Welcome to Premium Calculation!

In this chapter, we transition from calculating the value of future benefits to figuring out how much a policyholder needs to pay for them. Think of this as the "Pricing Department" of the actuarial world. We will explore three different ways to set a price: making it perfectly fair (Equivalence Principle), making it safe for the whole company (Portfolio Percentile Principle), and ensuring the company makes a target profit (EPV of Profit).

Don't worry if this seems tricky at first! While the notation can look like a bowl of alphabet soup, the underlying logic is just about balancing a scale. Let's dive in!

1. The Equivalence Principle: The "Fair Trade"

The Equivalence Principle is the most fundamental concept in actuarial pricing. It states that at the moment a policy is issued, the expected value of what the customer pays must equal the expected value of what the customer receives.

We define a random variable called the Loss-at-Issue (L). It is calculated as:
\( L = PV(\text{Benefits}) + PV(\text{Expenses}) - PV(\text{Premiums}) \)

Under the Equivalence Principle, we set the premium such that the Expected Value of this loss is zero:
\( E[L] = 0 \)
This is equivalent to saying:
\( E[PV(\text{Premiums})] = E[PV(\text{Benefits})] + E[PV(\text{Expenses})] \)

Net Premium vs. Gross Premium

There are two "flavors" of premiums you need to know:

  1. Net Premium: This ignores expenses. It only covers the benefits. (e.g., \( P \cdot \ddot{a}_x = A_x \))
  2. Gross Premium (G): This includes everything—benefits plus the costs of running the business (commissions, taxes, and administration).

Step-by-Step: Solving for a Premium

To find a premium using the Equivalence Principle, follow these steps:

  1. Write out the EPV of Benefits (e.g., \( A_x \) for a whole life policy).
  2. Write out the EPV of Premiums in terms of the unknown (e.g., \( P \cdot \ddot{a}_x \)).
  3. Write out the EPV of Expenses (watch out for "first-year" vs. "renewal" expenses!).
  4. Set them equal: \( \text{Premiums} = \text{Benefits} + \text{Expenses} \).
  5. Solve for the premium.

Quick Tip: Always check if premiums are paid at the beginning of the year (annuity-due \( \ddot{a} \)) or end of the year (annuity-immediate \( a \)). In FAM, premiums are almost always at the beginning of the year!

Key Takeaway: The Equivalence Principle means "On average, the company breaks even at the start of the policy."

2. The Portfolio Percentile Principle: The "Safety Buffer"

If a company only charges the Equivalence Premium, they have a roughly 50/50 chance of losing money on the portfolio due to random mortality fluctuations. The Portfolio Percentile Principle sets a premium so that the company has a high probability (such as 95% or 99%) of avoiding an aggregate loss.

The Math of many policies

Instead of looking at one person in isolation, we look at a portfolio of \( n \) independent policies. Let \( S \) be the aggregate future loss for the portfolio:
\( S = L_1 + L_2 + \dots + L_n \)

We want to determine the premium such that the probability of the portfolio suffering a loss is at most a specified small probability \( \alpha \) (e.g., \( \alpha = 0.05 \)):
\( \Pr(S > 0) \le \alpha \)

Using the Normal Approximation

Since \( n \) is usually large, we apply the Central Limit Theorem and approximate \( S \) by a Normal Distribution with:
Mean: \( E[S] = n \cdot E[L] \)
Variance: \( Var(S) = n \cdot Var(L) \)

Standardizing the probability condition gives:
\( \Pr(S > 0) = \Pr\left(\frac{S - E[S]}{\sqrt{Var(S)}} > \frac{0 - E[S]}{\sqrt{Var(S)}}\right) \le \alpha \implies \frac{-E[S]}{\sqrt{Var(S)}} \ge z_{1-\alpha} \)
Equivalently:
\( -E[S] \ge z_{1-\alpha} \cdot \sqrt{Var(S)} \)

Depending on the premium structure, this is solved in two common ways:

