Weather Derivatives and Other Insurance Derivatives

Why Index-Based Derivatives Need Their Own Settlement Model

Weather can affect an organization’s costs or revenue without causing a conventional insured loss. A cold winter may raise a utility’s heating demand; a dry growing season may reduce a farm’s revenue. A weather derivative can transfer some of that exposure by making its payment depend on a measured index, such as accumulated heating degree days or seasonal rainfall. Other insurance-linked derivatives can use a catastrophe loss index rather than the buyer’s own claim amount.

The key modeling task is to translate the contract’s index, trigger, payout rate, and limit into a settlement amount. The contract wording matters: a threshold may mark the point where payment begins, while a separate cap limits the maximum payment. Neither should be confused with an insurance deductible or a limit applied to an individual claim.

Definition: An index-based derivative is a contract whose settlement is determined by a specified index and contractual settlement function. The index may represent a measured weather quantity or an insurance-related quantity. Unless the contract says otherwise, settlement depends on that index—not on proof of the buyer’s actual loss.

Temperature and Precipitation Indices

A common temperature index is the heating degree-day total. For day $t$, let $T_t$ be the mean temperature, measured in the contract’s specified location and temperature scale. For a chosen base temperature $T_{\mathrm{base}}$, the daily heating degree days are $\max(T_{\mathrm{base}}-T_t,0)$. The period index adds the daily values. A cooling degree-day index reverses the difference: it adds $\max(T_t-T_{\mathrm{base}},0)$ over the specified period.

H=\sum_{t=1}^{n}\max(T_{\mathrm{base}}-T_t,0), \qquad C=\sum_{t=1}^{n}\max(T_t-T_{\mathrm{base}},0).

The base temperature, observation station, dates, treatment of missing observations, and units must all be specified by the contract. A cumulative rainfall index is often simply the total measured rainfall over a stated period, but the location, gauge, and units still need definition. Two contracts that both refer to “winter temperature” can produce different results if their observation periods, bases, or data sources differ.

A common capped call-style settlement pays for an index above a threshold $H_0$. If $K$ is the payment per index unit and $M$ is the maximum payment, the settlement is:

P(H)=\min\!\left(K\max(H-H_0,0),M\right).

A low-rainfall contract may instead pay when rainfall $R$ falls below a threshold $R_0$:

P(R)=\min\!\left(K\max(R_0-R,0),M\right).

These formulas specify settlement, not the probability distribution of the index. A distribution is needed to estimate probabilities or expected payments, but not to calculate a payment once the measured index is known.

Worked Example: A Capped Heating Degree-Day Contract

A hypothetical energy supplier buys a contract based on seven daily mean temperatures, all measured in degrees Fahrenheit at the location named in the contract. The base is $65^\circ$F. The recorded daily means are $60, 62, 65, 68, 63, 58,$ and $70^\circ$F. The contract pays $4,000 for each heating degree day above 12, subject to a maximum payment of $30,000. Find the settlement.

Assumptions and parameters. This is a deterministic settlement calculation using the stated temperatures and contract terms; no probability distribution is imposed. There is no mortality assumption. Interest is not applied because the requested amount is the undiscounted settlement at the contract’s settlement date.

Method. Calculate each daily heating degree-day contribution using $\max(65-T_t,0)$, then apply the threshold and cap.

The seven daily contributions are $5, 3, 0, 0, 2, 7,$ and $0$. Therefore:

$$ H=5+3+0+0+2+7+0=17. $$

The index exceeds the threshold by $17-12=5$ degree days. Before applying the cap, the payment is $5\times \$4{,}000=\$20{,}000$. The capped payment is $\min(\$20{,}000,\$30{,}000)=\mathbf{\$20{,}000}$.

Check. The uncapped amount is below the maximum, so the cap does not change the settlement. A common mistake is to apply the $4,000 rate to all 17 degree days, producing $68,000 before the cap. The threshold means only the five degree days above 12 earn a payment.

Thresholds, Caps, and Index-Based Settlement

A useful way to read a settlement formula is to separate it into three operations: determine the index shortfall or excess relative to the trigger, convert that amount into a payment using the rate, and apply any maximum. The order matters. The positive-part expression $\max(H-H_0,0)$ ensures that an index below the trigger produces no payment under a high-index contract. The cap then prevents the calculated payment from exceeding the contractual maximum.

For a low-index contract, such as a rainfall shortfall contract, the difference is $R_0-R$, not $R-R_0$. This sign is an important source of errors: rainfall above the threshold should not generate a positive payment under a contract designed to protect against insufficient rainfall.

Worked Example: A Rainfall Shortfall Derivative

A hypothetical agricultural cooperative enters a 30-day rainfall derivative. The specified gauge records 52 millimeters of rainfall. The contract pays $1,500 for each millimeter below 80 millimeters, with a maximum payment of $45,000. Calculate the settlement and compare it with a hypothetical cooperative loss of $70,000.

Assumptions and parameters. The rainfall total is the contract index; the payout rate is $1,500 per millimeter; the threshold is 80 millimeters; and the maximum payment is $45,000. This is a deterministic settlement calculation, so no probability distribution is imposed. There is no mortality assumption, and no interest adjustment is made to the settlement amount.

Method. Calculate the rainfall shortfall, multiply by the rate, and then check the cap.

The shortfall is $80-52=28$ millimeters. The uncapped payment is $28\times \$1{,}500=\$42{,}000$. Applying the maximum gives $\min(\$42{,}000,\$45{,}000)=\mathbf{\$42{,}000}$.

