Tax
in a Hurry

The Whole Thing in One Page

Tax arrives disguised as arithmetic. A rate is printed beside an income band, a receipt adds a percentage, a company reports a charge, an estate receives a bill. It looks as though government chooses a number and somebody pays it. That image is wrong in the way that matters most. The person named by law may collect the money from somebody else, alter a price, accept a lower return, move an activity, change its legal form or make the taxable amount disappear without making the underlying economic activity disappear.

To understand any tax, begin with the base. Governments do not tax wealth, work or consumption in the abstract. They tax definitions: taxable earnings in a stated period, the value added by registered firms, profits after permitted deductions, gains realised under specified rules, land assessed on a chosen date. Every definition includes one thing and excludes another. Every threshold creates two sides. Every date creates a before and an after.

Then separate statutory liability and remittance from economic incidence. An employer may send a payroll contribution, but wages and hiring can adjust. A shop may remit consumption tax, while customers, owners, workers or suppliers bear parts of the cost. A company is legally liable for corporation tax, and the immediate charge reduces its after-tax resources. The ultimate burden reaches people through lower shareholder returns, lower wages or employment, higher prices, or reduced land and business values. Who bears how much depends on alternatives, bargaining power, competition, mobility and time.

Rates come next, and there are several. A marginal rate applies to the next pound. An average rate divides the total bill by total income. An effective rate reflects the base after allowances, deductions and credits. The tax wedge compares what labour costs an employer with what reaches a worker. A headline rate without its denominator is an invitation to be misled.

No tax stands alone. Income tax interacts with payroll charges, benefits, consumption taxes, property taxes and the timing of saving. A tax can look progressive by annual income and different by lifetime resources. A relief can help a sympathetic group while rewarding richer members of that group most. Fairness depends on the whole system and on what comparison you choose.

People respond. They work, spend, save and invest differently; they shift income across years; they relabel labour as capital; they move profits across borders; they comply, make mistakes, avoid or evade. Administration therefore belongs inside the tax, not after it. Withholding, invoices, bank reports, payroll records and audits determine which elegant rules collect money in the world.

This is why tax is so complicated. Governments want revenue, fairness, economic growth, family support, regional policy, environmental change and political peace from the same machinery. Each exception solves one complaint and draws another boundary. Each anti-avoidance rule guards a boundary and lengthens the code. Complexity is the accumulated record of choices about who should count, what should count and who can prove it.

The permanent question is not, “What is the tax rate?” It is: what is being taxed, who sends the money, who ends up worse off, what changes in response, and can the state measure it?

That is the book.

Why You Should Care

In a Californian grocery store, researchers changed one detail. Selected shelf labels showed the price including sales tax rather than leaving the tax to appear at the till. Nothing else about the products changed. Demand for the labelled goods fell by 8 per cent relative to comparison products. The customers had always faced the same final price. They responded differently when the tax became visible before they reached the checkout.

That experiment contains the subject in miniature. Tax is money, but it is also timing, presentation and information. The same charge can feel different depending on where it appears. A deduction before wages reach a bank account is experienced differently from an annual bill. A levy folded into a pump price is noticed differently from one itemised on a receipt. A contribution labelled “employer” can be politically invisible even when it affects the cost of employing somebody. The design changes behaviour before anyone begins a speech about fairness.

You should care because tax is one of the largest systems acting on your life and one of the least accurately described. It shapes the gap between what work costs and what workers receive. It changes the price of alcohol, fuel, houses and inheritances. It affects whether income appears as salary, dividends, rent, interest or a capital gain. It helps determine which businesses remain small, which investments cross borders and which transactions acquire an accountant. Across the OECD, social contributions, personal income taxes and consumption taxes together account for most tax revenue. Yet public argument keeps shrinking that system to one headline rate.

The first benefit of understanding it is defensive. You stop making category errors. You no longer assume that the firm writing the cheque bears the cost, that moving into a higher bracket reduces take-home pay, or that a tax described as progressive must make the full system more progressive. You ask for the base, the denominator and the time period. Many confident tax claims fail before the political disagreement has even begun.

The second benefit is practical without becoming personal tax advice. You can read a payslip, company account, Budget announcement or campaign promise with the right questions. A new relief may narrow the base. A frozen threshold may raise effective rates without a rate rise. A tax on a mobile base may produce more reclassification than revenue. A simple-looking exemption may require pages defining who qualifies. The machinery becomes visible.

The third is economic. Tax changes do not stop at the Treasury account. They alter relative prices, returns and risks, then travel through wages, investment, rents and consumption. Sometimes the intended target bears much of the burden. Sometimes the burden moves. Knowing that movement exists does not decide against taxation. It tells you what a serious decision has to measure.

