“I turned down the raise because it would push me into a higher bracket” is the most expensive misunderstanding in personal finance, and I have heard an adult with a real job say it out loud.

It is wrong, and the way it is wrong is worth understanding properly, because the same confusion underlies a lot of bad decisions about retirement accounts, side income, and whether a bonus is worth taking.

The thing people think happens #

The mental model is that brackets work like a switch. Earn $100,000 and you are in the 22% bracket, so you pay 22% on everything. Earn one dollar more, cross into the 24% bracket, and now you pay 24% on everything, which means that one extra dollar cost you thousands.

If that were how it worked, turning down a raise would sometimes be rational.

It is not how it works, and it never has been.

What actually happens #

Brackets apply to slices of income, not to all of it. Each bracket taxes only the dollars that fall inside its range.

Think of it as filling a set of buckets. The first bucket is taxed at 0%, and it holds an amount equal to your standard deduction. When that fills, income spills into the 10% bucket, then the 12% bucket, and so on. Each bucket taxes only what lands in it.

The federal brackets are currently 10, 12, 22, 24, 32, 35, and 37 percent. The dollar thresholds between them are indexed to inflation and move every year, so check IRS.gov for the current figures rather than trusting a blog post, this one included.

Work an example with round numbers. A single filer with $100,000 of gross income, taking the standard deduction of roughly $16,000, has about $84,000 of taxable income. That income gets sliced:

That person is “in the 22% bracket,” and their actual federal tax is about $13,400 on $100,000 of gross income, which is 13.4%, not 22%.

Marginal versus effective, which is the whole idea #

Two numbers, and confusing them is the source of nearly every mistake in this area.

Your marginal rate is what the next dollar you earn gets taxed at. In the example above, 22%.

Your effective rate is total tax divided by total income. In the example above, about 13.4%.

The marginal rate is always higher than the effective rate, because the effective rate is an average that includes all the cheaper slices underneath. This gap is not small. Someone in the 24% bracket often has an effective federal rate in the mid-teens.

Which resolves the raise question completely. A $5,000 raise for that person is taxed at 22% federal, so they keep about $3,900 of it before state tax and payroll tax. Their effective rate ticks up slightly. Their take-home pay goes up. Earning more money never reduces your take-home pay through the bracket system. The structure makes it arithmetically impossible.

Where the myth comes from #

It is not pure invention, and there are three real things underneath it.

Withholding on bonuses looks brutal. Supplemental income like a bonus is frequently withheld at a flat federal rate, historically 22%, plus state withholding, plus payroll taxes. A $10,000 bonus can land as $6,000 in your account, which feels like a 40% tax. It is not a tax, it is withholding, and the difference gets reconciled when you file. If too much was withheld, you get it back as a refund.

Payroll taxes are real and separate. Social Security and Medicare come out on top of income tax and they are not part of the bracket system. Social Security is 6.2% up to an annual wage cap, and Medicare is 1.45% with no cap. An employee sees 7.65% before income tax has done anything, which makes the total bite larger than the bracket table alone suggests. Self-employed people pay both halves, which is 15.3%, and this is the biggest surprise for anyone going independent.

Benefit cliffs genuinely exist. This is the legitimate core of the concern, and it is covered below.

The cliffs that are real #

Brackets are smooth. Some other things in the tax and benefits system are not, and at those specific edges an extra dollar of income really can cost you more than a dollar.

Health insurance subsidies. Premium tax credits on the ACA marketplace phase out with income, and depending on the rules in force in a given year the phase-out can be gradual or can end abruptly at a threshold. Where a hard cliff applies, a small amount of additional income can cost thousands in subsidy. This is the most consequential real cliff for ordinary earners, and it is worth checking against the current year’s rules if you are near the line.

Roth IRA contribution limits. Direct Roth IRA contributions phase out over an income range and then stop. Crossing it does not cost you money directly, and it removes an option, and the backdoor Roth exists precisely as the workaround.

Income-driven student loan payments, various means-tested benefits, and some state and local programs have their own thresholds.

Medicare IRMAA, which raises Medicare premiums for higher-income retirees, is a genuine cliff rather than a phase-out, and it is one of the reasons retirement withdrawal planning cares about managing taxable income year by year.

None of those are brackets. All of them are worth knowing if you are near one. And for the vast majority of people in the vast majority of years, none of them apply, and the raise is simply worth taking.

California on top #

If you live here, state tax is a second, separate stack of brackets applied to a similar but not identical measure of income.

California’s rates run from 1% up to 12.3%, plus an additional 1% mental health services surcharge on income above one million dollars, which is where the widely quoted 13.3% top rate comes from. The brackets are their own, indexed separately, and the state has no preferential rate for long-term capital gains, which are taxed as ordinary income at the state level.

For a middle-income earner in San Francisco, the combined marginal rate on the next dollar is commonly in the low-to-mid thirties once you add federal, state, and the employee share of payroll tax. That number is the one worth knowing, because it is what a raise, a bonus, or a side income dollar is actually worth to you.

It also sets the value of every pre-tax dollar. A traditional 401(k) contribution for that person saves federal and state tax at the combined marginal rate, which is why the traditional-versus-Roth question tilts the way it does in a high-tax state, especially if you might retire somewhere without an income tax.

What this changes in practice #

Take the raise. Always. There is no bracket at which more gross income means less net income.

Know your marginal rate, not your bracket. Federal plus state plus payroll. It is the number that tells you what a deduction is worth, what a pre-tax contribution saves you, and what an hour of side work actually pays.

Value deductions at the marginal rate. A $1,000 deduction saves you $1,000 times your marginal rate, not $1,000. This is why people are consistently disappointed by deductions and consistently surprised by credits, which reduce tax dollar for dollar and are worth far more.

Adjust withholding rather than celebrating a refund. A large refund is an interest-free loan you made to the government. The IRS withholding estimator will get your W-4 close, and the money is more useful in a high-yield account during the year than as a lump in April.

Check the cliffs if you are near one. Marketplace subsidies especially.

One thing worth saying plainly #

I am not a tax professional and none of this is tax advice. Brackets, thresholds, and phase-outs change every single year, and several change more often than that. Everything above is the structure, which is stable, rather than the numbers, which are not.

If your situation involves equity compensation, self-employment, rental income, multiple states, or anything unusual, an hour with a CPA costs a few hundred dollars and routinely saves multiples of that. That is one of the few places where paying for expertise is straightforwardly correct, and it is the same reasoning as the rest of Money: spend where the return is real, and stop paying for defaults nobody checked.