## _wp04175

## Source details

**Canonical URL:** [_wp04175](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04175.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04175.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04175.pdf.json)

---

### I. Introduction and observed patterns
- Global reserves (in percent of world GDP) have exhibited an upward trend since the 1960s (Flood and Marion (2001); Edison (2003)).
- Over the same period, exchange rate flexibility has increased; no evidence has been put forward that exchange rate flexibility has decreased (Reinhart and Rogoff, 2003).
- Some countries accumulated international reserves of more than 20 percentof their GDPs by late 2003.
- The stock of international reserves held by China alone surpassed US$ 400 billion by end-2003, exceeding the capital of the International Monetary Fund.
- The capital of the IMF, measured by the paid-in quotas of member countries, amounted to SDR 212 billion (somewhat below US $ 300 billion) at end-2002.

### II. Conceptual framing: reserves as self-insurance
- Traditional view: demand for international reserves is a remnant of the fixed exchange rate era.
- Self-insurance view (dating to Heller (1966)): reserves are held as a cushion against undesired shortages of international liquidity that could damage the economy.
- Post–capital account–driven currency crisis discourse elevated the latent insurance view and proposals to maintain reserves large enough to provide against capital market–driven drains on international liquidity (see Mulder (2000) for references).

### III. Equivalence between reserves and financial options
- Holding reserves is interpreted as functionally equivalent to holding a put option providing identical insurance coverage.
- Merton (1976) discussed equivalence of insurance and a put option; by analogy the self-insurance value of reserves can be approximated by the market cost of obtaining equivalent insurance via option pricing theory.
- Formal notation and equivalence:
  - Time-t price of a put maturing at T with exercise price D: G(D, Vt, T, t).
  - Time-t insurance value of holding reserves C: I(C, Vt) ≡ G(C, Vt, T, t). (Equation (5))
  - Reserve coverage ratio: λ = C/D.
  - Insurance benefit of holding reserves: G(D, Vt, T, t) − G(˜D, Vt, T, t) where ˜D ≡ (1 − λ)D. (Equation (6))

### IV. Quantitative framework and optimal coverage
- Opportunity cost of holding reserves (approximation): cost = (R−r)(T−t) C. (Equation (7))
- Net benefit of self-insurance with τ ≡ T − t:
  - B(λ) = G(D, V, τ) − G(˜D, V, τ) − (R−r) τ (D − ˜D). (Equation (9))
- Marginal benefit and first-order condition:
  - MB(λ) ≡ ∂B(λ)/∂λ = D [ ∂G(˜D, V, τ)/∂˜D − (R−r) τ ]. (Equation (10))
  - Optimal coverage λ* solves ∂G(˜D, V, τ)/∂˜D = (R−r) τ. (Equation (11))
- Properties and comparative statics:
  - By monotonicity ∂G/∂˜D ≥ 0; meaningful solutions exist.
  - Full self-insurance (λ = 1, ˜D = 0) is optimal when R − r = 0 because MB(1) ≥ 0.
  - Implicit mapping H(λ*, D, R−r, V, σ) = 0 defines λ* as a function of D, R−r, V, σ. (Equation (12))
  - For given D, C = λ* D. (Equation (13))
  - λ* = λ*(σ, R−r | D). (Equation (14))

### V. Parameterization by European (Black–Scholes) option
- Rationale:
  - Black–Scholes closed-form European put is used for illustrative, tractable quantitative metrics; insurance value and optimal coverage depend on ratios (D or C to V) under Black–Scholes.
  - Limitations acknowledged: American options, credit derivatives, and macroeconomic insurance may yield different (often larger) cost estimates, but European option captures the lion’s share of market insurance value for illustrative analysis.
- Black–Scholes formula (log-normal asset process dVt / Vt = μ dt + σ dω(t)):
  - G(D, V, τ) = D e^{−rτ} N(x2) − V N(x1). (Equation (15))
  - x1 = [log(D/V) − (r + σ^2/2) τ] / (σ √τ); x2 = x1 + σ √τ.
- Average insurance value per unit of reserves:
  - I(C, V) / C = e^{−rτ} N(x2) − (V/C) N(x1) τ ≡ IA(C/V). (Equation (16))
  - IA depends only on C/V; IA increases in C/V and in σ.
- Marginal benefit in Black–Scholes:
  - ∂G(˜D)/∂˜D = e^{−rτ} N(˜x2).
  - MB(λ) = D [ e^{−rτ} N(˜x2) − (R−r) τ ]. (Equation (17))
  - MB(λ) decreases in λ; internal solution exists for 0 < (R−r) τ < 1.

