Tax Brackets and Cliff Edges: Why Crossing a Threshold Can Cost You Money

October 4, 2026 · 6 min read

Two opposite things both get called "the tax bracket system", and confusing them explains almost every misstatement you will hear about progressive taxation. One is a well-designed structure that taxes each slice of income according to its size. The other is a design flaw that makes earning more leave you worse off. They have nothing in common, and the second one is far more often the story people tell about the first.

How brackets actually work

Brackets apply to slices of income, not to the whole amount. A common bracket structure:

  • 0% on the first $15,000
  • 10% on income from $15,000 to $50,000
  • 20% from $50,000 to $100,000
  • 30% above $100,000

Worked example at $70,000, ignoring allowances:

  • First $15,000 at 0% = $0
  • Next $35,000 at 10% = $3,500
  • Final $20,000 at 20% = $4,000
  • Total tax = $7,500, effective rate 10.71%

The common misreading is "I am in the 20% bracket, so 20% of my $70,000 is $14,000". That is wrong by $6,500, and the error is what makes progressive systems sound punitive when they are not. The tax bracket calculator shows the slice-by-slice breakdown rather than a single number.

What crossing a threshold actually costs

Take income moving from $48,000 to $55,000 under the structure above.

At $48,000: tax = 3,500 (everything above $15,000 at 10%), effective rate 7.29%.

At $55,000: tax = 3,500 + 1,000 = $4,500 (the $1,000 above $50,000 at 20%), effective rate 8.18%.

You earned $7,000 more and paid $1,000 more tax — 14.3% of the raise, not 20%. The threshold cost you the 20% on one thousand dollars, and nothing more. This is the ordinary arithmetic, and the marginal versus effective rate guide covers why the effective rate stays below the marginal one.

Now the part that genuinely bites: the cliff edge

A cliff edge exists where a benefit, allowance or tax credit is withdrawn abruptly as income passes a threshold. The discontinuity is what matters — nobody is taxed at 20% on the extra dollar; instead an entire benefit disappears at once.

Worked example. Suppose a benefit of $4,000 is available below $50,000 and withdrawn completely above it, alongside the standard brackets.

At $49,000: tax = 3,500, benefit = 4,000, so net position = 49,000 − 3,500 + 4,000 = $49,500.

At $50,000: tax = 4,000, benefit = 0, so net position = 50,000 − 4,000 = $46,000.

Earn one dollar more and you are $3,500 worse off. That is a cliff, and it is a genuine design failure — which is why careful systems taper the benefit rather than cutting it. The effect on behaviour is severe: some people deliberately earn less than they could, which is exactly the opposite of what a tax system is meant to encourage.

A notch is the milder version: the marginal tax rate on the extra income exceeds 100%, so the net is worse, but the loss is gradual rather than a single step. Above the threshold, the loss grows with every additional dollar earned rather than jumping.

How to spot which one you are facing

Three checks, each quick:

  1. Look for a taper, not a cutoff. A benefit that reduces gradually by a percentage of additional income produces a high marginal rate, not a cliff. A benefit that stops entirely produces a cliff. The wording usually says which: "reduced by 50% of the excess" is a taper; "not available if income exceeds" is a cliff.
  2. Test a point just above the threshold. Compare net income at the threshold minus $1 with net income at the threshold plus $1. If the second is lower, you have a cliff. The income tax calculator will do this comparison for a specific set of rules.
  3. Check whether the threshold is indexed. A threshold frozen for years while wages rise slowly is how a system that was progressive when written becomes punishing as incomes climb through it. This is a real and well-documented effect in several countries, and it is a legitimate criticism of a system that was designed for a different income distribution.

The thresholds worth knowing about

Most tax systems have a handful of places where behaviour changes sharply, and they cluster in a few places:

The personal allowance taper. The most common notch. As income rises, the allowance reduces by a set amount per $1 of income — 50p is a well-known example, which means the top half of your allowance costs you 50p of the marginal rate. Once your income is high enough that the allowance is fully reduced, further income is taxed normally and the marginal rate falls back.

Child tax credit thresholds. These commonly phase out entirely, historically producing a real cliff for families around the threshold. Many systems have since moved to a taper, and the direction of travel has been consistently toward tapering for exactly this reason.

Benefit eligibility cut-offs. Means-tested benefits in most systems, and the interaction with a tax allowance creates a combined threshold that is easy to overlook and expensive to cross.

Capital gains exemptions and higher-rate bands. Different thresholds, same structure.

Company size limits and other statutory thresholds, where the effect is similar in kind.

What to do about it

For an individual the practical steps are unglamorous but effective:

Map your own thresholds. The single most useful exercise. List every allowance, credit and benefit you or your household rely on, with the income level at which each changes. Most households find one or two, and knowing where they are is worth more than any general knowledge of the system.

Do the arithmetic either side before deciding. Not the arithmetic you expect — the actual calculation at one dollar below and one dollar above. If the answer is a cliff, you now know the number, and the decision becomes a straightforward comparison of a known loss against the gain.

Look for timing flexibility. Many thresholds can be managed by timing income — deferring a bonus, bringing forward expenses to use an allowance, choosing which year to realise a gain. This works for cliffs because the discontinuity is only in the timeline, not the total.

Do not confuse a notch with a cliff. A notch is a high marginal rate over a range; you may still come out ahead overall. A cliff is a step down. They need different responses, and conflating them leads to people making large decisions about a modest problem.

Frequently asked questions

Do tax brackets work the way people think?

No. Each slice of income is taxed at its own rate — a 20% bracket does not mean 20% of everything you earn. That misunderstanding is what makes a progressive system sound punitive when the actual arithmetic is usually favourable.

What is a cliff edge in taxation?

A point where a small increase in income causes a large decrease in net income, because a benefit or allowance is withdrawn abruptly. It is a design flaw rather than an intended feature, and it is what people mean when they say the system penalises earning more.

What is the difference between a cliff and a notch?

A cliff is where net income drops as gross income rises — you lose an entire benefit. A notch is where the marginal tax exceeds the extra income earned, so you would be better off not earning it, but the loss is gradual rather than a single step.

How do I avoid a cliff edge?

Map the thresholds in every benefit or allowance you rely on, and check the arithmetic at one dollar above and one dollar below each before making a decision that crosses it. Where a taper exists instead, the marginal rate is high but the loss is gradual and much easier to reason about.

Related guides