The fourth is political. Tax arguments are arguments about membership. Who owes what to whom? Which differences between people count as relevant? Should two households with the same annual income but different needs pay the same? Should inherited resources be treated differently from earnings? How much should present taxpayers finance for future citizens? There is no technical formula that settles those choices. Economics can expose costs, trade-offs and likely responses. It cannot choose the society.

There are limits. This book will not tell you the right overall level of tax, produce a universal ideal system or replace jurisdiction-specific professional advice. Rules change, countries differ and the legal detail can turn on one word. What it can give you is the model beneath those differences.

Once you can follow the base, the burden, the response and the evidence trail, tax stops looking like a wall of rates. It becomes a machine whose choices can be examined.

The Core Ideas

A Tax Needs a Base

A government cannot tax “the rich”, “business” or “consumption” until law turns the label into something measurable. The measurable thing is the tax base. It may be a flow during a period, such as wages or company profit; a stock measured on a date, such as land value; a transaction, such as a purchase or property transfer; or a physical quantity, such as litres of fuel. The rate receives the publicity. The base does most of the intellectual work.

Consider income. A person receives salary, employer pension contributions, interest, dividends, rent, a gift and an increase in the value of a house. Which of those is income? Economics can propose broad concepts, but a tax authority must specify when each amount arises, what expenses may be deducted, how losses carry across years, how inflation is treated and whether an unrealised gain counts before an asset is sold. The answer is a legal construction built for collection. It is not a fact waiting in nature.

The same problem appears in a company. Its accounts report profit under financial-reporting rules designed to inform owners and creditors. Taxable profit begins with those accounts in many systems, then changes them. Some expenses are disallowed. Capital spending may be deducted over a schedule rather than at once. Losses may be carried forwards or backwards under conditions. Interest, research, royalties and transactions with related companies receive special treatment. A corporation-tax rate tells you almost nothing until you know which profit it applies to.

Every base creates a boundary. Earnings below a threshold sit outside while earnings above it sit inside. Food may be exempt from a consumption tax while restaurant meals are taxable. A repair may be deductible now while an improvement is capitalised. A worker may be an employee under one set of facts and self-employed under another. These distinctions can pursue sensible aims. A threshold may protect low incomes or spare tiny firms heavy compliance costs. An exemption may reflect distributional concerns. The difficulty is that conduct changes around the line.

Measurement adds a quieter choice. Market value may be current but volatile; historic cost is verifiable but can become detached from economic gain. Cash accounting follows money received and paid; accrual accounting recognises rights and obligations earlier. A tax on real gains would adjust for inflation, yet doing so requires an index and records across years. Administrative convenience can therefore shape the base as much as theory does.

Base design also chooses between breadth and discrimination. A broad base treats more activities alike, can raise a given sum at a lower rate and usually leaves fewer classification games. A narrower base can target ability to pay or a desired activity more closely. It can also reward people who fit the preferred category, intentionally or through planning. There is no rule that breadth always wins. A tax on pollution should distinguish pollution from everything else. A disability-related relief requires a boundary because the policy has a boundary. Simplicity and targeting pull in opposite directions.

This is the first habit to acquire. Before asking whether a rate is high, ask what enters the base, what leaves it, when the measurement occurs and who supplies the evidence. A debate about percentages before those questions is usually a debate about a number floating free of a tax.

The Collector Is Not Necessarily the Payer

A payslip names the employee. A payroll return names the employer. A receipt names the shop. A corporation-tax account names the company. These documents show legal liability and remittance: who must calculate, report and send money. They do not settle economic incidence, which asks which people end up with less real income or wealth because the tax exists.

Take a tax collected from sellers. If customers can switch easily to untaxed substitutes while sellers have few alternatives, sellers may have to absorb much of it through lower margins. If customers need the product and competing sellers face the same charge, prices may rise and customers may bear more. Suppliers, workers or landlords can enter the chain as contracts renew. The legal direction of the payment does not control the eventual direction of the burden.

The governing idea is responsiveness. Economists use elasticity to describe how strongly supply or demand changes when a price changes. The side with fewer good alternatives tends to bear more of a tax because it cannot escape the changed bargain as easily. That is a mechanism, not a moral judgement. A patient needing urgent medicine may have little ability to reduce demand. A highly mobile financial activity may move jurisdiction. Land itself cannot leave, though owners and uses can change. The relevant alternatives differ across markets and across time.

Time matters because many burdens are sticky before they become mobile. A tax introduced tomorrow may first reduce the profits of firms with fixed prices and contracts. Over several years, prices, wages, investment and location can adjust. A tax on an expected stream of land rent may be capitalised into a lower sale price, making the owner at the moment of announcement bear a large part of the burden even if later annual bills are sent to future owners. “Who pays?” needs a date as well as a person.

