Best investment accounts for kids: custodial, 529 and beyond

Saving for a child starts with an unfair advantage no adult account ever gets: an 18-year runway before the money has a job to do. The account-type debate, 529 or UTMA or Roth or bonds, is worth having, but it is a second-order debate about taxes, control and financial aid layered on top of a first-order fact about time. This page takes the accounts in order, rules dated and sourced, and keeps the first-order fact in view throughout. Every threshold below carries its tax year, because in this domain an undated number is already a suspect one.

Nothing here assumes prior investing vocabulary; where foundations help, learning to invest from scratch is the base camp this page builds from.

1. The 18-year edge

Eighteen years is long enough to change what an account is. At the horizons most savings decisions live on, one to five years, outcomes are dominated by luck and rates, which is why short-term investment options is a page about safety and access rather than growth. At eighteen years, the compounding term takes over: a fixed monthly contribution multiplies through more than two hundred compounding periods, and the growth term, explorable at any rate and horizon in the compound interest calculator, starts to dwarf the contribution total itself. What this page adds to that generic arithmetic is the historical distribution specific to the 18-year window, computed rather than assumed. The historical record makes the edge concrete, and this page computed it rather than asserting it from folklore: across every rolling 18-year window on US stocks since 1871, 1,617 of them, the median outcome was 8.3% a year nominal, the 10th percentile 5.5%, the 90th 14.7%, and no 18-year window in the sample compounded below 1.4% a year in nominal terms (Eco3min calculations, Shiller total return data, dividends reinvested, 1871–2023).

Two honest caveats keep the edge from becoming a slogan, and both are structural rather than rhetorical. Those figures describe one market’s past, the best-documented equity record in existence but still one country’s, not any account’s future; and they are nominal, which flatters them systematically, a correction section 5 makes in full. But the structural point survives both caveats: a child’s account is the one place in household finance where the calendar itself is the largest asset, and where starting at birth versus starting at ten changes the outcome more than most fund choices ever will.

The arithmetic of that comparison deserves one concrete line. At the historical median of 8.3% a year, a contribution stream started at birth compounds through eighteen full years; the same stream started at age ten gets eight. The late stream does not merely end smaller by the missing contributions; it ends smaller by the missing compounding on every contribution, which is the larger of the two losses and the one no later generosity fully repairs. This is also why the account-opening date matters in ways the rules quietly reward: the 529’s fifteen-year clock for Roth rollovers, for instance, starts at account opening, so an account opened at birth clears the gate at fifteen, while the same account opened at age ten keeps its beneficiary waiting until twenty-five.

2. The account types

Four structures carry most of the American conversation, and they differ on rules rather than on investments: the same index fund can live inside three of them, at three different tax and control outcomes. What varies is what goes in, how growth is taxed, who controls the account and when, and what it does to a financial aid application. The table compresses the comparison; the subsections date the load-bearing rules, and every figure in both carries its tax year.

529 plans

The 529 is the education wrapper, run by states with federal tax treatment, and its rules have widened twice in recent legislation, each widening dated below because the details move. Growth is federally tax-free for qualified education uses, a list that 2025 legislation extended: from 2026, K-12 withdrawals are allowed up to $20,000 a year per student, and credentialing programs qualify. The escape hatch for over-saving is dated and capped: under SECURE 2.0, up to $35,000 lifetime can roll to a Roth IRA owned by the beneficiary, provided the account is at least 15 years old, the funds at least 5 years old, and each year’s rollover fits inside the beneficiary’s Roth contribution limit ($7,500 in 2026) with matching earned income. State tax benefits exist as a general category and vary by state; they are a reason to read one’s own state plan, not a national rule. Two structural comforts complete the picture: the beneficiary can be changed to another eligible family member without tax consequences, which converts an over-funded account for one child into a head start for another; and the owner, not the child, controls the account throughout, the exact opposite of the custodial design below.

UTMA/UGMA custodial accounts

The custodial account is the open-ended wrapper: any asset, any purpose for the child’s benefit while a minor, no contribution ceiling beyond gift-tax mechanics, and a legal design older than every other row of the table. Its two structural features are simple to state, permanent in effect, and the ones families most often discover late, usually in that order. The gift is irrevocable and control transfers by law at the age of majority, 18 to 25 depending on the state: the eighteen-year-old owns the money, whatever the original intent, and no clause in the paperwork restores the giver’s hand. Families comfortable with that transfer treat it as the point; families uncomfortable with it discover the account’s defining feature at its least convenient moment. And the earnings are taxable each year under the kiddie tax: in 2026, the first $1,350 of a child’s unearned income is untaxed, the next $1,350 is taxed at the child’s rate, and everything above $2,700 at the parents’ marginal rate (IRS Form 8615 rules). The investing mechanics inside the account are ordinary brokerage mechanics, for which picking a brokerage account supplies the checklist. For cash parked along the way inside any of these wrappers, high-yield savings versus money market compares the two standard homes on an after-tax basis.