### VI. Numerical findings (summary)
- Average insurance value calculations for C/V from 0.5 to 3 and σ from 0.05 to 0.5:
  - Volatility has little effect on average insurance value except when C = V; C/V is the primary driver.
- Optimal reserve coverage ratio (with D/V = 1) results:
  - λ* decreases in spread (R−r) and increases in volatility σ.
  - Elasticity examples:
    - For σ = 0.05 and 0.1, λ* falls by about one-half as spread rises ten-fold from 100 to 1,000 basis points.
    - For σ = 0.5, decline in λ* with spread is less sharp.
    - For low spreads (1 or 2 percent), λ* rises nearly tenfold as σ increases from 0.05 to 0.5.
    - For higher spreads (9 or 10 percent), λ* rises by more than tenfold over the same σ change.
  - Full coverage (λ = 1) rarely optimal; λ* < 0.5 for most parameter values except the upper-right corner.
  - Plausible central range for λ* is 20-50 percent (visualized as a diagonal bloc from upper left to lower right across parameter space).
- Selected table excerpts (preserved exactly):
  - Table 1 first-row values for C/V = 0.5 and volatilities: 0.000 0.000 0.000 0.000 0.001 0.008 0.023
  - Table 2 first-row (C/V = 0.1) across volatilities: 0.000 0.000 0.000 0.000 0.000 0.000 0.00
  - Table 3 sample first-row (spread = 0.01) across volatilities: 0.083 0.187 0.280 0.364 0.508 0.623 0.714
  - Table 4 sample first-row (spread = 0.0001) across volatilities: 0.167 0.328 0.458 0.566 0.728 0.821 0.885

### VII. Illustrative applications and policy implications
- A. Cost of an Asian Monetary Fund (AMF)
  - Empirical context:
    - Emerging-market economies in Asia accumulated large reserves; diagnosis of overaccumulation (Edison (2003)).
    - Chiang Mai Initiative (CMI) mutual swap agreement cumulative total: 35 billion dollars.
  - European option–based insurance cost per unit for institutional arrangement approximated by C/V = 1: average cost ranges between 0.01 and 0.1 when C/V = 1.
  - Applying those costs:
    - Insurance cost of reserves amounting to 10 percent of GDP would be between 0.1 and 1 percents of GDP.
    - Countries with excess reserve holdings around 20 percent of GDP carry insurance with market value (cost) ranging between 0.2 and 2.0 percents of GDP. Even on the low side of the range, this is probably too expensive to be a
  - Opportunity cost example:
    - Cost of holding reserves amounting to 20 percent of GDP: opportunity cost about 0.2 percent of GDP for countries with sovereign spreads around 100 basis points.
  - Comparatives:
    - European option–based estimate will be, if any, lower than cost estimates based on American options or customized derivatives.
    - For C/V = 0.5 (first row of Table 1): cost of insurance < 0.01 per unit for most parameter values; insurance amounting to 10 percent of GDP < 0.1 percent of GDP.
    - If C/V = 0.1: cost of insurance < 10−5 per unit; 10 percent of GDP insurance < one-thousandth percent of GDP.
- B. Raising the value of the underlying asset (lowering C/V)
  - Pooling: arranging options for subgroup use increases effective underlying asset value; pure benefit of pooling.
  - Mutual commitments with conditionality: emergency loan with conditionality expands effective underlying asset beyond pledged external assets; cost of insurance can be paid via forgone investment income on the capital (quota) pledged to the insurance pool.
  - Conditionality roles:
    - Ensures repayment ex post.
    - Lowers ex-ante insurance costs by expanding effective underlying asset.
    - Scope of conditionality should match source of risks; where private-sector structural risks dominate, structural conditionality beyond traditional macro policy is coherent and necessary.
  - Most cost-effective arrangements:
    - All participants subscribing to conditionality and sufficiently large participant number approximating perfect pooling minimizes cost.
    - Internal provision without market options becomes possible with large pooling; IMF-scale pooling is strictly more efficient than a smaller regional AMF.
- C. Excess reserves — advanced versus emerging markets
  - Working assumption: desired overall insurance (D) equals short-term external debt.
  - Observed ratios:
    - Advanced economies: actual reserve coverage ratios most often fall in the 20-40 percent range, close to calculated optimal reserve coverage ratios in the 20-50 percent range.
    - Emerging markets: observed ratios of reserves to short-term debt are much higher (often close to or higher than 100 percent) and rose sharply in several Asian countries after the Asian crisis.
  - Constructed excess-reserve measure (benchmark: about half of short-term external debt is optimal self-insurance):
    - Table 8 reports average reserve holding by emerging markets over 2000-2002, optimal reserve holding implied by average short-term external debt, and implied excess reserves (all in percent of GDP).
    - Sample Table 8 excerpt (preserved exactly): Argentina 6 9 1; Brazil 7 6 4; Chile 2 2 1 4 15; China 1 9 2 18; Korea 2 4 8 20; Singapore 8 9 7 0 54
  - Interpretation:
    - Excess reserves gap between emerging and advanced economies quantifies financial-market limitations confronting emerging markets (weak institutional development, policy credibility); expected to narrow slowly.
    - Short-term pragmatic solution: maintain reserves at least equal to short-term external debt; some argue expanding target to potential domestic-origin capital outflow.
    - CCL and precautionary arrangements: optimal ratio of reserves to size of credit line corresponds to λ* (Table 3). CCL recommended quota-to-CCL range: 1/5 and 1/3; these ratios are on the low side compared with numerical calculations where self-insurance ratio often exceeds 1/3. Stringent qualification requirements limited interest in CCL.