Labour taxes show the distinction cleanly. Governments often divide charges into employee and employer components. The employer remits its component in addition to gross wages, but it forms part of the cost of hiring. In a competitive labour market, some of that cost may be reflected over time in wages or employment. The exact split depends on institutions, bargaining and worker and employer responses. Calling one part an employer tax does not make employers a separate species capable of bearing it without consequences for people.

Corporation tax creates the same temptation on a larger scale. The company is the legal taxpayer, and the direct effect is lower after-tax profit and fewer resources available to distribute or reinvest. Economic incidence asks which people ultimately experience the loss. Shareholders may receive lower returns; workers may face lower wages or employment; customers may pay higher prices; owners of less mobile assets may see values or rents adjust. Evidence and theory support all of those channels in some settings. They do not justify a universal percentage assigned to each. Company size, market power, profit type, capital mobility and the tax base all matter.

Incidence analysis does not prove that a tax is undesirable or that its stated target escapes. It stops the government cheque from ending the investigation. The name on the return is the collection point. The economic question begins there.

The Rate Has More Than One Meaning

Suppose a system charges no income tax on the first £20,000, 20 per cent on the next £30,000 and 40 per cent above £50,000. A person earning £60,000 does not pay 40 per cent of £60,000. The first slice is untaxed, the next £30,000 produces £6,000 of tax and the final £10,000 produces £4,000. The total is £10,000. The marginal rate is 40 per cent because that is the rate on the next pound. The average rate is one sixth, because £10,000 is one sixth of £60,000.

This distinction explains why entering a higher bracket does not normally reduce take-home income. The higher rate applies to the slice above the threshold. A genuine cliff is different. If crossing a threshold removes a fixed benefit, allowance or eligibility status, one extra pound can trigger a loss larger than the pound earned. Good analysis separates a kink, where the slope changes, from a notch or cliff, where the amount jumps.

The statutory rate is the percentage written in law. The effective rate relates the tax paid to a chosen economic measure after deductions, credits, exemptions and timing rules. A company facing a 25 per cent statutory rate may have a lower effective rate on one investment because accelerated deductions reduce the present value of tax. Another investment may face a higher burden because some costs are not deductible. Two people with equal cash receipts may face different effective rates because the receipts have different legal labels.

Then there is the tax wedge. For labour, it compares the employer’s total cost with the worker’s net take-home pay, usually including personal income tax and employee and employer social contributions, and sometimes cash benefits depending on the measure. The OECD calculated an average wedge of 35.1 per cent in 2025 for a single worker without children earning the average wage across its members. That is a useful comparison for one standardised household type. It is not the share of everything the worker gives government, because it excludes consumption taxes, many other taxes, state services and other income.

Effective marginal rates can also include benefit withdrawal. A worker may face income tax, payroll contributions and the loss of a means-tested payment on the same next pound. None of the individual rates looks extreme, but the combined withdrawal can be high. The reverse can happen when a refundable credit rises with earnings. Tax and benefits are two directions in the same cash-flow system even when administered by different departments.

Inflation changes rates without legislation. If incomes and prices rise while nominal thresholds remain fixed, more income moves into higher bands and the average burden rises. This is fiscal drag, often called bracket creep. A government can therefore raise revenue while announcing no change to headline rates. Indexing thresholds limits that effect, but then the index, timing and treatment of real income growth need rules of their own.

A rate is always a ratio. Ask which tax is in the numerator, which base or income concept is in the denominator, whether the figure is marginal or average, whether benefits are included and which period is measured. Most public arguments use the word “rate” as though it names one object. It names a family.

Taxes Work as a System

A single tax can be progressive while the combined system is less so. Another can look regressive in isolation while financing transfers that make the package progressive. The distributional question is not answered by staring at one schedule. It requires the taxes, benefits, household resources and time period that belong in the comparison.

Progressive usually means that tax rises as a share of the chosen measure of ability to pay. Proportional means the share stays constant. Regressive means the share falls as the measure rises. Those definitions sound clean until the denominator moves. A consumption tax often takes a larger share of annual income from low-income households because they spend a high fraction of current income. Rank households by current expenditure, which is often smoother than current income but is not a direct measure of lifetime resources, and the pattern can look different. Neither view is automatically correct. A student, a temporarily unemployed worker and a permanently poor household can report the same annual income while occupying different economic positions.

Households complicate the unit. Should tax depend on individual income or combined family resources? Individual assessment avoids making one partner’s rate depend directly on the other’s earnings and can support work incentives for second earners. Household assessment recognises shared living standards and economies of scale. Children, disability, caring duties and housing costs affect needs but are hard to capture without creating more categories and withdrawal rules. Equal income does not guarantee equal capacity, yet adjusting for every difference would make the return resemble a biography.

Fairness has at least two directions. Vertical equity asks how burdens should differ between people with different resources. Horizontal equity asks whether people in similar positions are treated similarly. Tax systems often violate the second while trying to pursue another aim. Two people earning the same amount through salary and dividends may pay different sums. Two owner-occupiers with equal housing consumption may face different bills because property valuations are old. A relief for pension saving may reward the same saving differently depending on the taxpayer’s marginal rate.

