Introduction: When Two Lives Are Linked
Welcome to one of the most practical chapters in the ALTAM syllabus! Up until now, you might have been assuming that the death of one person has nothing to do with the death of another. In the actuarial world, we call this independence. But let’s be real: if a couple lives together, eats the same food, and travels in the same car, their "future lifetimes" are likely connected.
In this chapter, we are going to learn how to move beyond the simple assumption of independence. We will explore how to model the dependence between two lives (usually denoted as \(x\) and \(y\)). Understanding this is crucial because if we ignore these links, we might seriously misprice insurance products or underestimate the risk for a pension fund. Don't worry if this seems a bit "math-heavy" at first—we'll break it down piece by piece!
Quick Review: Remember that \(T_x\) and \(T_y\) are the random variables representing the remaining lifetimes of \(x\) and \(y\). When they are independent, the probability of both surviving is simply \(P(T_x > t) \times P(T_y > t)\). In this chapter, that simple multiplication rule gets an upgrade!
The Reality of Dependence: Why Does it Happen?
Before we dive into the formulas, let’s think about why these lifetimes are dependent. In the exam, you might see these concepts referenced in qualitative questions:
- Common Environment: A couple living in the same house is exposed to the same risks, like radon gas, poor air quality, or even just the same diet.
- Common Shocks: An event that happens to both people at once, such as a car accident or a natural disaster.
- The "Broken Heart" Syndrome: Statistically, the death of a spouse often increases the mortality risk for the surviving partner due to grief and lifestyle changes.
Did you know? Actuaries have found that for many joint-life annuities, the "broken heart" effect is a very real statistical factor that must be accounted for to keep the insurance company solvent!
Model 1: The Common Shock Model
The Common Shock Model is the most intuitive way to handle dependence. Imagine three "clocks" ticking. Clock 1 only kills person \(x\). Clock 2 only kills person \(y\). Clock 3 kills both \(x\) and \(y\) at the same time.
The Math of Common Shock
We assume there are three independent sources of mortality with constant forces (hazard rates):
- \(\lambda_1\): Force of mortality affecting only \(x\).
- \(\lambda_2\): Force of mortality affecting only \(y\).
- \(\lambda_{12}\): Force of the "common shock" affecting both simultaneously.
The total force of mortality for \(x\) is \(\mu_x = \lambda_1 + \lambda_{12}\).
The total force of mortality for \(y\) is \(\mu_y = \lambda_2 + \lambda_{12}\).
The survival function for the joint life status (which ends when the first person dies) is:
\(_{t}p_{xy} = P(T_x > t, T_y > t) = e^{-(\lambda_1 + \lambda_2 + \lambda_{12})t}\)
Common Mistake to Avoid: When calculating the joint life force \(\mu_{xy}\), don't just add \(\mu_x + \mu_y\). If you do that, you are counting the common shock (\(\lambda_{12}\)) twice! The correct joint force is:
\(\mu_{xy} = \lambda_1 + \lambda_2 + \lambda_{12} = \mu_x + \mu_y - \lambda_{12}\).
Key Takeaway: Common Shock
The common shock model increases the chance of both lives ending at the exact same time. It reduces the probability that the joint status survives compared to the independent model.
Model 2: Copulas – The Mathematical "Bridge"
If the Common Shock model is a hammer, Copulas are a Swiss Army knife. They are a more flexible way to link two individual (marginal) distributions into one joint distribution.
What is a Copula?
Think of a Copula as a recipe. You have two separate ingredients (the survival probabilities of \(x\) and \(y\)), and the Copula tells you how to mix them to get the joint survival probability.
Mathematically, for survival functions \(S_x(t) = P(T_x > t)\) and \(S_y(t) = P(T_y > t)\), the joint survival function is expressed as:
\(S(t, u) = C(S_x(t), S_y(u))\)
Where \(C\) is the Copula function. In ALTAM, you generally need to know how to plug values into specific Copula formulas provided in the question.
Common Copulas You Might Meet
While there are many, the syllabus focuses on these types of structures:
- Independence Copula: \(C(u, v) = uv\). (This is just our old friend where we multiply the probabilities).
- Gumbel Copula: Often used to model "upper tail dependence" (situations where if one person lives a very long time, the other is likely to as well).
- Frank Copula: Good for modeling a wide range of dependence, from negative to positive.
- Clayton Copula: Focuses on "lower tail dependence" (links between the lives when they are both expected to end early).
Step-by-Step: How to Solve a Copula Problem
1. Find the individual survival probabilities \(_{t}p_x\) and \(_{t}p_y\) using the provided mortality table or formula.
2. Identify the Copula formula given in the exam (e.g., \(C(u, v) = \text{some complex function}\)).
3. Set \(u = _{t}p_x\) and \(v = _{t}p_y\).
4. Plug \(u\) and \(v\) into the Copula formula to find \(_{t}p_{xy}\).
5. If the question asks for the last survivor probability (\(_{t}p_{\overline{xy}}\)), remember the fundamental identity: \(_{t}p_{\overline{xy}} = _{t}p_x + _{t}p_y - _{t}p_{xy}\).
Key Takeaway: Copulas
Copulas allow us to keep the individual mortality models we already like, but join them together in a way that reflects real-world correlation.
Common Pitfalls and Pro-Tips
1. Confusion between \(T_{xy}\) and \(T_{\overline{xy}}\)
Remember: \(T_{xy} = \min(T_x, T_y)\) (Joint life ends when someone dies).
\(T_{\overline{xy}} = \max(T_x, T_y)\) (Last survivor ends when everyone is dead).
Dependence usually affects \(T_{xy}\) more noticeably than individual survival.
2. The "Independence Assumption" Trap
Always read the question carefully. If the question says "the lives are independent," use the multiplication rule. If it mentions a "Common Shock" or a "Copula," you must use the dependence formulas.
3. Logic Check
If two lives are positively dependent (they tend to die together), then the probability that they both survive (\(_{t}p_{xy}\)) will be higher than it would be if they were independent. If you calculate a value lower than the independent result for a positive dependence model, double-check your math!
Quick Review Box
The Essentials:
- Independence: \(_{t}p_{xy} = _{t}p_x \times _{t}p_y\)
- Common Shock: \(\mu_{xy} = \mu_x + \mu_y - \lambda_{12}\)
- Copula: \(_{t}p_{xy} = C(_{t}p_x, _{t}p_y)\)
- Last Survivor: \(_{t}p_{\overline{xy}} = _{t}p_x + _{t}p_y - _{t}p_{xy}\) (This formula works regardless of whether the lives are dependent or independent!)
Summary
In this chapter, we acknowledged that the "real world" is messy and lives are often linked. We learned that the Common Shock Model uses a shared hazard rate (\(\lambda_{12}\)) to represent simultaneous deaths. We also learned that Copulas provide a flexible mathematical framework to link marginal survival functions into a joint survival function. Master the formula for \(_{t}p_{\overline{xy}}\) and the algebra of the specific copula provided to you, and you'll be well on your way to success on this part of the ALTAM exam!