### VIII. Main conclusions and policy recommendations
- Key analytical conclusion:
  - The unobservable insurance value of holding external reserves can be quantified by the price of a put option with equivalent insurance coverage using European option parameterization.
- Policy implications:
  - Pure regional swapping arrangements are too costly to be implemented on a large scale (10 or 20 percent of GDP).
  - Arrangements adopting conditionality to increase the effective value of the underlying asset (IMF-like) can greatly reduce insurance arrangement costs at regional and global levels.
  - The theoretically derived optimal reserve coverage ratio is close to actual reserve holdings of advanced economies, while emerging markets hold reserves well above optimal ratios—quantifying the access gap.
  - The quantitative framework can aid narrowing the gap by evaluating tradeoffs between size of access and stringency of requirements.
- Numeric highlights (preserved exactly):
  - Opportunity cost for 20 percent of GDP with sovereign spread ~100 basis points: 0.2 percent of GDP.
  - C/V examples:
    - C/V = 0.5 → cost of insurance < 0.01; 10 percent of GDP insurance < 0.1 percent of GDP.
    - C/V = 0.1 → cost of insurance < 10−5 per unit; 10 percent of GDP insurance < one-thousandth percent of GDP.
  - Advanced economy reserve coverage ranges: 20-40 percent (observed); calculated optimal: 20-50 percent.
  - CCL recommended quota-to-CCL range: 1/5 and 1/3.

*Source: _wp04175 - Section IV turns to the European option formula for a simple parameterization of the*

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

### References

### I. Introduction and observed patterns
- Global reserves (in percent of world GDP) have exhibited an upward trend since the 1960s (Flood and Marion (2001); Edison (2003)).
- Over the same period, exchange rate flexibility has increased; no evidence has been put forward that exchange rate flexibility has decreased (Reinhart and Rogoff, 2003).
- Some countries accumulated international reserves of more than 20 percentof their GDPs by late 2003.
- The stock of international reserves held by China alone surpassed US$ 400 billion by end-2003, exceeding the capital of the International Monetary Fund.
- The capital of the IMF, measured by the paid-in quotas of member countries, amounted to SDR 212 billion (somewhat below US $ 300 billion) at end-2002.

### II. Conceptual framing: reserves as self-insurance
- Traditional view: demand for international reserves is a remnant of the fixed exchange rate era.
- Self-insurance view (dating to Heller (1966)): reserves are held as a cushion against undesired shortages of international liquidity that could damage the economy.
- Recent policy discourse (post–capital account-driven currency crises affecting former Asian tigers) gave prominence to the latent insurance view and proposals to maintain reserves large enough to provide against capital market-driven drains on international liquidity (see Mulder (2000) for references).