  1. Single Premium Case: If a single premium \( U \) per policy (or aggregate premium \( nU \)) is paid upfront, let \( Y = \sum_{i=1}^n PV(\text{Benefits}_i) \) be the total present value of benefits. The aggregate loss is \( S = Y - nU \). Then \( \Pr(Y > nU) \le \alpha \), leading directly to:
    \( \text{Total Single Premium} = nU = E[Y] + z_{1-\alpha} \cdot \sqrt{Var(Y)} \)
  2. Annual Premium Case: If annual premiums \( P \) are paid throughout life, \( L_i = PV(\text{Benefits}_i) - P \cdot \ddot{a} \). Substitute \( E[S] = n(E[PV(\text{Benefits})] - P \cdot \ddot{a}) \) and \( \sqrt{Var(S)} = \sqrt{n \cdot Var(L)} \) into \( -E[S] = z_{1-\alpha} \sqrt{Var(S)} \), and solve for \( P \).

Analogy: Imagine you are hosting a party. The Equivalence Principle says you buy exactly enough pizza for the average appetite. The Percentile Principle says you buy extra pizzas just in case your guests are hungrier than usual so you don't run out!

Did you know? Because \( \sqrt{Var(S)} = \sqrt{n} \cdot \sigma_L \), the safety margin per policy is proportional to \( \frac{1}{\sqrt{n}} \). As the portfolio size \( n \) grows, the required risk load per policy decreases! This is the Law of Large Numbers in action.

Key Takeaway: This principle accounts for the risk of random fluctuation, not just the average cost.

3. Expected Present Value (EPV) of Profit

Insurance companies need to make a profit to support capital and stay solvent. Profit is simply the negative of Loss.

If we define the Profit Random Variable (\( Pr \)) as:
\( Pr = -L = PV(\text{Premiums}) - PV(\text{Benefits}) - PV(\text{Expenses}) \)

Then the EPV of Profit is just the expected value of that variable: \( E[Pr] = -E[L] \).

Targeting a Profit

In many exam problems, you will be asked to find a premium that results in a specific Profit Margin or a specific Expected Profit.
Example: "Find the gross premium \( G \) such that the EPV of profit is 10% of the EPV of gross premiums."
Equation: \( E[Pr] = 0.10 \cdot E[PV(\text{Gross Premiums})] \)
Which translates to: \( E[PV(\text{Premiums})] - E[PV(\text{Benefits})] - E[PV(\text{Expenses})] = 0.10 \cdot E[PV(\text{Premiums})] \).

Common Pitfalls to Avoid:

  1. Variance of L: Remember that \( Var(L) = Var(PV(\text{Benefits}) - PV(\text{Premiums})) \). For level benefits \( b \) and level net premium \( P \) on a whole life policy, this simplifies to:
    \( Var(L) = \left(b + \frac{P}{d}\right)^2 \left( ^2A_x - (A_x)^2 \right) \). Don't forget to square the \( (b + P/d) \) factor!
  2. Timing: Be careful with expenses that happen "at issue" (Time 0) versus expenses that happen "at the beginning of each year" or "at the end of each year."
  3. Interest Rates: Note whether the question gives you \( d \) (discount rate), \( v \) (discount factor), or \( i \) (effective annual interest rate).

Key Takeaway: Profit is "negative loss." Setting a target profit is simply formulating an equation with \( E[L] = -\text{Target Profit} \).

Quick Review Box

  • Equivalence Principle: Set \( E[L] = 0 \). (EPV of Premiums = EPV of Outgo)
  • Loss-at-Issue \( L \): \( PV(\text{Benefits}) + PV(\text{Expenses}) - PV(\text{Premiums}) \).
  • Portfolio Percentile Principle: Set \( \Pr(S > 0) \le \alpha \), leading to \( -E[S] = z_{1-\alpha} \cdot \sqrt{Var(S)} \). For single premiums, \( \text{Total Premium} = E[Y] + z_{1-\alpha} \cdot \sigma_Y \).
  • Normal Approximation: \( E[S] = n E[L] \) and \( Var(S) = n Var(L) \).
  • Target EPV of Profit: Set \( E[Pr] = -E[L] = \text{Target} \).

Final Encouragement: You are building the foundation for long-term actuarial modeling! Practice moving between the loss formulas and the annuity symbols, and you will master this chapter with confidence.