From One Illustration to 10,000 Futures: Stress-Testing an IUL

By the InsuriShield Analytics Desk

Every life insurance illustration tells one story: a single, smooth line marching up and to the right. It is a useful story and a misleading one, because the future does not arrive in a straight line. Markets cluster — calm years follow calm years, then volatility erupts and persists. Carriers move their index caps. And the order in which good and bad years arrive changes the destination. To see what a policy is really likely to do, you have to replace the one line with thousands.

Why not just draw random returns?

The naive way to simulate an indexed account is to draw each year's return independently from a bell curve. It is easy, and it is wrong, because real equity volatility is not independent from year to year — it clusters and persists. A model that ignores that produces unrealistically tidy sequences and understates the runs of consecutive bad years that actually break policies.

Instead we use a GARCH process — generalized autoregressive conditional heteroskedasticity, a mouthful that essentially means ‘volatility that remembers.’ It lets each year's turbulence depend on the recent past, reproducing the calm-then-storm clustering real markets exhibit. The result is thousands of return paths that look like markets actually behave, not like a textbook bell curve.

The cap is a moving target, too

The other half of an indexed account's fate is the growth cap — the ceiling on how much index gain gets credited. Carriers do not hold caps constant; they adjust them as interest rates and options budgets shift, and across the industry caps have drifted down substantially over the past decade. So we do not hold the cap fixed either. We model it as a mean-reverting process — free to wander year to year but pulled toward a long-run average — so every simulated future carries its own plausible path of cap compression. That single choice captures one of the most important and most overlooked drivers of IUL underperformance.

What the picture looks like

Run all of that — thousands of trials, each a full month-by-month projection of the policy's real mechanics under its own market path — and the single illustration line becomes a fan: a dense band of possible cash-value trajectories that widens with time. Collapse the endpoints into a histogram and you get the distribution of where the policy might actually land. The illustration's smooth line is just one path through that cloud, and usually an optimistic one.

“An illustration answers ‘what if everything goes to plan?’ A simulation answers the only question that matters: ‘how likely is the plan?’”

Turning pictures into probabilities

The payoff is not the visualization; it is the probabilities you can read off it. Because every trial is a complete inforce projection, we can count how often specific, real-world outcomes occur: the policy accumulating enough value to do what it was designed to do, repaying a premium-finance loan, sustaining coverage to maturity without new premiums, or matching the illustrated cash value. Each becomes a percentage instead of a hope.

In the engagements we run this way, the gap between the illustrated outcome and the simulated reality is often stark: the as-sold result sits far out in the optimistic tail, achieved in only a small minority of trials. That is not a market-timing accident — it is what the math implied was likely all along.

This is also why ongoing monitoring matters. A distribution is not a one-time verdict; as markets move and caps reset, the odds shift. Re-running the simulation on a schedule turns a static illustration into a living estimate of whether a policy is still on track.

InsuriShield analyses are mathematically derived and provided for informational purposes — they are not legal, financial, or medical advice.