Tax expenditures sit inside this system. An exemption, deduction, credit, reduced rate or deferral can pursue the same aim as a spending programme. A childcare credit and a childcare grant send support through different administrative pipes. Calling the first a tax cut can hide its cost and distribution, while calling every departure from an imagined benchmark a subsidy can overstate certainty. Measuring a tax expenditure requires choosing what the normal tax base would have been, and that benchmark is contestable.

Public services make the comparison wider still. A household can pay more tax and receive more in health care, schooling, insurance against unemployment or public infrastructure. Valuing those services by what government spends is not the same as valuing them by what each household would have paid, and some benefits are collective rather than divisible. A claim about net redistribution should say whether services are included and how they are valued.

The mix matters for resilience and behaviour as well as fairness. Income taxes respond strongly to employment and profits. Consumption taxes follow spending. Recurrent property taxes rest on a base that is difficult to move but politically visible. Resource taxes can capture location-specific rents yet fluctuate with prices. Social contributions may be linked to entitlements or function much like other labour taxes. Different countries assemble different portfolios because their institutions, histories and administrative capacities differ.

A serious distributional statement therefore names the package, unit and period. It asks what happens to disposable resources after taxes and cash benefits, and sometimes after the value of public services. The result may still leave room for moral disagreement. At least the disagreement will be about the system people live under rather than one attractive or infuriating line within it.

Behaviour Changes the Base

Taxpayers are not numbers waiting to be multiplied. They can change how much they work, what they buy, when they sell, how a business is financed, where an activity occurs and what label it receives. A forecast that applies a new rate to yesterday’s base assumes away the main reason tax policy is difficult.

Some responses change real activity. A higher tax on cigarettes can reduce smoking. A charge on congestion can alter travel time or mode. A tax on earnings can affect hours, participation, effort, training or migration, though the size varies across groups and institutions. A deduction for investment can bring spending forward. These responses may be the policy’s purpose, an unwanted cost or both. Revenue is not the only outcome.

Other responses change timing or form more than substance. An owner can delay realising a capital gain. A company can bring a deductible expense into one year and income into another. A professional may receive income through a company rather than salary where the tax treatment differs. A multinational group can adjust internal financing or prices within legal limits. The economic activity may remain while the reported tax base moves.

Visibility matters. In the grocery experiment, showing tax-inclusive shelf prices reduced demand by 8 per cent even though the final price was unchanged. Consumers underreacted when the tax appeared only at the till. This does not mean hidden taxes are painless. It means presentation changes attention, and attention changes behaviour. Withholding has a similar psychological feature: it improves collection and smooths payment, but many people focus on net pay rather than the full labour cost or annual liability.

Tax returns reveal response at boundaries. Emmanuel Saez found clear bunching around the first kink in the United States Earned Income Tax Credit schedule, concentrated among self-employed filers. The finding is informative precisely because it is limited. People with control over reported income responded more visibly than wage earners whose income was reported by employers. A sharp line in law can leave a mark in data, but the mark depends on the ability to adjust.

Avoidance and evasion belong on different sides of legality, though the practical border can be disputed. Evasion conceals or falsifies facts against the law. Avoidance arranges affairs within legal form to reduce tax, sometimes in ways legislators expected and sometimes against the apparent purpose of the rule. Ordinary use of an approved pension relief and an elaborate transaction with no commercial purpose can both be lawful yet raise different policy questions. Mistakes and inability to comply form further categories. Treating every shortfall as deliberate cheating produces bad diagnosis.

The revenue response therefore depends on the starting rate, breadth of the base, available alternatives, enforcement and time. At a zero rate, revenue is zero. At some extreme rate, the base may shrink dramatically. That observation does not show that any proposed tax cut will raise revenue. The location and shape of the revenue curve are empirical questions and differ by tax and setting.

Behavioural response is neither a trump card against taxation nor a nuisance to ignore. It is part of what the policy does. The right question is which margin changes, by how much, for whom, and whether the change is a cost, a benefit or a relabelling exercise.

Administration Is Part of the Tax

A beautifully designed liability that cannot be observed is a thought experiment. Real tax systems need records, identifiers, payment dates, filing rules, matching, audit, penalties, appeals and people or software capable of operating them. Administration does not sit beneath policy as clerical detail. It determines which policies exist outside a statute book.

Withholding is one of the decisive inventions. An employer calculates and sends tax as wages are paid rather than leaving every worker to save for a yearly bill. The system collects from a relatively small number of organised payers, uses payroll records and spreads payment across the year. Britain adopted this collection method in 1944 under Pay As You Earn; variants now make wage taxation feel automatic across many countries. The worker still has a liability, but the employer becomes an information source and collection agent.

