What rate of return should I assume for my plan?

Nobody can tell you, and nobody publishes one. No statistics office or regulator measures a future long-run return, because it has not happened yet. Every figure you see — in a calculator, a forum, or Nivritee — is an assumption. The useful questions are whether it is modest, whether it matches the mix you actually hold after costs, whether it is consistent about inflation, and whether your plan still works if reality comes in lower.

What forecasters publish, and what it is

Large asset managers publish capital market assumptions — J.P. Morgan’s 2026 edition, its thirtieth, covers a 10- to 15-year horizon. These are reasoned estimates from current yields and valuations. They differ between firms and are revised every year, which is the point: they are opinions about the future, not measurements of it, and none covers the forty or fifty years a plan may run.

What history says, and what it cannot

• In the US from 1928 to 2025, $100 in the S&P 500 grew to about $1.16 million — roughly 10.0% a year nominal. 10-year Treasuries compounded at about 4.5% and Treasury bills about 3.4%.
• The Dimson–Marsh–Staunton database covers 35 countries from 1900. It records that global equity investors earned 3.5% a year after inflation over the past 25 years, and that returns in the first quarter of this century were lower than in the last century.
• Single markets can disappoint for decades: Japan’s Nikkei 225 took until February 2024 to regain its December 1989 close.

History is one path out of many that could have happened, and a single country’s record is one country’s luck. A plan that assumes the best long stretch on record is a plan that fails quietly.

Real or nominal: pick one and stay with it

A nominal return includes inflation; a real return has it taken out. Use a nominal return with costs that rise with inflation, or a real return with costs held in today’s money — never a nominal return with flat costs (which ignores inflation) or a real return with rising costs (which counts it twice). Nivritee works in nominal terms: each cost inflates at its own rate, so the returns you enter are nominal too.

Adjust for what you actually hold

Two more adjustments are yours to make. Nivritee does not model tax, so if tax will take part of your returns, a lower figure allows for it. And returns earned in one currency but spent in another carry an exchange-rate risk on top.

Nivritee’s two assumptions

In You & your assumptions, under Rates we’ll project with, your plan carries Investment return while you are working and Investment return after you stop earning. Both start from your country’s defaults, and the note behind each says what it is: an assumption, not a measurement, anchored to long-run index history and set deliberately below it. The second describes a bond-tilted portfolio and is lower by design. Press Edit these to change either; Reset to country defaults puts them back.

Test it rather than trust it

1. Lower each return by one point, generate the plan, and see whether the verdict changes.
2. Run the stress tests to see bad sequences of years, not just a lower average.
3. Once you list holdings, Return against your plan on Investment Performance compares what you have earned with the rate your plan assumes, and once three years are behind it and at least half of what you hold is priced, Your plan at your actual return reruns the projection at that rate.

Nivritee’s levers never include “earn a better return”: it is not a lever anybody can pull, and pricing it would be investment advice.

Common mistakes

• Using the last ten years’ return as the next forty years’.
• Mixing real returns with costs that rise with inflation, or the reverse.
• Forgetting fees, tax and cash drag.
• Using one figure for an all-equity portfolio when half your money sits in deposits.