Put the client's health spending on the portfolio page

The client arrives with the binder—Social Security estimates, a withdrawal schedule that holds at 4%, a Roth conversion ladder mapped through the early sixties, and a travel line indexed at 3%—but there is no line for health. The couple is 62, both still working part time, both carrying an employer plan that ends when they stop, and both healthy, which is why the line is missing and why the plan has a hole in the middle of it.

The dollar figures here are illustrative; the structure is the part worth copying, because the mistake most files make is answering the healthcare question with a single number when what a client actually needs is a position: an inflation rate, a funding source, and a date on which it gets revised.

Health spending behaves unlike anything else in a retirement budget: it does not shrink when a client cuts back, it does not track the general price level, and it is not discretionary the way travel is. Marsh projects health costs to climb 8.2% over a single year rather than a quarter century, and that figure should be carried with the caveat attached, because compounded across a retirement it becomes a bigger assumption than most plans make about returns—at that pace, medical spending doubles in about nine years and multiplies roughly sevenfold in twenty-five.

Joseph Coughlin, who leads MIT's AgeLab, has argued that advisors should treat monthly health spending as a managed position rather than a personal line item, and his October paper hands them a script for the conversation. Health is the last retirement input most advisors miss, and the file is how it stops being missed.

This couple's own ledger—premiums, dental, vision, and the out-of-pocket that never quite makes it into a budget—runs about $14,000 a year. Against a $9,000 monthly spending target, health is already roughly 13% of the plan, and it is the only line in that budget with a trend rate attached that runs well past general inflation.

Hold the ledger at Marsh's rate and the health line reaches about $31,000 in ten years and $68,000 in twenty; at 3% as the comparison case, the same $14,000 becomes $18,800 in ten years and $25,300 in twenty. A 3% assumption and an 8.2% assumption, applied to the same client, are $42,400 apart in year twenty, a gap that lands in the withdrawal rate rather than in a footnote.

Marsh is projecting a single year's trend, not a quarter century of it, and 8.2% will not hold for twenty-five years. The problem is the shape of the error: health is the one line in a retirement budget that rises because the client is getting older rather than because the client chose to spend, so an underestimate does not average out across the plan—it accumulates on the wrong side of the ledger and arrives in the years when the client has the least room to adjust. The bridge between the employer plan and Medicare eligibility gets its own row in the file for the same reason: the rules change on the other side of 65, and so does the client's ability to shop.

The first approach is already in the binder—fold health into general inflation and move on—and it requires nothing, produces no number to review, and understates the liability year after year, right up until the client is 78 and the withdrawal rate no longer works.

The second is a national average: one published lifetime health-cost estimate, applied at the first meeting and rarely revisited. It helps less than a file because a lifetime average is a stock where the plan needs a flow—an annual number to fund, a trigger for revisiting it, and a way to see whether the client's own spending is drifting away from the average. It also imports somebody else's distribution, blending the client who dies at 68 after a year of nursing care with the client who lives to 97 on a pharmacy regimen; the plan has to be funded for the second one.

The third is the recommendation: give health its own sleeve with three properties. It carries its own inflation rate, re-marked every year against the client's actual trailing spending, pulled from statements and pharmacy receipts rather than from a survey; it carries its own funding source; and it carries its own date on the calendar, because a position reviewed whenever it comes up is a position that never gets reviewed.

Health spending in the first decade of retirement is the least deferrable money in the plan, which makes it a portfolio decision before it is a tax decision: a client can skip a trip in a bad market, but cannot skip a prescription in a bad market or schedule a procedure around a drawdown, and sequence risk does its worst damage where spending is inflexible. So the near-term health line belongs in the reliable assets—cash, short duration, the money the client needs inside ten years—while the tail beyond it stays in the growth portfolio with everything else. Discounted, a decade of this couple's health line is on the order of $200,000 of the balance sheet doing nothing but covering health and taking no market risk to do it.

Implementation is brisk and unglamorous: the sleeve gets a line in the investment policy statement, a target funded amount, a maturity ladder matched to the next decade of health spending, and a rule for refilling it out of the growth assets after a strong year rather than in the middle of a weak one.

Then stress the line against the two events that move it: Medicare premiums reset annually and drug coverage rules reset with them, so a plan funded to the average premium is quietly unfunded against the year the premium lands above the average, the same asymmetry as the inflation assumption one level down. The tail, a year of custodial care the health line was never sized for, is answered with a separate reserve rather than a bigger withdrawal, since the plan should not be sized to the worst year in every year.

Two second-order effects belong in the file because they change the rest of the plan. The first is the withdrawal rate: a mandatory line growing faster than the portfolio's distribution leaves the sustainable rate on the remaining budget lower than the rate on the whole of it, so a client who has been told 4% has been told a number that is too high. The second is Social Security claiming: the larger the share of the budget that is a rising, non-negotiable cost, the more the client needs the one inflation-indexed income stream available, which tilts claiming later rather than earlier—the opposite of what a plan concludes when health is folded into 3% general inflation.

Coughlin's case is that the position gets managed monthly; the advisory version is one review every twelve months, and the detail that decides whether it happens is where it sits on the calendar—put it in the tax meeting. The client is already gathering statements, drug-plan documents arrive in the same season, and the conversation that produces a return can produce a revised health number in the same hour. The deliverable is a single page: last year's actual spending from statements and receipts, the inflation rate the plan will carry next, the funded amount and when it matures, and the result of the premium stress test.

How the page gets presented matters as much as what is on it. A number that arrives as a prediction invites an argument; a number built from the client's own receipts invites a correction. The framing that works at the desk is to treat health as a funding requirement, the same way a plan treats a mortgage, which no client describes as a prediction about housing.

The assets side of the plan gets marked constantly and the health side gets marked almost never, which is how the two drift. Fidelity's average 401(k) balance sat at $155,800, up 10.5%, alongside a record number of millionaire accounts in the same data—rising balances sitting on an assumption from the first meeting, with nothing in a quarterly statement to suggest the liability moved. It moved. At Marsh's trend rate, this couple will be spending close to five times on health in twenty years what they spend today.

Pull ten written plans and count how many carry an annual health figure, an inflation rate attached to it, and a date on which it was last revised; most practices will count fewer than they expect. The couple in this file will spend about $14,000 on health this year and something near $68,000 in twenty, and the only version of that second number a plan can fund is the one an advisor has written down and dated.