - No other economic impact on output and exports except depreciation in ε.
- Optimal reserves implied:
  - Caribbean: less than 0.01 months of import.
  - Sahel: 2.5 months of imports.
- Interpretation:
  - In Sahel, prolonged consumption of foreign goods close to subsistence during ToT shocks motivates substantial reserves holding.
  - In Caribbean, ToT shock lowers imports from 50 percent of GDP to slightly less than 45 percent — a small welfare loss given concavity of utility; minimal reserves needed.

### C. Self-insurance against combined natural disasters and terms of trade shocks
- Model extension:
  - Three-state Markov process: “normal”, “terms of trade shock”, or “natural disaster”.
  - Transition probabilities as previously calibrated.
- Optimal reserves (combined):
  - Caribbean = 1.78 months of imports.
  - Sahel = 4.13 months of imports.
- Simulated dynamics and behavioral differences:
  - Caribbean (Figure 7a):
    - After a ToT shock, imports initially kept close to normal; reserves usage slowed to preserve capacity against hurricanes when ToT shock duration exceeds expectation (example: shock lasting 3.5 years instead of 1 year).
    - After a hurricane, exports drop by 50 percentage points but imports decline by only 10 percentage points due to rapid use of reserves.
  - Sahel (Figure 7b):
    - Droughts and ToT shocks are milder but frequent; declines in foreign exchange inflows more persistent.
    - Sahel countries do not slow down reserves usage progressively as Caribbean countries do; they can smooth reserves accumulation at a slower pace.
- Key quantitative sensitivity points:
  - Optimal reserves very sensitive to parameter calibration.
  - Example ranges:
    - Caribbean optimal reserves in baseline calibration can vary between 0 and 2 months of imports depending on export sector size.
    - Shock persistence of 2 months calls for roughly 3 months of imports in reserves; persistence of 3 months calls for 5 months.
  - Opportunity cost of holding reserves:
    - Negligible for countries with a small export sector.
    - Determinant when exports represent a large share of GDP.
  - Average reserves over time can differ from steady-state optimal level:
    - Example: Sahel optimal steady-state reserves-to-import ratio above 4 months, but average reserves level over 16 years of simulation = 3.2 months.

### Policy implications, recommendations, and scenarios
- Calibration-critical policy advice:
  - The model requires careful country-specific calibration; small changes in key parameters (size and persistence of shocks, importance of export sector, degree of risk aversion) substantially affect optimal reserves.
- Practical benchmarks:
  - Rules of thumb provide imprecise guidance; example calibrated benchmarks:
    - Caribbean average country: ~1.8 months of imports as optimal reserves.
    - Sahel average country: ~4.1 months of imports as optimal reserves.
- Policy levers and considerations:
  - Prioritize accurate measurement of:
    - Return period and persistence of natural disasters and ToT shocks.
    - Exports-to-GDP ratio and composition of export sector.
    - Subsistence consumption levels (especially where population lives near subsistence).
    - Opportunity cost of holding reserves.
  - Account for heterogeneous effects: parameters that are critical in one country may be irrelevant in another.
  - Recognize that historical average reserves may understate optimal steady-state targets; simulate dynamic paths when assessing adequacy.

### Conclusion and extensions
- Main conclusion:
  - Optimal reserves level as insurance against large external shocks is highly sensitive to parameter calibration; country-specific, carefully calibrated analysis is necessary.
- Suggested extension:
  - Apply framework to evaluate optimal size of Sovereign Wealth Funds for commodity-reliant economies:
    - Rationale: natural resource income provides large foreign exchange inflows but price volatility and uncertainty about resource amounts call for accumulation to smooth price fluctuations and provide alternative revenue sources.

*Italic: Source — _wp08149 - Section II on the impact of disasters and terms of trade shocks, I can calibrate the parameters (PDF chapter/section).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08149.pdf_