The cooperative’s hypothetical $70,000 loss does not change this result. The contract pays $42,000 because of the measured rainfall, not because the cooperative substantiated a $70,000 loss. It may therefore undercompensate this loss. Under different circumstances, an index contract could pay when the buyer has little or no corresponding loss.

Check. The shortfall is positive and the uncapped payment is below the cap. A common mistake is to treat the derivative as indemnity insurance and substitute the cooperative’s $70,000 loss for the formula’s calculated amount.

Other Insurance-Linked Derivatives

An insurance-linked derivative can also be based on a regional or industry catastrophe index. One design pays in proportion to the part of the index that falls within a specified layer. Let $X$ be the index in the contract’s stated units, $A$ the attachment point, $B$ the exhaustion point, and $N$ the maximum payment. For $B>A$, a linear layer settlement is:

P(X)=N\min\!\left(\max\!\left(\frac{X-A}{B-A},0\right),1\right).

Below $A$, payment is zero. Between $A$ and $B$, the payment increases linearly from zero to $N$. At or above $B$, it remains at $N$. The index must be defined precisely; the buyer’s own loss is not automatically the settlement measure.

The layer formula follows by measuring how much of the interval from $A$ to $B$ has been traversed. For $A<X<B$, that fraction is $(X-A)/(B-A)$: it is zero at $X=A$ and one at $X=B$. Multiplying the fraction by $N$ gives a payment that increases linearly across the layer. Clamping the fraction below zero and above one gives the formula for all index values.

Worked Example: A Catastrophe-Index Layer

A hypothetical insurer enters a derivative with a regional catastrophe loss index $X$, measured in millions of dollars. The contract has attachment point $A=60$, exhaustion point $B=80$, and maximum payment $N=\$2{,}000{,}000$. Suppose the observed index is $X=74$. Also suppose, solely to calculate an expected payment, that the index at settlement is modeled as continuous uniform on $[0,100]$ million dollars. Find the payment at $X=74$ and the expected payment under that distribution.

Assumptions and parameters. The observed settlement calculation uses the stated index and layer terms. For the expectation, $X\sim\operatorname{Uniform}(0,100)$, with all index amounts in millions of dollars. No mortality assumption applies. Interest is not applied; the expected amount is the undiscounted payment at settlement, not a market price.

Observed settlement. Since $60<74<80$, the index is partway through the layer. Its fraction is $(74-60)/(80-60)=14/20=0.7$. The payment is $0.7\times \$2{,}000{,}000=\mathbf{\$1{,}400{,}000}$.

Expected payment. For indices from 0 to 60, the payment is zero. From 60 to 80, it increases linearly from zero to $2,000,000; above 80, it remains $2,000,000. Under the uniform distribution, the expected fraction of the maximum payment is the area under the payment fraction divided by the total interval length:

\frac{\tfrac12(20)(1)+(20)(1)}{100} =\frac{10+20}{100} =0.3.

The first area is the triangle across the layer, and the second is the full-payment interval from 80 to 100. Thus the expected payment is $0.3\times \$2{,}000{,}000=\mathbf{\$600{,}000}$.

Check. The result is between zero and the $2,000,000 maximum. A common mistake is to multiply the maximum by $P(X>60)=0.4$, obtaining $800,000. That treats every index above attachment as a full-limit payment and ignores the partial payment within the layer.

What the Settlement Does—and Does Not—Measure

Index derivatives and indemnity insurance respond to different measures. An indemnity contract generally bases payment on covered loss, subject to its terms, conditions, and limits. An index derivative bases payment on the contractual index and settlement function. An index can be objective and relatively quick to verify, but it may not move exactly with the buyer’s economic loss. That mismatch is basis risk.

For a sound calculation, identify the following before evaluating the formula:

Index definition: the measured quantity, units, source, location, and observation period.
Trigger direction: whether payment begins above or below a threshold.
Payout rate and shape: fixed per-unit payment, linear layer, or another stated function.
Limits: the maximum payment and any other boundary in the settlement terms.
Settlement basis: whether payment depends only on the index or on substantiated actual loss.

To estimate an expected payment, the model must additionally specify an index distribution and assumptions consistent with the contract period and data definition. The expected payout is not automatically the derivative’s fair price: valuation may also depend on discounting, risk compensation, and market terms. The payout calculations here establish the contractual amount or a model-based expectation, not a complete pricing framework.

Key takeaway: First map the measured index into the contract’s settlement function. Apply the threshold in the correct direction, calculate the payment rate only on the eligible portion, and enforce the cap. Do not replace the index-based amount with the buyer’s actual loss unless the contract explicitly requires indemnification.

Check Your Understanding

For each problem, use the stated contract terms and distinguish a settlement calculation from any loss amount that is not part of the formula.

A hypothetical heating contract uses a base temperature of $60^\circ$F. The period’s total HDD index is 34. It pays $2,500 per HDD above 20, capped at $30,000. What is the payment? (Answer: $30,000)
A cooling contract has a CDD index of 27, a threshold of 18, a rate of $1,200 per CDD, and a $15,000 cap. What is the settlement? (Answer: $10,800)
A rainfall derivative pays $900 per millimeter below 65 millimeters, capped at $18,000. The measured rainfall is 44 millimeters. Find the payment. (Answer: $18,000)
A catastrophe-index layer has attachment 45, exhaustion 75, and maximum payment $3,000,000. The index is 60, with all three index amounts measured in millions of dollars. Find the payment. (Answer: $1,500,000)
An index $X$, in millions of dollars, is uniformly distributed on $[0,120]$. A derivative pays a maximum of $2,000,000, increasing linearly from zero at 70 to the maximum at 90. Calculate the expected payment. (Answer: $666,666.67)

What You'll Learn