Lifetime Allocation

Most glide-path rules (“100 minus your age”, target-date funds) are heuristics with no link to your income target, risk tolerance or guaranteed income. This model derives an allocation from stated preferences and return assumptions instead.

What “optimal” means here

The optimizer picks the equity weight for each 5-year step that maximizes the expected utility of retirement spending, scored with a CRRA (constant relative risk aversion) function: a dollar counts for more when you would otherwise fall short than when you are already comfortable. The result is the best risk-adjusted outcome for the inputs entered, not the highest expected return and not the least volatile.

Two preferences shape it. γ sets how much you dislike uncertain retirement spending: a higher γ gives safer, lower-equity plans once spending is flexible, but barely moves the allocation with constant-dollar spending, because that plan is driven by avoiding shortfall years. β sets how much later retirement years count compared with earlier ones.

It maximizes welfare, not success rate. Where no stock/bond mix can close a funding gap, the fix is in your inputs (retire later, spend less, save more), not the allocation. Money left at the end of the plan has no value in the model.

How it optimizes

Monte Carlo coordinate ascent: hold every step’s weight fixed but one, try every candidate weight for that step on the same simulated market paths (common random numbers), keep the best, and cycle through the steps until nothing improves. No shape (flat, rising, falling) is assumed.

By default each year’s return is an independent draw from our PWL / FP Canada capital-market assumptions, in real (today’s) dollars. Those assumptions are compound (geometric) returns, so the simulation converts each one to the matching average annual return; otherwise volatility drag would make every mix compound below its stated rate. The Historical sequences option instead replays stock and bond sequences from the JST Macrohistory cross-country panel, rescaled to the same assumptions and block-bootstrapped. That keeps history’s multi-year patterns: crashes followed by recoveries, and inflation decades.

Reading your result

CE income (certainty-equivalent income) is the steady yearly income that would feel as good as the uncertain plan. Shortfall is the share of simulated retirements with at least one year the portfolio cannot fund your target. It is reported from the expected retirement balance (drawdown) and from today (full-path), which adds the luck of markets before you retire.

The tool prefers the best single constant equity weight unless the optimized glide path beats it by more than 3% of CE income or 1 point of full-path shortfall. Smaller gaps are inside the model’s own uncertainty, and one fixed allocation is easier to hold through market swings. Guaranteed income starts at the age you set; before then (a “bridge” after an earlier retirement) the portfolio funds your whole target.

What the model finds

The return model sets the equity level. With independent returns, constant-dollar spending derisks into a moderate “bond tent” near retirement. With historical sequences, where equities partly recover while long nominal bonds suffer bad decades, the optimum is about 100% equity. The two can disagree by 30 percentage points, so compare both settings of the return model.

About 100% equity is a scenario, not the default. That result leans on pre-1990 monetary regimes, surviving developed markets, and long nominal bonds being the only alternative. With a real, short-duration holding (such as a short inflation-linked bond fund) the optimum drops to roughly 50–70% equity in our tests.

The shape is worth little; the level matters. Out of sample, the optimized glide path beats the best constant weight by only about 0.5% of CE income with independent returns, and by essentially nothing with historical sequences.

A wrong guess about the regime costs little. Any constant weight from 70% to 100% equity cost at most about 2% of CE income in every historical era we tested.

How well funded you are sets the tent. The more your savings exceed what the spending target needs, the less upside you can use and the more the optimizer protects the floor. If every allocation already delivers your target, the tool says the plan is over-funded: the weight shown is the middle of a tie.

Flexible spending holds about 100% equity. That is true under both return models. Compare spending rules at the same spending level, though: a 4% draw on a large balance can be double a constant-dollar target, so most of a flexible plan’s higher CE income is spending more, not spending flexibly. Check the year-1 portfolio draw shown with your result. At the same starting draw, flexibility is worth a few thousand dollars a year in our default case.

Guaranteed income is your bond floor. A larger guaranteed income (CPP + OAS + DB) lets the portfolio take more equity risk. With none, rare depleted years dominate the score, so the tool requires at least $10,000 a year. If it covers your whole target and spending is fixed, no allocation is better than another and the tool asks you to adjust the inputs. Starting it after you retire lowers income: retiring at 60 with CPP/OAS from 65 cost about 6% in our example.

Leverage earns little. For a young saver, allowing up to 150% equity raised CE income by under 1% under both return models, and only in the first decade of saving, before margin calls, rate changes or nerves. In this model, equity above about 110% is unrewarded.

Bonds here are nominal. The app’s bond asset is an ordinary (nominal) bond. Our research finds that a mostly real-return bond sleeve pairs with a lower equity weight, but that rests on an assumed real-bond series and is not an option in the app.