Third-party reporting changes compliance because the taxpayer is no longer the only witness. In a Danish field experiment involving more than 40,000 individual filers, evasion was close to zero for income reported by third parties and substantial for self-reported income. Audit threats affected self-reported income but not the income already visible through third-party records. Denmark has unusually strong institutions and extensive reporting, so the magnitude should not be exported everywhere. The mechanism is hard to miss: verifiable information narrows the room for concealment.

Value added tax builds information into transactions between firms. A registered business charges tax on sales and claims credit for tax paid on inputs. To claim the credit, it wants an invoice from its supplier. In Chile, two randomised experiments covering more than 400,000 firms found that the paper trail created preventive deterrence and enforcement effects along the VAT chain. VAT and a retail sales tax can look equivalent in a frictionless textbook. They are not administratively equivalent when each stage creates evidence about another taxpayer.

Collection design also distributes burdens among compliant taxpayers. A small firm may face bookkeeping costs out of proportion to the revenue at stake. A threshold can spare it those costs, but then firms near the threshold may limit sales, split activity or stay informal. Pre-populated returns can remove mistakes, but only if the administration receives timely and accurate data. Digital records can make compliance easier for some and difficult for people without access, skills or stable systems.

Enforcement requires selection and restraint. Auditing everyone would be ruinous and intrusive. Auditing nobody would invite non-compliance. Authorities use risk rules, third-party matches and increasingly automated tools to choose cases. Errors in those systems can concentrate scrutiny unfairly or make opaque decisions hard to challenge. Taxpayer rights, reasons, independent review and appeal are therefore part of effective collection, not luxuries added after revenue is secured.

Tax gaps illustrate the uncertainty. HM Revenue and Customs estimated the United Kingdom’s 2024 to 2025 gap at 6.4 per cent of theoretical liabilities, or £59.2 billion. “Theoretical” matters. The number is modelled from audits, surveys, administrative data and assumptions; it is not a count of banknotes known to be missing. Other countries use different methods, so rankings can mislead. A tax gap is a management estimate, not a moral census.

The administrative test is blunt and useful. Who knows the fact being taxed? Who reports it? Can another record verify it? What does compliance cost? What happens when the record is wrong? Those questions can overturn a policy that looks ideal on a whiteboard.

Every Boundary Creates a Game

Return to the base. Law must distinguish salary from dividends, an employee from a contractor, interest from profit, debt from equity, a repair from an improvement, residence from source, a gift from payment and business consumption from private consumption. Those distinctions are unavoidable because different rules cannot operate without categories. They are also invitations.

The invitation need not involve dishonesty. Suppose labour income is taxed more heavily than returns received through an owner-managed company. A person who can choose the form of work has a reason to incorporate. If debt interest is deductible while the cost of equity is not, a company has a tax reason to borrow. If capital gains are taxed only on sale, an investor has a reason to hold an appreciated asset. If a relief ends at a threshold, a firm has a reason to remain below it. The tax system becomes one input into the design of the transaction.

Governments respond by refining the line. Employment-status rules examine control, substitution and economic dependence. Anti-avoidance provisions can look through arrangements lacking commercial substance. Thin-capitalisation or interest-limitation rules restrain debt loading. Transfer-pricing rules ask related companies to use terms resembling those between independent parties. Controlled-foreign-company rules attribute some offshore income back to owners. Each response protects revenue or neutrality. Each requires definitions, exceptions, evidence and procedures.

Reliefs add another layer. Some recognise costs needed to measure the base properly. A business must usually deduct genuine operating expenses before profit means anything. Others pursue social or economic aims: research, pensions, home ownership, charities, regional investment or family support. Once a relief exists, removing it creates visible losers while its cost is dispersed across the wider base or higher rates. The rule acquires defenders, advisers and transactions organised around it.

History remains in the machinery. New taxes are often built beside old ones because replacing the old system would create abrupt gains, losses and administrative risk. Temporary provisions survive. Different levels of government retain different bases. Contributions keep separate names because they once financed specific benefits. Property valuations remain frozen because updating them would reveal changes in liability. Complexity is partly the fossil record of compromises that were politically easier to add than to unwind.

International taxation multiplies the boundary problem. Countries divide taxing rights by residence, source and destination. Multinational groups operate as one business but file through many legal entities. Internal prices, loans, licences and services allocate profit among jurisdictions. Treaties prevent some double taxation and create common concepts, yet treaty networks add another legal layer. The OECD’s Pillar Two rules set out a coordinated 15 per cent minimum effective-tax framework for large multinational groups through jurisdictional calculations and top-up taxes. The effort to place a floor under tax competition requires a formidable information and rule system because “15 per cent of profit in a country” contains several definitions inside every word.

