Effective Marginal Rate and Benefit Cliffs, Explained

After reading this you can compute what you truly keep from the next unit you earn, spot where a raise can leave you poorer, and name which benefit causes the damage.

What the effective marginal rate actually measures

Your tax bracket tells you only part of the story. If you sit in a 30% bracket, an extra 100 earned costs 30 in tax, so you keep 70. That 30% is your statutory marginal rate. But many benefits shrink as income rises. If earning that same 100 also strips away 25 of a housing subsidy, you keep 45, not 70. Your true marginal rate is 55%, not 30%.

The effective marginal rate counts both effects: tax owed on the next unit plus benefits lost because of it. It answers a blunt question: of the next unit of gross income, how much reaches your pocket?

The hook is the extreme case. Suppose a benefit worth 4800 per year vanishes completely the moment your income crosses 20000. Earn 20001 instead of 20000, and one extra unit of gross income triggers the loss of 4800. You are worse off by roughly 4799. That is a benefit cliff, and its effective marginal rate is far above 100%.

When to use this, and when not

Use this when your income sits anywhere near a means-tested benefit that tapers or cuts off: child benefits, housing aid, income-tested credits, subsidised childcare, or student support. It is most useful before you accept a raise, add overtime hours, or take a second job, because those are exactly the moves that can push you across a threshold.

Do not use it as a tax filing tool. It models a simplified bracket schedule and linear phase-outs. Real systems layer in deductions, credits with their own rules, payroll taxes, and household composition. The tool is country-agnostic on purpose: you supply the numbers, so its accuracy is exactly the accuracy of what you enter.

A high effective marginal rate is not a reason to earn less on its own. Earning more almost always leaves you better off. The rare exception is a genuine cliff, a narrow band where the effective rate reaches or exceeds 100%. Those bands are the only place a raise can make you poorer.

The formula and the intuition behind it

Start with net income as a function of gross income g. Net income is gross minus tax plus whatever benefits survive at that income:

N(g) = g - T(g) + \sum_i B_i(g)

Here T(g) is total tax computed from the bracket schedule, and each B_i(g) is the amount of benefit i that remains at income g. A benefit that starts at its full amount A_i, begins phasing out at s_i, and reaches zero at e_i is defined piecewise:

B_i(g) = \begin{cases} A_i & g \le s_i \\ A_i \cdot \dfrac{e_i - g}{e_i - s_i} & s_i \lt g \lt e_i \\ 0 & g \ge e_i \end{cases}

The effective marginal rate is one minus the slope of net income. Slope means how much net income changes per unit of gross income:

\text{EMR}(g) = 1 - \frac{dN}{dg} = 1 - \left(1 - T'(g) + \sum_i B_i'(g)\right)

That simplifies to a clean statement: the effective marginal rate equals your statutory rate plus the sum of every active phase-out rate.

\text{EMR}(g) = T'(g) - \sum_i B_i'(g)

Each phase-out contributes a slope of -A_i / (e_i - s_i) while active, so its taper rate is A_i / (e_i - s_i). A benefit of 3000 phased out over a 15000-wide band adds 3000 / 15000 = 0.2, that is 20 percentage points, to your effective rate across that band. A hard cliff has zero width, so its slope is infinite: the tool treats it as a single point where net income drops by the full amount.

A worked example with the demo data

Reproducing income = 40000

Load the demo, which sets your income to 40000 against the default schedule. The tax brackets are: 0% is not listed, but the first line 0 10 means 10% from 0, then 20% from 12000, 30% from 50000, and 40% from 100000. The benefits are ChildBenefit 3000 30000 45000 and HousingAid 4800 20000 20000 (a hard cliff at 20000).

At 40000, which bracket applies? Income is between 12000 and 50000, so the statutory marginal rate is 20%.
Is HousingAid active? Its phase-out band is 20000 to 20000, a cliff already behind you. At 40000 it contributes 0 to the marginal rate (its damage happened at 20000).
Is ChildBenefit phasing out? Its band is 30000 to 45000, and 40000 is inside it. Its taper rate is 3000 / (45000 - 30000) = 3000 / 15000 = 0.2, that is 20 percentage points.
Add them: effective marginal rate = 20% + 20% = 40%. Of your next unit earned, you keep 0.60.

Compare that to your naive expectation. You are in a 20% bracket, so you might expect to keep 0.80. The disappearing ChildBenefit quietly doubles your real marginal rate.

