There are 400 individuals, each of whom has a 0.2 chance of incurring a claim. If the distribution of the individual claim amount, given that a claim has occurred, is uniform on the interval (0, 200), find the mean and variance of the total claim.

Essentially, what I have going here is that if each $X_i$ follows $U(0,200)$ and the random variable $N$ is binomial such that $N$~$B(400,0.2)$, then I want to find the mean and variance the following random variable (since the $X_i$ are independent): $$\sum_{i=1}^{N}X_i$$ I suspect that the mean is simply $E(N)E(X_i)=80*100=8000$, but I have no idea how to calculate the variance. Any help would be appreciated!

  • $\begingroup$ For example, look here: math.unl.edu/~sdunbar1/ProbabilityTheory/Lessons/Conditionals/… $\endgroup$ – NCh Jan 29 '18 at 18:14
  • $\begingroup$ Your suspicion is correct but unjustified. Try a google or peruse through your notes looking for Wald’s formula or law of total expectation for justification. For calculating the variance, try applying the law of total variance. $\endgroup$ – Nap D. Lover Jan 29 '18 at 19:16
  • $\begingroup$ if the solution below answered your question, consider accepting it by clicking on the tick $\endgroup$ – dimebucker Feb 13 '18 at 0:35

$S = \sum_{i=1}X_i$ is a random sum, where the claim size $X_i \sim f_X(x) $ are iid 'portfolio-wide' distributions, and the claim number $N \sim p_N(n)$ is typically poisson, binomial or negative binomial.

To find the mean of a random sum, we need to make use of conditional probabilities. Intuitively, if we knew what the value of $N$ is, say $10$, then it would be quite simple:

$$ E(S|N=10) = \sum_{i=1}^{10} E(X_i) = 10 E(X_1) $$

but since $N$ is itself random (we do not know how many claims we will have in any given month, we must condition on the value of $N$. To do so, we use the law of iterated expectation, which I'll break down very finely:

\begin{align*} E(S) &= E_N[E_{X_i}(S|N)]\\ &= E_N \left ( E_{X_i} \left (\sum_{i=1}^N X_i ~ \bigg |~N \right ) \right)\\ &= E_N \left ( \sum_{i=1}^N E_{X_i}(X_i|N) \right )\\ & = E_N \left ( \sum_{i=1}^N E_{X_i}(X_i) \right )\\ & = E_N \left ( \sum_{i=1}^N \mu_i \right )\\ & = E_N \left ( N\mu_1 \right )\\ & = E_N ( N) \mu_1\\ & = \nu \mu_1 \end{align*}

where $\nu$ in your case is $400 \times 0.2$, and $\mu_1 = \frac{200}{2}=100$

Note that we are able to drop the conditioning on $N$ when calculating the expectation of $X_i$ as we assume the claim amounts and the number of claims are independent. The idea is that we don't know $N$, so we average over all possible values of N in order to solve the expectation of the random sum.

As for the variance, we need the law of the iterated variance:

\begin{align*} \text{Var}(S) &= E_N(\text{Var}_{X_i}(S|N)) + \text{Var}_N(E_{X_i}(S|N)\\ & = \dots \\ & = \nu \mu_2 + \mu_1^2(\tau^2 - \nu) \end{align*}

where $\nu = E(N)$, $\tau^2 = \text{Var}(N)$, $\mu_1 = E(X_i)$, $\mu_2 = E(X_i^2)$.

I'll leave the intermediate steps in the iterated variance to you


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.