Politics can also prefer opacity. A visible rate rise attracts attention; a frozen threshold, narrowed deduction or new surcharge may raise money less dramatically. Labelling a charge as a contribution can preserve the appearance of a link to benefits even when the revenue enters a broad pool. Complexity can hide distribution, but concealment is only one source among several.

None of this proves that a short tax code is impossible or that every complication is defensible. Poor drafting, obsolete reliefs, duplicated taxes and political concealment produce needless burdens. Coherent reform can align bases, remove cliffs, integrate taxes and benefits, update valuations and replace narrow favours with direct spending. But simplicity is not achieved by deleting detail while keeping every objective.

The loop is now complete. Tax needs a boundary to exist. The boundary changes behaviour. Government adds rules to preserve fairness, revenue or the policy aim. Those rules draw fresh boundaries. Complexity is the armour around the base and the record of every promise made through it.

How It Actually Works

Before the first pound

Before money changes hands, the tax system has already made several decisions. It has defined the taxpayer, the accounting period, the jurisdiction and the records that will count as evidence. A human being may be taxed as an individual, a household member, a sole trader, a partner, a trust beneficiary or the owner of a company. The same economic activity can pass through different legal persons and meet different rules.

The calendar matters. Governments usually divide continuing life into tax years or accounting periods. Wages earned on 31 December may belong to one period while wages earned a day later belong to another. Businesses need rules for invoices raised but not paid, stock held at year-end, bad debts and contracts spanning several years. The period creates a stopping point at which an unfinished economic story becomes a number.

Jurisdiction supplies another line. Residence often gives a country a claim over a person’s broad income, while source rules give a country a claim over income arising within it. Citizenship matters in a few systems. Property is usually taxed where it sits. Consumption is increasingly assigned to the destination where it occurs. None of these connecting rules is self-executing. They require tests for days present, homes available, management exercised, customers located or work performed.

States with limited records often favoured bases that could be seen: land, windows, hearths, goods crossing ports, salt leaving official stores. Modern income and profit taxes became practical as governments, employers, banks and firms accumulated records. Tax capacity grew with the information system. The state did not become able to tax income merely by deciding that income was fairer than a customs duty. It became able to tax income when employers, accounts and administrative identifiers made millions of separate receipts legible enough to assess.

Earn

An employer agrees a gross salary, but the employment bargain has several layers. The worker sees gross pay, deductions and net pay. The employer sees salary, employer social contributions, pension costs and other employment expenses. Government sees a stream that can be measured each pay period and collected through payroll.

Withholding turns the employer into a collection agent. The payroll system applies a code or schedule, subtracts tax and employee contributions, adds any employer charge, reports the figures and sends money to the authority. Britain introduced Pay As You Earn in 1944, after wartime expansion had brought millions more workers into income tax. Weekly or monthly deduction allowed tax to be collected more efficiently than annual or twice-yearly payment and aligned payment more closely with wages. The tax remained the worker’s liability, while the employer became the recurring information source and remitter.

The labels can obscure the full burden. Suppose an employer’s total cost of employing someone is 120 units. The worker’s contractual gross wage is 100, employee deductions are 25 and employer contributions are 20. Net pay is 75. The personal average rate measured against gross wage is 25 per cent. The tax wedge measured between employer cost and net pay is 37.5 per cent. Neither figure is fraudulent. They answer different questions.

Benefits in kind and expenses complicate the base. A company car, health cover, meals, travel or share awards may substitute for cash. If cash is taxed and equivalent benefits are not, remuneration moves into benefits. Rules then distinguish business use from private consumption, genuine reimbursement from pay and a widely available staff facility from a personal perk. Each distinction exists because compensation is more inventive than a wage line.

Self-employment breaks the payroll chain. The person receives gross business receipts, deducts permitted costs and reports profit. There may be no employer supplying an independent figure. Payment can arrive irregularly, expenses can have mixed purposes and the boundary between labour return and return on capital becomes less obvious. Many systems use advance payments, platform reports, presumptive rules or simplified expenses to collect without pretending that a sole trader is a monthly employee.

Buy

At a shop, consumption tax may appear as an extra line or sit inside the displayed price. A retail sales tax is collected at the final sale. Value added tax is collected in stages, with registered firms charging tax on sales and claiming credit for tax on business inputs.

Imagine a timber supplier sells wood to a furniture maker for 100 plus 20 of VAT. The supplier sends 20 to the state. The maker turns the wood into a table and sells it to a retailer for 300 plus 60. The maker owes 60 on the sale but claims credit for the 20 already paid, so sends 40. The retailer sells to a household for 500 plus 100, claims the 60 input credit and sends 40. Total revenue is 100, equal to 20 per cent of the final pre-tax price. Each business has remitted tax on the value it added.

The invoice is doing two essential administrative jobs. It calculates the credit and creates a record linking buyer and seller. A firm claiming an input credit wants evidence that its supplier charged tax. That self-enforcing feature is incomplete because fraud can exploit false invoices, missing traders and refunds, but it gives VAT an information structure that a single final-stage tax lacks.