Roth IRA for minors

The minor’s Roth is gated by one requirement that is easy to state and strict in practice: contributions require the child’s own earned income, wages or self-employment, up to the lesser of that income or the annual limit ($7,500 in 2026). Allowance and investment income do not qualify, and the IRS standard is documented compensation, which is why the family-business payroll shortcut attracts scrutiny in proportion to its creativity. For a teenager with a real W-2, the structure is unusually clean: decades of tax-free compounding, contributions withdrawable without penalty, and no weight in federal aid formulas, which exclude retirement assets. The account is custodial until majority; the earned-income gate, not the paperwork, is the real constraint. One mechanical footnote connects it back to the 529: the earned-income requirement also governs 529-to-Roth rollovers, so the teenager’s summer job is the key that unlocks both doors at once, the direct contribution and the rollover alike.

Savings bonds

Series I and EE bonds are the formula-locked corner of the map: rates set by published formulas rather than markets, federal tax deferred until redemption, state and local tax exempt, annual purchase caps per person, and an education-use interest exclusion subject to income limits and conditions. How the two formulas actually work, and what they pay in the current environment, is the territory of series I and EE savings bonds; on this page they stand as the reminder that not every child’s account needs market exposure to have a defined role. In an 18-year design they are the anchor line, the position whose nominal value is never in question, and section 5 prices what that certainty costs in real terms.

3. Control, taxes and aid: the observable trade-offs

Set side by side, the four structures price three questions, and every answer is a rule rather than a forecast. Control: the 529 owner keeps it indefinitely and can change beneficiaries; the custodial gift surrenders it at a statutory birthday; the Roth transfers at majority; bonds follow registration. Taxes: the sheltered wrappers defer or eliminate tax on growth while the custodial account generates taxable income annually, the general trade-off sheltered versus taxable investing maps in full. Aid: the federal formula assesses student-owned assets at 20% and parental assets, including parent-owned 529s, at a maximum of 5.64% (FAFSA rules), so identical dollars weigh differently on an aid application depending on whose name they sit under, a gap of roughly three and a half times on the assessed amount. None of these questions has a universally right answer; all three have dated, checkable ones, which is what separates an account choice from a guess.

The trade-offs also interact, which is where families most often mis-sort. The custodial account’s tax cost and aid weight are the price of its open purpose; the 529’s favorable treatment is the price of its purpose restriction, softened but not removed by the Roth escape hatch; the Roth’s cleanliness is gated by income the child must actually earn. Choosing among them is therefore not ranking three products but matching three restrictions to what a family actually knows about an eighteen-year future, which is usually less than the account forms assume.

4. Gift limits and funding mechanics

Feeding any of these accounts runs through the same federal gift machinery, which is worth one factual section because it is where the account types stop differing. In 2026, gifts up to $19,000 per giver per recipient ($38,000 for a married couple electing to split gifts) stay under the annual federal exclusion, and 529s carry a specific accelerator: five years of annual exclusions can be front-loaded into a single contribution, $95,000 per giver per beneficiary ($190,000 for a couple electing gift-splitting), a mechanism aimed precisely at the long-horizon logic of section 1, since it moves the compounding start date rather than the total (IRS rules; provider documentation, 2026). Above those amounts, gifts consume lifetime exemption rather than triggering immediate tax in most cases, but they enter reporting territory. The mechanics are factual and this page states them as such; how any family should use them is estate planning, which is neither this page’s mandate nor its competence. One reading habit transfers regardless: gift limits are per giver per recipient, so the funding capacity of a child’s account scales with the number of adults involved, grandparents included, and the paperwork thresholds are annual, which rewards planning by calendar rather than by impulse.

5. What 18 years compound into, in real terms

The nominal figures of section 1 need their deflator, because eighteen years is long enough for inflation to change the meaning of every number on the statement. Rerunning the same 1,617 windows in purchasing-power terms moves the median from 8.3% to 6.6% a year and the 10th percentile from 5.5% to 2.0%, and, most instructively, breaks the “no negative windows” record: the worst 18-year stretch compounded at −0.9% a year in purchasing power, and about one window in a hundred ended below zero real (Eco3min calculations, Shiller data, 1871–2023). The lesson is not that the edge disappears; roughly ninety-nine windows in a hundred still ended positive in purchasing power. It is that the edge is real-denominated, the discipline why real returns matter formalizes. A child’s account measured in nominal euros or dollars is a scoreboard; measured in purchasing power at age 18, it is an answer.