### III. Equivalence between reserves and financial options
- Holding reserves is interpreted as an act of self-insurance that is functionally equivalent to holding a put option providing identical insurance coverage.
- Merton (1976) first discussed the equivalence between insurance and a put option in the context of deposit insurance, proposing to calculate the cost of deposit insurance using option pricing theory.
- By analogy, the self-insurance value of reserves can be approximated by the market cost of obtaining equivalent insurance, derived from option pricing theory.

### IV. Quantitative framework and its uses
- The option-pricing–based approach provides a quantitative metric of the insurance value of reserves based on observable parameters.
- Advantages:
  - Applicable to a large variety of situations without relying on a single theoretical model.
  - Can calculate the cost of an insurance arrangement that provides insurance coverage of a desired size.
  - For any desired level of overall insurance, can investigate how large a fraction of the total insurance need will be covered by holding reserves (self-insurance ratio).
  - With additional working assumptions, the reserve coverage ratio can be used to assess the amount of excessive reserves in various countries.

### V. Caveats and limitations
- Market value of insurance does not fully coincide with the welfare value of insurance.
- The quantitative metric from option pricing does not form a self-contained basis for determining the optimal level of overall insurance.
- The framework should be viewed as a starting point for quantifying the insurance value of reserve holding, subject to the above caveat.

### VI. Paper organization (overview)
- Section II: reviews existing literature on reserves and discusses its relationship to the insurance interpretation.
- Section III: discusses the equivalence between a financial option and the self-insurance role of reserves.

*Source: _wp04175 - References..........................................................................................................*

### Section IV turns to the European option formula for a simple parameterization of the

### _wp04175 - Section IV turns to the European option formula for a simple parameterization of the quantitative framework.

### II. MANY FACES OF OFFICIAL RESERVES
- Official reserves are held to guard against an undesired shortfall in international liquidity and to mitigate its adverse consequences, while incurring an opportunity (interest rate) cost; this parallels classic insurance: guarding against downside risk at the cost of an insurance premium.
- Traditional reserve-demand motives:
  - Transactions motive: reserves as medium for international transactions (Harrod (1953)); exists in absence of uncertainty.
  - Precautionary motive: reserves held due to uncertainty to limit downward adjustment in external balance (Heller (1966)); optimal reserves affected by cost of adjustment, cost of maintaining reserves, and probability of relying on reserves.
- Literature developments:
  - Clark (1970): general equilibrium tradeoff between domestic adjustment and external financing (reserves).
  - Frenkel and Jovanovic (1981): stochastic model with fixed adjustment cost when reserves depleted.
  - Post-1990s currency-crisis literature emphasized insurance/“war chest” motive—maintaining reserves that exceed short-term external liabilities (Guidotti and Greenspan; Mulder, 2000).
  - Others recognizing insurance role: Feldstein (1999), Kletzer and Mody (2000), Caballero (2003).
  - Aizenman and Marion (2002) explicitly emphasized the insurance value of reserves.
- Distinct contribution of this paper:
  - Brings quantitative implications of the insurance view to the fore while abstracting from specific sources of insurance need.
  - Treats reserve holding as self-insurance and compares it to market-based insurance alternatives; analyzes value and desirability of reserve holding as a quantifiable financial problem.

### III. OPTIONS AND RESERVES
A. Two Methods of Insurance
- Setup:
  - Agent (country) needs guaranteed cash flow D at time T against an underlying asset value Vt (0 ≤ t ≤ T).
  - Net cash flow at maturity: VT − D; insurance with coverage D pays D − VT when VT < D.
- Equivalence:
  - Market-based insurance: put option with exercise price D on underlying asset; payoff max{D − VT, 0}.
  - Self-insurance: hold cash reserves C (C = D for full self-insurance); drawdown equals max{D − VT, 0}.
  - Underlying asset identity varies by context (exports, project value, commodity exports, government bond as pledgeable asset).
- Contractual arrangements can expand underlying asset base, reducing regional insurance costs (to be discussed in Section A).

