Introduction: Moving Beyond "Life and Death"

Welcome! So far in CM1, you’ve likely spent a lot of time looking at simple "Alive" or "Dead" models. But life is rarely that simple, right? People get sick and recover, they retire, or they move between different levels of care.

In this chapter, we explore Multiple State Models. Think of these as a "map" of someone's life journey. We will learn how to value cashflows that depend on which "state" a person is in (like being in a hospital) or moving between states (like the moment someone retires). Don't worry if this seems a bit more complex than what you've seen before—we'll break it down step-by-step!

1. The Basics: What is a Multiple State Model?

Imagine a board game where your piece can move between different squares. In actuarial terms, those squares are States, and the rules for moving are Transitions.

A common example is the Health-Sickness Model:

  • State 0: Healthy
  • State 1: Sick
  • State 2: Dead

In this model, a person can move from Healthy to Sick, then back to Healthy again, or move to the Dead state from either.
Key Term: A state from which you cannot leave (like "Dead") is called an Absorbing State. A state you can leave is a Transient State.

The Markov Property

To keep the math manageable, we usually assume the Markov Property. This is a fancy way of saying: "The future depends only on where you are right now, not how you got there or how long you've been there."
Analogy: It's like a video game save point. It doesn't matter if you fought 100 monsters or 0 to get there; once you load the save, your options for the next level are exactly the same.

Quick Review:

  • State: The status of the life (e.g., Active, Retired).
  • Transition: Moving from one state to another.
  • Markov Property: Only the current state matters for future probabilities.

2. Transition Intensities: The "Speed" of Change

To value cashflows, we need to know how likely someone is to move between states. We use Transition Intensities, denoted by \( \mu_{x+t}^{jk} \).

This represents the forces acting on a life aged \(x+t\) to move from State \(j\) to State \(k\). It’s very similar to the "force of mortality" you’ve already studied, just expanded to include other life events.

The Probability of Staying Put
The probability that someone currently in State \(j\) is still in State \(j\) after time \(t\) is denoted as \( {}_t p_x^{\overline{jj}} \).
The formula looks scary, but it's just an extension of the survival probability you know: \( {}_t p_x^{\overline{jj}} = \exp \left( -\int_{0}^{t} \sum_{k \neq j} \mu_{x+s}^{jk} ds \right) \)

Why the sum? Because to stay in State \(j\), you must avoid all possible exits to any other state \(k\).

3. Valuing Benefits (Cash Outflows)

In actuarial work, we need to find the Expected Present Value (EPV) of benefits. In multiple state models, benefits usually fall into two categories:

A. Benefits Paid While in a State (Annuities)

Example: A disability income policy that pays $1,000 per month as long as the policyholder is in the "Sick" state.
To value this, we look at the probability of being in State \(s\) at every moment in time, discount it back to today, and integrate: \( EPV = \int_{0}^{n} e^{-\delta t} \cdot {}_t p_x^{as} \cdot B_t dt \)

Where:

  • \( e^{-\delta t} \) is the discount factor.
  • \( {}_t p_x^{as} \) is the probability that a person currently in State \(a\) will be in State \(s\) at time \(t\).
  • \( B_t \) is the rate of benefit payment.

B. Benefits Paid Immediately on Transition (Lump Sums)

Example: A death benefit paid the moment someone moves from "Healthy" to "Dead".
This requires three things to happen at time \(t\): 1. The person must be in the starting State \(j\). 2. The person must transition to State \(k\) at that exact moment. 3. We discount that payment back to time 0.

\( EPV = \int_{0}^{n} e^{-\delta t} \cdot {}_t p_x^{aj} \cdot \mu_{x+t}^{jk} \cdot S_t dt \)

Where \( S_t \) is the sum assured paid on transition.

Key Takeaway: - Use \( {}_t p_x^{jk} \) for annuity-style payments (being in a state). - Use \( {}_t p_x^{aj} \cdot \mu_{x+t}^{jk} \) for lump sum payments (moving between states).

4. Valuing Premiums (Cash Inflows)

Premiums are usually paid only while the policyholder is in a specific state (e.g., the "Active" or "Healthy" state). They are valued exactly like the annuities mentioned above.

The Equation of Value:
Just like in simpler models: EPV of Premiums = EPV of Benefits + EPV of Expenses

Common Mistake: Students often forget that someone can move out of a state and then move back in. Make sure you check if the transition is "once-and-for-all" or if the state is "recoverable." If you can return to a state, you must use the probability \( {}_t p_x^{jk} \) which accounts for all possible paths, not just staying there continuously.

5. Step-by-Step: How to Solve Valuation Problems

If you face a multiple-state valuation question, follow these steps:

Step 1: Draw the State Space. Sketch the boxes and arrows. This helps you see which transitions are possible.
Step 2: Identify the Cashflows. Is it a payment for being in a state or for moving?
Step 3: Write down the Probabilities. Identify the correct \( {}_t p_x \) or intensity needed for each cashflow.
Step 4: Set up the Integral. Combine the discount factor, the probability, and the cashflow amount.
Step 5: Solve (or Simplify). Often, intensities and interest rates are assumed to be constant to make the integration easier.

6. Summary and Quick Review

Did you know? Multiple state models are the foundation of "Critical Illness" insurance, where payments depend on specific diagnoses!

Key Points to Remember:

  • Markov Property: Past history doesn't matter; only the current state does.
  • Transition Intensity (\(\mu\)): The "instantaneous" risk of moving.
  • Valuing Annuities: Integrate \( (Discount) \times (Prob. of being in state) \).
  • Valuing Lump Sums: Integrate \( (Discount) \times (Prob. of being in state) \times (Intensity of leaving) \).

Don't worry if this seems tricky at first! The notation is the hardest part. Once you realize that these are just more detailed versions of the life tables you've already mastered, the logic starts to click. Practice drawing the diagrams first—it makes the math much clearer!