Your financial planner discounts your retirement spending at 7% — the "expected return" on a balanced portfolio. Every pension fund in the world discounts the same obligation at the risk-free rate. One of them is wrong. The difference is 20% of the capital you think you need. This post steelmans the 7% argument, shows why it fails, and tells you what the gap costs in years of your life.
The Best Argument for 7%
Before I demolish it, let me give the other side its strongest case. The argument for discounting at the expected portfolio return goes like this:
-
Over long horizons, equities always recover. The S&P 500 has never delivered a negative real return over any rolling 30-year period in US history. If your retirement is 30+ years, you are virtually guaranteed a positive outcome. Why price the liability as if the money will earn only 2% when history says it will earn 7%?
-
Discounting at 2% overstates your liability by 20%. At a 7% real return, $50,000/yr for 30 years costs $620,000. At 2.3% real, the same spending costs $1,075,000. That is $455,000 more capital — perhaps 5–8 additional years of working and saving. If the higher rate is the true cost of funding the obligation, the lower rate condemns people to work years longer than necessary.
-
US public pension funds do this every day. Under GASB (the Government Accounting Standards Board), US state and local pension funds discount liabilities at the expected return on plan assets — typically 7–7.5%. Hundreds of trillions of dollars of obligations are priced this way. CalPERS uses 6.8%. The New York Common Retirement Fund uses 5.9%. These are not amateurs.
-
Even sophisticated thinkers defend it. Cliff Asness — founder of AQR Capital Management, one of the most rigorous quantitative investors alive — has argued that discounting pension liabilities at the risk-free rate creates perverse incentives: it makes liabilities look unaffordable, pressures sponsors to close defined-benefit plans, and drives participants into inferior defined-contribution plans. The argument has an equity dimension: if risk-free discounting causes pension funds to stop offering guaranteed retirement income, the cure is worse than the disease.
This is not a straw man. It is a serious position held by serious people. And it is wrong.
Why It Fails
The error is a category mistake. The question is not: "What will my portfolio probably earn?" The question is: "At what rate can I guarantee delivery of a known real cash flow?"
These are different questions with different answers.
The Replication Argument
Suppose you owe someone $50,000 in real purchasing power, ten years from now. What does it cost you today to guarantee that payment?
You have two options:
Option A: Buy a 10-year TIPS bond with a face value of $50,000. It contractually delivers $50,000 in real terms on the maturity date. The cost today: $50,000 / (1.023)^10 = $39,800. Done. Risk eliminated. Obligation settled.
Option B: Invest $31,000 in equities, which you expect to grow at 7% real to ~$61,000 in ten years. If it works, you have more than enough. If it doesn't — if you hit a lost decade, a secular bear market, or a bad-beta event — you have $20,000 and an unpayable obligation.
Option A is a hedge. Option B is a bet. They are not the same thing, and they do not have the same price. The price of extinguishing the obligation with certainty is the TIPS price — $39,800, discounted at 2.3%. The $31,000 figure is not a price — it is a hope.
This is the Merton replication principle. The present value of a known obligation is determined by the cost of the portfolio that replicates it — not by the expected return on some other portfolio you happen to hold. You cannot extinguish a certain liability with an uncertain asset at a discount. The market does not allow it.
Or, more simply: you cannot eat expected returns. You can only eat realised cash flows.
The Pension Fund Evidence
The argument that "pension funds discount at 7%" is true — for US public pensions under GASB. It is false for virtually every other institutional standard:
| Standard | Discount Rate | Who Uses It |
|---|---|---|
| GASB (US public pensions) | Expected return on assets (~6–7%) | CalPERS, state/local plans |
| IFRS / IAS 19 | High-quality corporate bond yield (~4–5%) | Every listed company globally |
| UK Pensions Act | Gilt yield + prudence margin (~3–4%) | UK private sector DB plans |
| ERISA / PBGC (US private) | Corporate bond yield curve (~4–5%) | US corporate pension plans |
| Financial economics (Merton, Gold, Bader) | Risk-free real rate (~2%) | Academic consensus |
US public pensions are the outlier, not the norm. And the consequences of discounting at 7% are visible: as of 2025, US state and local pension plans report aggregate funded ratios of roughly 75–80% using their own discount rates. Re-priced at risk-free rates, many are below 50%. The gap is not a theoretical curiosity — it is trillions of dollars of unfunded obligations that will eventually be paid by taxpayers, benefit cuts, or both.
Jeremy Gold and Lawrence Bader — two actuaries who spent their careers fighting this battle — summarised it in 2007: discounting at expected returns "conflates the cost of a promise with the expected return on assets set aside to meet that promise." The cost of the promise is determined by the promise itself — its timing, its inflation character, its certainty — not by whatever speculative portfolio happens to sit next to it.
When CalPERS switches from a 7% discount rate to 6.8%, their unfunded liability increases by roughly $8 billion on a $500 billion obligation. When they switch from 6.8% to 2% — the real risk-free rate — the liability approximately doubles. The money was always owed. The accounting was hiding it.