Exemptions disturb the chain. An exempt business may charge no VAT but also be unable to recover tax on inputs. Its unrecovered input tax becomes a cost that can affect prices or production choices. A zero-rated sale differs: the rate on the output is zero while input credits remain available. The consumer may see the same zero charge, but the business and Treasury see different mechanics.

Excises target particular goods or harms. Fuel, tobacco and alcohol are often taxed by physical quantity or product category as well as through general consumption tax. A quantity-based duty does not rise automatically with price and may erode in real terms unless adjusted. A percentage duty moves with price. Product boundaries invite redesign: strength, ingredients, emissions or packaging can determine the category. Where the aim is to change behaviour, a fall in the base may be success even while revenue falls.

Customs duties sit at the border and depend on origin, classification and value. A shoe, component or food preparation needs a commodity code. Trade agreements ask where a product counts as originating. A tariff may be legally paid by the importer, reflected in consumer prices, absorbed by margins or transmitted to foreign suppliers. Once again, the customs account records remittance, not the final burden.

Run a company

A company receives revenue and incurs costs, but taxable profit is not cash in the bank. Customers may pay late. Equipment bought today may serve for years. Inventory becomes a cost when sold. A loan brings cash without creating profit, while repaying principal uses cash without creating an ordinary deduction. Accounts organise these differences before tax rules adjust them.

Depreciation illustrates the split. Financial accounts estimate how an asset’s value or usefulness is consumed. Tax law may replace that charge with capital allowances set by policy. Immediate deduction makes an investment cheaper in present-value terms than a deduction spread over many years. A special allowance can stimulate spending or compensate for inflation; it can also favour qualifying assets over other costs. The statutory corporation-tax rate remains unchanged while the effective tax on an investment moves.

Losses raise the question of symmetry. If government takes a share of profit, should it share a loss at once? Most systems do not send every loss-making company a full refund. They permit losses to offset profits in other periods or within groups, subject to limits. A young firm with years of losses may therefore receive less present value from deductions than an established profitable firm making the same investment. The timing of tax changes risk.

Finance draws another boundary. Interest on debt is commonly deductible in measuring profit while dividends paid to shareholders are not. That can encourage borrowing relative to equity and allows multinational groups to place debt where deductions are valuable. Interest-limitation rules respond by restricting excessive deductions. The response protects the base but creates tests for group ratios, earnings measures and genuine third-party finance.

Owners can receive value as salary, dividends, retained profit, loans, benefits or capital gains. Different combined rates on company profit and personal receipts make the choice consequential. Rules for closely held companies try to keep similar work from facing radically different burdens because it passed through a company. The design must still distinguish entrepreneurial return, invested capital and labour, which do not arrive in separate envelopes.

Corporation tax is paid from the company account. In the short run it can reduce after-tax profit. Over time it can affect investment, location, wages, prices and asset values. Taxing location-specific economic rent is different from taxing the normal return required to attract mobile capital. A mine cannot move its ore body; an intellectual-property licence can be assigned elsewhere with less physical disturbance. One corporate rate covers unlike kinds of profit unless the base separates them.

Own land and property

Property gives governments a visible, immobile base, then makes them politically cautious. Recurrent taxes may use market value, rental value, area, bands or old assessments. Current valuation tracks differences more accurately but creates changing bills and disputes. Frozen valuation avoids annual upheaval while gradually severing tax from present value.

A tax on pure land value has an unusual property: the supply of land in a location cannot shrink because of the tax. That makes the base hard to move and, in the pure case, does not reduce the physical supply of land. Buildings are different. Taxing improvements can discourage construction or maintenance at the margin. Separating land from structures is economically attractive and administratively demanding because their values are observed together.

Transaction taxes collect when property changes hands. The event is easy to identify and payment can be tied to registration. The cost is a wedge against moving, downsizing or reallocating property. A household can remain in the wrong home rather than trigger a charge. Recurrent taxes are harder to sell politically because they arrive repeatedly and may create cash-flow problems for owners with valuable property but low current income. Deferral can address cash flow and introduces interest, eligibility and recovery rules.

Owner-occupied housing mixes consumption and investment. Rent paid to a landlord is visible. The housing service an owner receives from living in their own property has no cash transaction. Capital gains may be taxed, exempt or deferred; mortgage interest may be deductible or ignored; local charges may bear little relation to current value. The result often reflects decades of housing policy rather than one coherent tax base.

Save, invest and transfer

Saving moves consumption through time, so its taxation can occur at several points. Income can be taxed before it is saved, returns can be taxed as they arise, and withdrawals can be taxed later. Pension systems often exempt contributions and investment returns, then tax withdrawals; other accounts tax contributions but exempt later returns. With equal tax rates at contribution and withdrawal, and no other differences, taxing on entry rather than exit can produce the same after-tax return. Once rates, limits, matching contributions, benefit tests or behaviour differ, the equivalence breaks.