B. Insurance Value of Reserves
- Notation:
  - G(D, Vt, T, t) = time-t price of put maturing at T with exercise price D.
- Net payoff of purchasing put (approximation as T−² → T): max{VT − D, 0} − G(D, VT, T, T).
- Terminal put price condition: G(D, VT, T, T) = max{D − VT, 0}.
- Net period-T payoff of both put purchase and holding reserves of level C = D simplify to VT − D.
- Time-t insurance value of holding reserves of level C: I(C, Vt) ≡ G(C, Vt, T, t). (Equation (5))
  - This is the market price of equivalent insurance coverage and the implicit cost of setting up market-based insurance if not already available.

C. Optimal Reserve Coverage
- Reserve coverage ratio λ = C/D.
- With reserves C = λD, market-based insurance required has exercise price ˜D ≡ (1 − λ)D.
- Insurance benefit of holding reserves: G(D, Vt, T, t) − G(˜D, Vt, T, t). (Equation (6))
- Opportunity cost of holding reserves:
  - Investment interest rate R (return on less-liquid investment or borrowing rate) vs. liquid asset rate r.
  - Present value of foregone investment income (approximation e^{(R−r)(T−t)} − 1 ≈ (R−r)(T−t)) yields cost = (R−r)(T−t) C. (Equation (7))
- Net benefit of self-insurance (for overall insurance cost) with τ ≡ T − t:
  - B(λ) = G(D, V, τ) − G(˜D, V, τ) − (R−r) τ (D − ˜D). (Equation (9))
- Marginal benefit of increasing λ:
  - MB(λ) ≡ ∂B(λ)/∂λ = D [ ∂G(˜D, V, τ)/∂˜D − (R−r) τ ]. (Equation (10))
- First-order condition for optimal coverage λ*:
  - ∂G(˜D, V, τ)/∂˜D = (R−r) τ. (Equation (11))
- Properties:
  - By monotonicity ∂G/∂˜D ≥ 0, a meaningful solution exists.
  - Full self-insurance (λ = 1, ˜D = 0) is optimal when R − r = 0 because MB(1) ≥ 0.
  - Implicit mapping H(λ*, D, R−r, V, σ) = 0 defines optimal λ* as function of D, spread R−r, asset value V, volatility σ. (Equation (12))
  - For given D, C = λ* D. (Equation (13))
  - λ* = λ*(σ, R−r | D). (Equation (14))

### IV. PARAMETERIZATION BY EUROPEAN OPTION
- Rationale:
  - Use Black-Scholes formula for closed-form solutions to derive indicative ratios.
  - Under Black-Scholes, insurance value and optimal coverage depend on ratio D (or C) to V, independent of the absolute level of V.
  - Acknowledged limitations: American options, credit derivatives, macroeconomic insurance can provide richer alternatives, but European option captures the lion’s share of market insurance value for illustrative analysis.

A. Simple Calculus of Self-Insurance
- Under log-normal asset process dVt / Vt = μ dt + σ dω(t), European put formula:
  - G(D, V, τ) = D e^{−rτ} N(x2) − V N(x1). (Equation (15))
  - x1 = [log(D/V) − (r + σ^2/2) τ] / (σ √τ); x2 = x1 + σ √τ.
- Average insurance value per unit of reserves:
  - I(C, V) / C = e^{−rτ} N(x2) − (V/C) N(x1) τ ≡ IA(C/V). (Equation (16))
- Properties:
  - IA depends on C/V only.
  - IA increases in C/V; ∂IA/∂(C/V) = (V/C)^2 N(x1) > 0.
  - IA increases in volatility σ.
- Marginal benefit using Black-Scholes:
  - ∂G(˜D)/∂˜D = e^{−rτ} N(˜x2).
  - MB(λ) = D [ e^{−rτ} N(˜x2) − (R−r) τ ]. (Equation (17))
- Optimal coverage properties:
  - λ* depends on D (or C) to V ratio only.
  - MB(λ) decreases in λ (convexity of put price); ∂MB/∂λ = D e^{−rτ} N′(˜x2) (−1/(σ √τ (1 − λ))) < 0.
  - Internal solution for 0 < (R−r) τ < 1; MB(0) > 0 and MB(1) < 0 when R > r.

