How are income scenarios ranked in the Test Plan?

When looking at the Example Income Scenarios in the "Test Plan" section, you may notice that the income lines for the example scenarios may cross each other. For example, in the graph below you'll see that the dark blue line showing the 'Median' example is higher than the green 'Best' line early in the plan and the red 'Worst' line is higher than several of the other lines at various points through the plan. What's going on here?

There is no way to guarantee that "better" income scenarios will always have higher income in every month of the simulation than "worse" outcomes. In fact, it would be truly incredible if different simulated income experiences didn't cross over the course of the plan. Some will be higher early in the simulation but then go down. Others the reverse of this. Others will zig and zag a bit. So, the only ranking that is possible for income experiences is based on an overall experience, not a month-by-month experience. Our goal in this ranking is to match what the preferences of a retiree would be if faced with a choice between outcomes. While this ranking may not perfectly match any given retiree's preferences, the following are the core components of the ranking.

1. Prefer income now or soon to income later or never: Scenarios are ranked by mortality-adjusted income over the life of the plan. This means higher income earlier in the plan is more valuable than higher income late in the plan.
2. Prefer to avoid large drops in income: Furthermore, example scenarios are demoted if they experience periods of very low income, even if they otherwise have reasonable or even high income levels.

Technically, in providing these example scenarios across a range of outcomes, we are applying a "constant relative risk aversion utility function" to mortality-adjusted income, but the end goal is to rank income scenarios in a way that reflects the relative ordering that a retiree might make when give the choice between different income experiences.