The Hidden Slope in a Tax Schedule
The Dutch experiment’s crucial object is not the headline tax bracket but the marginal return hidden inside the tax-and-transfer schedule. This chapter explains how to read that return: what happens to disposable income when gross income rises by €100, why credits and withdrawn benefits can change the answer without changing the statutory rate, and why an individual effective marginal tax rate must not be confused with an average tax burden or an OECD tax wedge. That distinction gives the study’s later evidence a firm economic reference point.
A tax schedule is a stack of slopes
Suppose a person’s gross monthly income increases by €100. The person does not keep a fixed percentage simply because a statutory income-tax rate is printed next to their income band. The €100 may generate additional income tax. It may also change the value of tax credits or means-tested allowances. Some provisions add to disposable income as earnings rise; others are withdrawn. The relevant result is the combined change in the person’s disposable income.
If disposable income rises by €60, then €40 of the gross increment has been absorbed by taxes or reduced transfers. The effective marginal tax rate is therefore 40% for that increment. If disposable income rises by €90, the effective marginal rate is 10%. The calculation concerns the next €100, not the person’s entire income. It is marginal because it asks what happens at the edge of the current position.
This is the question the researchers put to participants in the Dutch study. They asked respondents to estimate how much they would gain on net if their gross income increased by €100 per month. Framing the question this way avoids a common ambiguity. A person may know that their statutory rate is somewhere around a particular percentage and still not know what an additional hour of work would add to their household’s disposable income. The second answer is the one that bears directly on an hours decision.
The Dutch system makes the calculation difficult because several kinds of provisions meet at that edge. Alongside progressive income tax, the system studied includes income-dependent tax credits and means-tested allowances. Their eligibility or value can depend on gross income and on circumstances such as household composition, childcare, housing, and other characteristics. Each provision may be understandable in isolation. The difficulty arises when their slopes overlap.
The paper illustrates the problem with a 2024 example of a single household with two children aged six to eleven, renting a home for €697 per month. In that example, the statutory income-tax line is about 37% through lower and middle income ranges. Yet the first €9,100 of taxable income is effectively tax-free because of the general tax credit. That credit does not make the statutory line disappear; it changes the relationship between gross income and disposable income.
Other credits add different shapes to the schedule. The combination credit and earned-income tax credit have their own phase-in and phase-out patterns. Several allowances begin at a positive level, remain stable for a while, and then start declining as income rises. In the example, childcare, child-budget, rent, and health-insurance allowances begin phasing out at approximately €27,000 of taxable income. A child-benefit scheme remains independent of income. The point is not to memorize these provisions. It is to see why a person would have to combine several changing lines to know the return on the next euro.
A phase-out is especially important because it acts like an additional marginal tax. If an allowance falls by €20 when gross income rises by €100, that withdrawal absorbs 20% of the increment even though it is not labelled an income-tax bracket. Several withdrawals can operate at once. In the ranges where the Dutch allowances and credits are being withdrawn, the paper reports effective marginal rates that can reach 80–90% for some representative workers. This does not mean that every worker faces such a rate. It means that a schedule designed from separate policy choices can produce a combined slope far steeper than its most visible component.
The reverse can happen at low incomes. A credit or allowance can initially rise or remain intact as earnings increase. The person may therefore keep most of the next €100 even while a statutory income-tax line exists on paper. At higher income levels, credits and allowances may have fully phased out, reducing some of the heterogeneity. But the slow phase-out of the earned-income tax credit can keep the effective marginal rate above the statutory rate of about 50% for many people in the higher-income range described by the paper.
The schedule is thus better understood as an interface between several policy purposes. Progressive taxation can raise revenue and redistribute income. Credits and allowances can target support toward households judged to need it. Earned-income provisions can be intended to strengthen work incentives. When these components meet, however, the marginal return is governed by their combined movement. A policy can be well-motivated at the component level and still be difficult for a household to interpret at the point where a work decision is made.
The direction of a provision’s movement matters as much as its existence. A credit that phases in can make the next increment more valuable; a benefit that phases out can make it less valuable. If a household crosses two thresholds at nearly the same time, a modest increase in gross income can trigger several changes in disposable income at once. The schedule may be continuous, but its slope can change sharply. That is enough to make two people with similar wages face different marginal returns, or to make one person’s return change as their income moves through the schedule.