B. Numerical Calculations (summary of findings from calculations described)
- Average insurance value computed for C/V from 0.5 to 3 and σ from 0.05 to 0.5 (5 percent to 50 percent):
  - Volatility has little effect on average insurance value except when C = V; C/V ratio is the primary driver.
- Optimal reserve coverage ratio calculations set D/V = 1:
  - λ* decreases in spread (R−r) and increases in volatility σ.
  - Elasticity of λ* to spread varies with volatility: for low σ (0.05 and 0.1), λ* falls by about one-half as spread rises ten-fold from 100 to 1,000 basis points.
  - For higher σ (e.g., 0.5), decline in λ* with spread is less sharp.
  - Elasticity of λ* to σ varies with spread: for low spreads (1 or 2 percent), λ* rises nearly tenfold as σ increases from 0.05 to 0.5; for higher spreads (9 or 10 percent), λ* rises by more than tenfold over the same σ change.
  - Full coverage (λ = 1) rarely optimal; Table 3 shows λ* < 0.5 for most parameter values except upper-right corner.
  - Table 4 (spreads 1 to 10 basis points) still shows λ* substantially below 1; likely λ* frequently exceeds 0.2 except when σ is low and spread high.
  - Plausible central range for λ* is 20-50 percent (visualized as a diagonal bloc from upper left to lower right across parameter space).

### V. TWO ILLUSTRATIVE APPLICATIONS
A. Cost of an Asian Monetary Fund (AMF)
- Observation: Several emerging-market economies, particularly in Asia, have accumulated large reserves; some economists diagnose overaccumulation (Edison (2003)).
- Chiang Mai Initiative (CMI) mutual swap agreement cumulative total: 35 billion dollars.
- Using Table 1 numerical results (average insurance cost per unit of reserves when C/V = 1 or C/V ≤ 0.5):
  - For institutional arrangement approximated by C/V = 1 (mutual company pooling reserves and purchasing put options), average cost of insurance per unit of reserves ranges between 0.01 and 0.1 when C/V = 1.
  - Applying these costs: insurance cost of reserves amounting to 10 percent of GDP would be between 0.1 and 1 percents of GDP.
  - Countries with excess reserve holdings around 20 percent of GDP (China and Korea in Asia, and Chile in Latin America) carry insurance with market value (cost) ranging between 0.2 and [text truncated at end of provided content].

*Italicized source attribution: _wp04175 - Section IV turns to the European option formula for a simple parameterization of the*

### 2.0 percents of GDP. Even on the low side of the range, this is probably too expensive to be a

### _wp04175 - 2.0 percents of GDP. Even on the low side of the range, this is probably too expensive to be a

### Insurance cost estimates and option framework
- Cost of holding reserves amounting to 20 percent of GDP: opportunity cost about 0.2 percent of GDP for countries with sovereign spreads around 100 basis points.
- Cost-estimate comparison: European option–based estimate will be, if any, lower than the cost estimate based on the American option or other more customized derivative arrangements.
- For C/V = 0.5 (first row of Table 1): the cost of insurance is less than 0.01 for most parameter values; cost of insurance amounting to 10 percent of GDP would thus be less than 0.1 percent of GDP.
- If C/V can be lowered to 0.1, the cost of insurance falls below 10−5 per unit, implying that the cost of insurance amounting to 10 percent of GDP would be less than a one-thousandth percent of GDP.

### Raising the value of the underlying asset (lowering C/V)
- Two mechanisms to raise the effective value of the underlying asset:
  - Pooling: arrange options for use by a subgroup of members rather than simultaneous use by all members; basing a small number of option arrangements on pooled asset that exceeds the combined notional value raises effective underlying asset value (pure benefit of pooling).
  - Mutual commitments with conditionality: mutual commitment to emergency loan accompanied by conditionality expands the scope and value of the underlying asset beyond pledged external assets; cost of insurance can be paid via forgone investment income on the capital (quota) pledged to the insurance pool (e.g., the AMF).
- Conditionality roles:
  - Ensures repayment ex post.
  - Lowers cost of ex-ante insurance arrangements by expanding the effective underlying asset.
  - Scope of conditionality should be adjusted according to the source of risks; if risks are structural in the private sector, structural conditionality beyond traditional macroeconomic policy is coherent and necessary.