The real-terms frame also settles the quiet competition between the market wrappers and the guaranteed corners of the map. A regulated or formula-locked rate that sits below inflation preserves the number and erodes the answer; a market path accepts visible volatility in exchange for the historical tendency, not guarantee, of outrunning the deflator over long windows. Neither is a mistake; they are different purchases, certainty of the nominal figure versus probability of the real one, and an eighteen-year horizon is precisely the setting where that distinction earns its keep.

6. The 18-year edge, simulated

The simulator compounds a monthly contribution from a chosen starting age to 18, under two documented paths rather than one promise: a regulated-savings path at a fixed low rate, certain but low, and a market band spanning the 10th-percentile-to-median range of the 1,617 historical 18-year windows computed above, 5.5% to 8.3% a year in nominal terms. The real-terms line, deflated at the inflation you set, overlays the median path. The asymmetry is stated rather than hidden: regulated rates are near-certain and revisable; market ranges are historical and not guaranteed. Read the gap between the nominal and real lines as carefully as the totals: it is the inflation assumption at work, and it is the part of the outcome no account structure controls. Nothing displayed is a projection of any product.

7. Read through the regime

An account opened this year will close in 2044, and it will cross macro regimes, plural, on the way. The real-terms arithmetic of section 5 is regime arithmetic in disguise: the windows that went negative in purchasing power are the ones that met sustained inflation, the environment documented in the inflationary regime explained. Eighteen years starting today will contain regimes nobody can name yet, which is an argument for understanding the mechanism rather than betting the account on any single configuration. Locating the present is a measurement, not a forecast: the current regime classification reads the classifier’s state, and asset classes under every regime shows what each configuration historically did to the building blocks these accounts hold. The account structure question of sections 2 through 4 does not change with the regime; what the account holds, and what its statements mean in real terms, does.

8. FAQ

What account types can be opened for a minor?

Four main structures: 529 plans (education-purposed, owner-controlled), UTMA/UGMA custodial accounts (open-purpose, irrevocable, child takes control at majority), custodial Roth IRAs (require the child’s earned income) and savings bonds (formula-rate instruments). Checking, savings and youth brokerage accounts exist alongside as cash-management layers. Each structure’s gate, purpose, income or control, matters more than its marketing, and the gates are checkable in the account agreements before any money moves.

How does the kiddie tax apply to custodial accounts?

Custodial account earnings, interest, dividends and realized gains, are the child’s unearned income, and in 2026 the first $1,350 is untaxed, the next $1,350 is taxed at the child’s rate, and amounts above $2,700 at the parents’ marginal rate, reported on Form 8615. The rule reaches dependents under 19, and full-time students under 24. Qualified 529 withdrawals generate no unearned income and sit outside the rule entirely, which is one of the quiet structural differences between the two wrappers.

What can 529 funds be used for?

Qualified higher-education expenses, tuition, fees, books, room and board within limits; K-12 costs up to $20,000 a year per student from 2026, with an expanded expense list; career credentialing and licensing programs; and, as an exit for leftover funds, up to $35,000 lifetime rolled into the beneficiary’s Roth IRA under the SECURE 2.0 conditions (account 15+ years old, funds 5+ years old, annual Roth limits, earned income). Non-qualified withdrawals tax the earnings portion and generally add a 10% federal penalty, which is the boundary that gives every other rule on the list its meaning.

When can a minor contribute to a Roth IRA?

As soon as they have earned income, wages or self-employment, and only up to the lesser of that income or the annual limit ($7,500 in 2026). A parent or guardian opens and manages the custodial Roth until the age of majority, at which point it converts to an ordinary Roth in the child’s hands. Sizing the habit matters more than maximizing the contribution in any single year; sizing a monthly contribution treats that question on its own terms.

How do custodial accounts affect financial aid calculations?

They weigh in as student assets, assessed at 20% in the federal aid formula, against a maximum of 5.64% for parental assets, a category that includes parent-owned 529 plans. On identical balances, the custodial structure can therefore reduce need-based aid several times more than the 529 structure, a mechanical difference in the formula rather than a judgment about either account. Aid formulas are also revised periodically, which is one more dated rule to re-check in the year it matters.

9. The horizon, then the wrapper

The reading order this page defends is the one the arithmetic imposes: the horizon first, because eighteen years of compounding is the account’s largest asset and it depreciates one year at a time, silently and irreversibly; the wrapper second, chosen on dated rules about taxes, control and aid rather than on brand familiarity; the contents last, built from the same blocks as any long portfolio and judged in real terms. Which vehicle for which regime connects the contents to their environment, and long-horizon allocation strategies places the whole account inside the family’s plan. For a child’s account, the horizon does most of the compounding work; the family’s job is mostly to not interrupt it. Every rule cited above has a date attached, and rules with dates get amended; the horizon’s arithmetic is the one part of the design that never files an update.

Last updated: 8 July 2026

Last updated — 8 July 2026