This creates a policy trade-off rather than a simple design mistake. Targeted support is meant to reach households with particular needs, and withdrawing that support as income rises can limit public spending or direct resources toward lower-income families. But the withdrawal also lowers the reward from earning the next euro in the range where it occurs. A person need not know which policy objective produced each line to experience the combined result. From the point of view of an hours decision, the lines meet in one number: the additional disposable income left after the increment.
It is also important not to mistake a high effective marginal rate for a high tax on all income. An 80–90% marginal rate in a phase-out range means that the next increment is mostly offset by tax and reduced benefits at that point. It does not mean that the household loses 80–90% of every euro it has already earned. The rate can fall again when a credit or allowance has fully phased out. That distinction is central to both public interpretation and personal decisions: a household may have a tolerable average burden and still see a particular extra shift as offering little net gain.
Four numbers that answer different questions
The first number is the statutory tax rate. It describes the rate written into the income-tax rules for a relevant band or category. It is an important legal and administrative fact, but it is not automatically the share of an additional €100 that a person loses. Credits and allowances can offset it, and their withdrawal can add to it.
The second is an average tax burden. This asks how much tax or transfer reduction applies across a person’s income as a whole. Someone may have paid a modest average share over the year while facing a much higher rate on the next increment. The average describes what has happened to the income already earned; the marginal rate describes what happens if income rises from here. A work decision is usually made at the margin.
The third is the effective marginal tax rate used in the IZA study. It is household- and income-specific. It asks how much of an additional increment in gross income fails to become additional disposable income after all relevant taxes, credits, and allowances are taken into account. The same statutory bracket can therefore coexist with different effective marginal rates for households whose credits, benefits, or circumstances differ. Even within one household, the effective rate can change as income moves through a phase-in or phase-out range.
The fourth is the OECD tax wedge. The OECD’s 2026 release defines it as total labor taxes paid by employees and employers, minus cash benefits received by working families, expressed as a share of labor costs. That measure is useful for comparing the overall tax burden on labor for a specified household type and wage point. It includes the employer side of labor taxation and is not a direct answer to the question “How much of my next €100 will I keep?”
The distinction becomes visible in the release’s cross-country figures. For a single worker earning the average wage, the OECD reports an average tax wedge of 35.1% of labor costs in 2025, with the measure rising in 24 of 38 countries. For a single-earner household with two children at the average wage, the average wedge was 26.2%, reflecting the role of cash benefits and household circumstances. These are meaningful comparisons, but neither figure can be inserted into the Dutch participant’s individual €100 calculation without changing the question being answered.
This is a difference in unit, not merely in terminology. The tax wedge is an average share of labor costs for a representative household type. The effective marginal rate is an incremental change in disposable income for a particular household situation. A country can have a moderate average wedge while some households face steep marginal withdrawal rates. Conversely, a high average burden does not by itself tell us whether a particular person believes an extra hour will pay enough to be worthwhile.
The distinction also matters for policy interpretation. If a government sees a weak labor-supply response and looks only at the statutory rate, it may miss a credit phase-out that makes an extra hour less rewarding. If it looks only at the average tax wedge, it may miss that the problem is concentrated in a particular income range or household type. If it assumes that the official effective marginal rate is already understood, it may attribute a weak response to preferences or laziness when part of the response reflects an incorrect calculation.
The Dutch study is not claiming that people need to reproduce the entire tax code in their heads. It is testing whether a compact explanation of the relevant ranges can correct a consequential misunderstanding. The treatment later used the same incremental logic: people were told approximate net gains from an extra €100 for broad monthly-income groups. The information was intentionally coarse, because an exact personalized calculation would require more details than a short survey can supply. Even a range, however, can reveal that a person’s assumed marginal loss is far above or below what the schedule actually implies.
That leads to a final caution about the phrase “the tax rate.” There is no single rate that fully describes the economic position of every Dutch worker. A statutory rate belongs to a rule. An average burden summarizes accumulated income. An effective marginal rate describes the next increment for a particular situation. A tax wedge compares labor costs and benefits at a broader household level. Treating these as interchangeable can make a policy debate sound precise while directing attention to the wrong margin.
For the experiment, the decisive comparison is between the actual effective marginal rate calculated from respondents’ income and household information and the rate they believed applied to them. The next question is therefore about perception, not further tax arithmetic: how far apart were those two numbers, and did the gap follow a systematic pattern across income groups? Once the hidden slope has been defined, “tax confusion” becomes a measurable error about a specific return—and a possible explanation for a different work decision.