The chart below shows the effective marginal rate across the whole income range for this schedule. Notice the flat 40% shelf between 30000 and 45000, and the vertical spike at the 20000 cliff.

The 40% shelf from 30000 to 45000 is the tax bracket plus the ChildBenefit taper. The narrow spike at 20000 is the HousingAid cliff, drawn tall because a point drop has no width.

Reading the two charts the tool draws

The tool sweeps the whole income range and produces two views. The first is net income against gross income. On a healthy schedule this line always rises: earning more leaves you with more. At a cliff, the net income line drops straight down. That downward step is the visual signature of being made poorer by a raise.

The second view is the effective marginal rate, like the chart above. Read it as a map of danger zones. Flat regions are ordinary brackets. Raised shelves are phase-out bands, and the height of the shelf tells you how many percentage points the taper adds. Any point at or above 100% is flagged: crossing it costs more than it pays.

Net income mostly rises, but drops sharply at the 20000 cliff where HousingAid disappears. Between 30000 and 45000 the line climbs more slowly because each unit earned is taxed and also shrinks ChildBenefit.

When you spot a cliff, the practical move is often to jump well past it rather than land just after it. If a raise takes you from 19000 to 21000, you cross the 20000 cliff and lose 4800. A larger raise to 26000 more than recovers the loss.

Watch a phase-out reshape your marginal rate

The single most instructive parameter is the width of a benefit's phase-out band. A wide band spreads the loss thinly. A narrow band concentrates it. Set the width to zero and you get a cliff.

Common mistakes

The first mistake is reading the tax bracket as the whole marginal rate. In the worked example the bracket was 20% but the effective rate was 40%. The gap is entirely the ChildBenefit taper.

The second mistake is ignoring stacked phase-outs. Two benefits each tapering at 25 percentage points over the same income band combine to 50 points. Add a 30% tax bracket and the effective rate is 80%. You keep 0.20 of every extra unit while inside that overlap. The tool sums all active phase-outs at each income, so the stacking is visible directly on the rate chart.

The third mistake is confusing a cliff with a steep taper. A cliff (end = start) makes net income fall. A steep taper only slows its rise. Both look alarming on the rate chart, but only the cliff can leave you strictly poorer, so only the cliff is worth reorganising your income to avoid.

The fourth mistake is using annual figures for one benefit and monthly for another. Every input here is yearly. Convert consistently before entering anything.

Related tools

Once you know your effective marginal rate, you can judge a raise properly. The Pay Raise Calculator shows the raise in percent, amount, and after inflation. To turn gross into net under ordinary deductions, use the Take-Home Pay Calculator. For the tax portion alone, the Income Tax Bracket Calculator handles bracket math directly.

If you are weighing extra hours against the effective rate they face, the Overtime Pay Calculator and the Salary to Hourly Calculator convert between pay bases. Freelancers setting a target income can start with the Freelance Hourly Rate Calculator, then check whether a chosen income lands them in a phase-out band here.

Frequently asked questions

Can the effective marginal rate really exceed 100%?

Yes, at cliffs and steep tapers. If a 3000 benefit tapers over a 2000 band, the taper rate is 3000 / 2000 = 1.5, that is 150 percentage points. Add a 20% tax and the effective rate is 170%. Across that band, every extra unit earned leaves you with less.

Does a high marginal rate mean I should turn down a raise?

Almost never. A high rate below 100% still means you keep something positive from each unit. Only a rate of 100% or more, a genuine cliff, makes more income leave you poorer, and even then only inside its narrow band. A big enough raise clears the band.

Why does the net income chart drop straight down at a cliff?

Because a cliff removes the whole benefit at one income. At 20000 the 4800 HousingAid disappears, so net income falls by roughly 4800 the instant you cross it. Between 19999 and 20001 your gross rose by 2 but your net fell by nearly 4800.

How do I model a benefit that never fully disappears?

This tool tapers each benefit to zero. To approximate a floor, enter only the portion that phases out. If a 5000 benefit shrinks to a permanent 2000, enter 3000 as the amount that tapers and treat the 2000 as untouched income elsewhere.

What if two benefits phase out over different but overlapping ranges?

The tool adds their taper rates only where they overlap. If one runs 20000 to 40000 and another 30000 to 50000, the band 30000 to 40000 carries both tapers stacked, and the rate chart shows a taller shelf there than on either side.

Try it yourself in the tool