The Parts That Move the Number Most

Required minimum distributions and the bracket spike

Under SECURE 2.0 the age at which you must start drawing from tax-deferred accounts is 73 if you were born before 1960, and 75 if you were born in 1960 or later. That distribution arrives whether or not you need the income, and for households who saved diligently for thirty years it can produce a higher tax bracket in retirement than they ever had while working. The reason is not the distribution by itself — it is what the distribution stacks on top of, in three specific ways.

It stacks on income you already have

A required distribution lands on top of Social Security, any pension, and whatever else you draw that year. A household comfortably inside a lower bracket on its ordinary income can be pushed into the next one, or two, by the distribution alone — and the increase applies to the year it lands in, not spread across the years the balance was built.

It increases how much of your Social Security is taxed

How much of your Social Security benefit becomes taxable is driven by combined income, and a required distribution raises it. Above $44,000 of combined income for a married couple filing jointly — or $34,000 filing single — up to 85% of the benefit can be taxable. Those thresholds are not indexed to inflation, so more households cross them every year without anything changing on their end.

It can trigger a Medicare surcharge, and that one is a genuine cliff

Medicare Part B and Part D premiums are income-tested, and the test looks back two years. Ordinary tax brackets are steps — only the income above the line is taxed at the higher rate. The Medicare surcharge is not a step. Cross the threshold by a single dollar and the full surcharge for that tier applies. A large distribution or conversion this year therefore sets your premium two years out, regardless of what your income does in between.

All of which is why this work happens in the years before required distributions begin rather than in the year they start. The window between your last paycheck and your first required distribution is usually the lowest-bracket stretch you will ever have, and it is the one most commonly left unused.

What Connecticut does differently

Retiring in Connecticut changes the arithmetic in ways that a federal-only plan misses. Four mechanics come up in almost every plan we build, and two of them are commonly described incorrectly — including by people who ought to know better.

Connecticut exempts retirement income, but only below an income line

Social Security is fully exempt from Connecticut tax below $75,000 of AGI filing single, or $100,000 filing jointly. Above that line, no more than 25% of the benefit becomes taxable to the state — a two-tier structure rather than a gradual taper. Pension, 401(k), 403(b) and 457(b) income works differently again: it phases out on a graduated scale, fully exempt at the same floors and reaching zero at $100,000 single or $150,000 jointly. Traditional IRA income reaches full eligibility for that same treatment in tax year 2026.

A Roth conversion is fully taxable in Connecticut — and it compounds

Connecticut's retirement-income exemptions explicitly exclude Roth accounts, and because the state starts from federal AGI, a converted amount is fully taxable at Connecticut's 2% to 6.99% rates with no state-level carve-out. The part people miss is the second-order effect: a large conversion raises AGI, and a higher AGI can shrink or eliminate the exemption on the pension, IRA and Social Security income you are receiving in that same year. The conversion has a state cost, and it can also hand you a state bill on income that would otherwise have been exempt.

The Connecticut estate tax is not a cliff

This is worth stating plainly because the opposite is repeated constantly. Connecticut was a true cliff before 2011, when a dollar over the threshold could cost six figures. It is not one now. Since a 2023 change the state taxes only the excess over the exemption, at a flat 12%. With a $15 million exemption per individual in 2026, an estate two million dollars above it owes tax on that two million, not on the whole estate. What does matter is that Connecticut offers no portability between spouses — so leaving everything outright to a surviving spouse can waste the first spouse's exemption entirely, which is what makes bypass and credit-shelter planning relevant for couples near the threshold. Connecticut is also the only state with a unified companion gift tax under the same exemption.

There is no Connecticut capital gains surcharge

Another one worth correcting. Connecticut has no capital-gains-specific tax at all; gains are simply ordinary income at the same 2% to 6.99% graduated rates as everything else. The only historical reference to a separate levy is a 1990 provision that Connecticut's own Department of Revenue Services flags as superseded. The real planning fact is the inverse of a surcharge: Connecticut gives no preferential rate for long-term gains the way the federal system does. A gain you would pay a reduced federal rate on is taxed by the state at your ordinary rate — which changes where it makes sense to hold what.

None of these are permanent. Connecticut's thresholds are set by budget act and are not automatically indexed to inflation, so exposure is worth re-checking periodically rather than assuming a number you read a few years ago still applies.

Three of the biggest risks to a retirement plan

Portfolios get most of the attention. In practice, the things that damage a retirement plan tend to be structural rather than about security selection, and three of them account for a large share of the problems we are asked to fix.

The order returns arrive in, not the average

Two retirees can earn identical average returns and end up in very different places, depending entirely on when the bad years land. It is usually described as the risk of a bad first year, which understates it. Michael Kitces' research found only a weak relationship — a correlation of about 0.21 — between a portfolio's first-year return and the withdrawal rate it can sustain. Run the same analysis against cumulative real return over roughly the first decade and the correlation rises to about 0.80. It is not the bad year that does the damage; it is the bad decade. The practical response is holding near-term spending in short-duration assets so a downturn is funded from the reserve rather than by selling depressed holdings.

Tax risk, which is mostly the risk of having no plan

A large tax-deferred balance is not simply savings; a portion of it belongs to a future tax bill that nobody has yet calculated. Ed Slott's framing is that this compounds the longer it goes unmanaged, because the balance grows and the eventual distributions grow with it. The counterweight is doing deliberate work in the low-bracket years — conversions sized to a target bracket rather than a round number — which functions less as a bet on future rates than as insurance against not knowing them.

Inflation, measured over the whole retirement rather than the first year

A retirement that has to fund twenty to thirty years is exposed to inflation for the whole of it, and the erosion is not intuitive at the front end because the first few years look manageable. Vanguard and Morningstar both treat inflation as a top-tier retirement risk rather than a background assumption, and so do we: the question is not what a plan costs today but what the same standard of living costs in year twenty-five.

These are three of the significant risks, not all of them — longevity and healthcare costs sit on the same tier and get their own treatment in a plan. What the three have in common is that each is best addressed by a decision made in advance rather than a reaction made in the moment.

Building the portfolio around the income

A portfolio built to accumulate and a portfolio built to distribute are not the same portfolio, even when they hold similar things. While you are saving, a decline is an inconvenience and arguably an opportunity. Once you are drawing an income from it, the same decline is a forced sale. So the allocation stops being a question of abstract risk tolerance and becomes a question of funding dated, sequenced withdrawals: what has to be spendable in the next few years is held in short-duration assets, and only what is genuinely long-dated carries market risk. Mapping known income — Social Security, a pension — against required distributions and voluntary withdrawals also shows which years land in a lower bracket than the ones around them, and those are the years worth using for a conversion or a change in asset location.

Retirement income here starts with a chosen withdrawal rate and a dollar amount. Each year that amount adjusts for inflation, unless the portfolio has had a negative return over the prior twelve months. If strong performance pushes your withdrawal rate well below where it started, spending can increase; if a downturn pushes it well above, spending is trimmed. The point of setting the boundaries in advance is that you know before it happens what would cause your income to move and by how much — which is a very different experience from deciding what to do in the middle of a bad market.