### Most cost-effective arrangement and comparison of pooling sizes
- Most cost-effective when all participants subscribe to conditionality and participant number is large enough to approximate perfect pooling of insurance need within the AMF.
- Internal provision of insurance without purchasing market options is possible when pooling is large; economies of scale imply the IMF (larger pooling) would be strictly more efficient than a smaller regional AMF.

### Excess reserves — advanced versus emerging markets
- Working assumption: desired level of overall insurance (D) equals the amount of short-term external debt.
- Observed ratios:
  - Advanced economies: actual reserve coverage ratios most often fall in the 20-40 percent range, close to calculated optimal reserve coverage ratios in the 20-50 percent range (Tables 3 and 4).
  - Emerging markets: observed ratios of reserves to short-term debt are much higher, except in years prior to outbreak of a currency crisis (e.g., Korea and Thailand in 1997). For most emerging markets, ratios used to be close to or higher than 100 percent and rose sharply in several Asian countries after the Asian crisis.
- Constructed measure of excess reserves: assumes about a half of short-term external debt is the optimal reserve coverage (self-insurance). Using this benchmark, Table 8 reports average reserve holding by emerging markets over 2000-2002, optimal reserve holding implied by average short-term external debt, and implied excess reserves (all in percent of GDP).
- Interpretation: excess reserves gap between emerging and advanced economies is a quantitative indicator of financial-market limitations confronting emerging markets; reflects weakness in institutional development and policy credibility and is expected to narrow only slowly.
- Short-term pragmatic solution: maintain reserves at a level no less than short-term external debt; some argue expanding target to potential amount of capital outflow originating in the domestic market.
- Potential aids: Contingent Credit Line (CCL) facilities and Precautionary Arrangements — under these, the optimal ratio of reserves to size of credit line corresponds to λ∗ (Table 3). CCL recommended quota-to-CCL range was between 1/5 and 1/3; these ratios are on the low side compared with numerical calculations where self-insurance ratio often exceeds 1/3. Stringent qualification requirements for CCL likely limited interest and led to abandonment.

### Main conclusions
- The unobservable insurance value of holding external reserves can be quantified by the price of a put option with equivalent insurance coverage using European option parameterization.
- Pure regional swapping arrangements are too costly to be implemented on a large scale (10 or 20 percent of GDP).
- Arrangements similar to the IMF that adopt conditionality to increase the effective value of the underlying asset can greatly reduce insurance arrangement costs at regional and global levels.
- The theoretically derived optimal reserve coverage ratio is close to actual reserve holdings of advanced economies, while emerging markets hold reserves well above optimal ratios—quantifying the access gap.
- The proposed quantitative framework can aid in moving toward narrowing the gap by evaluating tradeoffs between size of access and stringency of requirements.

### Key numeric values and table highlights (preserved exactly as in source)
- Opportunity cost for 20 percent of GDP with sovereign spread ~100 basis points: 0.2 percent of GDP.
- C/V examples:
  - C/V = 0.5 → cost of insurance < 0.01; 10 percent of GDP insurance < 0.1 percent of GDP.
  - C/V = 0.1 → cost of insurance < 10−5 per unit; 10 percent of GDP insurance < one-thousandth percent of GDP.
- Advanced economy reserve coverage ranges: 20-40 percent (observed); calculated optimal: 20-50 percent.
- CCL recommended quota-to-CCL range: 1/5 and 1/3.
- Selected table excerpt figures (preserved exactly):
  - Table 1 first-row values for C/V = 0.5 and volatilities: 0.000 0.000 0.000 0.000 0.001 0.008 0.023
  - Table 2 first-row (C/V = 0.1) across volatilities: 0.000 0.000 0.000 0.000 0.000 0.000 0.00
  - Table 3 sample first-row (spread = 0.01) across volatilities: 0.083 0.187 0.280 0.364 0.508 0.623 0.714
  - Table 4 sample first-row (spread = 0.0001) across volatilities: 0.167 0.328 0.458 0.566 0.728 0.821 0.885
  - Table 8 sample excess reserves (in percent of GDP; average over 2000-2002): Argentina 6 9 1; Brazil 7 6 4; Chile 2 2 1 4 15; China 1 9 2 18; Korea 2 4 8 20; Singapore 8 9 7 0 54

*Source: _wp04175 - 2.0 percents of GDP. Even on the low side of the range, this is probably too expensive to be a (PDF).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2004/_wp04175.pdf_