Time Diversification Does Not Save You
The strongest version of the 7% argument rests on time diversification: the idea that over long horizons, the probability of equity underperformance shrinks toward zero, so it is "safe" to use the expected return.
There are three problems with this.
First, the probability shrinks but the magnitude grows. Over 30 years, equities have never lost money in US data. But over 30 years, the range of outcomes is enormous. The difference between the 10th percentile and the 90th percentile real wealth outcome after 30 years of equity investing is roughly 4:1. "You'll probably be fine" is not the same as "the obligation is funded."
Second, you only get one draw. A pension fund with millions of participants can rely on the law of large numbers — some cohorts will experience bad markets, others good, and the pool averages out. You are a sample of one. You retire once, into one sequence of returns, in one macroeconomic regime. The expected value is not your outcome. Your outcome is one realisation from a wide distribution.
Third, the US historical record is a single sample from the winner. As Asness himself has argued, in "The Long Run Is Lying to You" (2021), simple past returns overstate expected future returns because the US market got progressively more expensive over the 20th century. Valuation expansion inflated realised returns above the true expected return. Adjusting for the fact that starting valuations are higher today, the forward-looking expected real equity return is lower than 7% — perhaps 4–5%. The gap between 5% and 2% is smaller than the gap between 7% and 2%, but the gap still exists — and it is still the difference between a bet and a hedge.
What the Gap Costs You
The practical cost of discounting at the wrong rate is not abstract. It is measured in years of your life.
| Spending | Real Rate | Liability PV (30yr) | Capital Gap | Extra Years Saving |
|---|---|---|---|---|
| $50,000/yr | 7.0% | $620,000 | — | — |
| $50,000/yr | 5.0% | $770,000 | $150,000 | ~2–3 years |
| $50,000/yr | 2.3% | $1,075,000 | $455,000 | ~5–8 years |
Read that table carefully. If you discount at 7% and target $620,000, you believe you are funded. If reality is closer to 2.3%, you are 58% funded — and you do not know it until it is too late.
The 7% assumption does not save you time. It hides a deficit. You feel funded. You retire. And then, at 72, when equities have not delivered their historical average, when real rates have moved, when the risk premium you were counting on failed to materialise, you discover the gap. At that point, the options are: cut spending, return to work, or depend on others. None of these are the retirement you planned.
The 2.3% assumption costs you 5–8 more years of saving. But those years buy certainty. When you reach 100% funded at the TIPS rate, you are funded in fact — not in expectation. You can buy an income stream that no market can take away.
The FIRE community optimises for the earliest possible retirement date. The funded ratio optimises for the most certain possible retirement income. These objectives are in tension — and the discount rate is where the tension lives. My argument is not that you must work 8 extra years. It is that you should know what the certain price of your retirement is before you decide how much uncertainty to accept.
The Right Way to Use Both Rates
The two rates are not competing answers to the same question. They answer different questions:
| Question | Rate | What It Tells You |
|---|---|---|
| "What is the minimum I need to guarantee my spending?" | TIPS yield (~2.3%) | Your funded ratio — the no-arbitrage price of your retirement |
| "What will my portfolio probably generate?" | Expected return (~5%) | Your expected surplus — the gap between likely outcome and certain cost |
The funded ratio uses the first rate. It tells you where you stand today, at market prices, with no assumptions about future equity performance. If you are 100% funded at the TIPS rate, you can retire and never worry about markets again.
The expected return tells you how fast you might get to 100% funded — and how much surplus you might accumulate along the way. It belongs in your planning model. It does not belong in your liability discount rate.
This is what institutional investors do. They price the liability at the risk-free rate. They invest the portfolio at the expected return. They measure the gap between the two — the funded ratio — and manage it actively. When surplus is large, they take risk. When it is thin, they hedge. The two rates coexist, but they serve different functions.
For you: know your funded ratio (at 2.3%). Know your expected surplus (at 5%). Decide how much risk to take based on the gap. But never — never — price a certain obligation at an uncertain rate and call yourself funded. That is not optimism. It is mispricing.
The pension industry spent two decades fighting over the discount rate. The financial economists won the argument. The actuaries implemented the compromise. And US public pensions are still pretending the problem does not exist. You do not have to make the same mistake. Price your retirement at the rate that guarantees it. Everything above that is surplus — welcome, useful, but never to be confused with the cost of the promise.
You cannot eat expected returns. You can eat TIPS coupons.
This is Post 12 of The Funded Ratio — a series on amifunded.com applying pension fund mathematics to individual retirement portfolios. Built on the Campbell-Viceira (2002) intertemporal asset allocation framework, the lifecycle consumption theory (Merton, 1969; 1971), and the pension finance literature (Gold & Bader, 2007; Novy-Marx & Rauh, 2009). Asness, "The Long Run Is Lying to You" (AQR, 2021). No financial advice is given. All methods use publicly available academic research and market data.