Interest is usually easier to observe when a regulated financial institution reports it. Dividends connect company profit to shareholders and may receive credits, exemptions or lower rates intended to recognise prior corporation tax or pursue other aims. The exact relationship differs by jurisdiction, so “double taxation of dividends” describes a structure, not a complete verdict.

Capital gains create the realisation problem. An asset can rise in value without producing cash. Taxing the gain each year would reduce deferral and treat retained and sold assets more alike, but valuation and liquidity become difficult for private businesses, art or property. Taxing only when sold uses an observable event and gives owners control over timing. That control produces lock-in: selling triggers tax, so an owner may keep an asset they would otherwise exchange.

An annual net wealth tax tries to measure assets minus debts at regular intervals. Bank accounts and listed securities are comparatively visible; private companies, trusts, art, pensions and homes raise valuation, ownership and liquidity problems. Exemptions change both distribution and planning. A broad low-rate base can differ sharply from a narrow high-rate charge filled with exclusions. The legal taxpayer is the owner, while an anticipated recurring charge can also be reflected in asset prices. Mobility differs: land cannot leave, while a portfolio owner can sometimes change residence or legal form.

The calculation needs a basis, normally linked to acquisition cost plus permitted adjustments. Inflation can make a nominal gain exceed the real gain. Death, gifts, emigration and transfers between spouses or entities require rules for whether the gain is realised, carried over, rebased or forgiven. Every answer changes incentives around the event.

Taxes on gifts and estates target transfers of wealth, the recipient’s acquisition, or the estate before distribution. They need valuation, relationship rules, exemptions, treatment of businesses and farms, and a period for gifts made before death. Without aggregation, a large transfer could be split into many small ones. With aggregation, the administration needs records across years. The moral argument concerns inheritance and opportunity; the operating problem concerns tracing value through time and legal forms.

Cross a border

Cross-border tax begins with overlapping claims. The country where a person lives may tax worldwide income. The country where work, property or business activity occurs may tax at source. The country where a customer consumes a service may claim destination-based consumption tax. Without coordination, the same income can be taxed twice. With gaps between rules, it can be taxed nowhere or later than either country intended.

Tax treaties allocate rights, reduce some withholding taxes, define residence and permanent establishment, and provide methods for relieving double taxation. They do not create one world tax system. A treaty is a bridge between two domestic systems, and thousands of bridges do not form a single road code.

Multinational groups intensify the problem because the business is economically integrated and legally divided. One subsidiary manufactures, another owns intellectual property, another lends money and another sells to customers. Transfer-pricing rules ask the entities to price internal dealings as independent parties would. Independent prices may not exist for unique intangibles or integrated risks. Documentation grows because the allocation cannot be observed directly.

Countries also use controlled-foreign-company rules, interest restrictions, anti-hybrid rules and reporting requirements to protect their bases. The OECD/G20 Pillar Two framework adds a coordinated minimum effective rate of 15 per cent for large multinational groups, calculated jurisdiction by jurisdiction, with top-up tax where covered income is taxed below the minimum. By 2026 the OECD had issued consolidated commentary and continuing administrative guidance. The mechanism aims to reduce profit-shifting incentives and place a floor under competition. Its complexity reflects the task: combining accounting data, tax adjustments, entity scope, ownership chains and several countries’ priority rules without charging the same top-up twice.

File, match, audit and appeal

At the end of the chain, the authority assembles information. Employers report wages. Banks and platforms may report payments or accounts. Firms file consumption and corporation-tax returns. Property registries record transfers. Customs systems record imports. The taxpayer confirms, supplements or disputes the picture.

Some systems rely heavily on self-assessment. Others pre-populate returns from third-party data and ask the taxpayer to correct them. Pre-population can reduce effort and error, but it can also turn a wrong official figure into the default. The taxpayer remains the person who knows some facts the databases do not: mixed business expenses, foreign income, household circumstances or eligibility for relief.

Automated matching identifies discrepancies. Risk models rank cases for enquiry. An audit can request records, test valuations and challenge legal treatment. Penalties usually distinguish lateness, carelessness, deliberate conduct and concealment, though categories and safeguards vary. Collection can include payment plans, interest, security or enforcement against assets.

Disagreement is normal because facts and law can both be uncertain. A functioning system needs explanations, time limits, review and an independent route of appeal. Tax certainty is not the absence of every dispute. It is the ability to know the rule in advance, receive reasons and contest the state through a process that does not depend on its goodwill.

How we know

Tax research has an unusual advantage: governments collect detailed administrative records because the system cannot run without them. Researchers can study returns, payroll data, invoices, audits and policy changes, often across millions of observations. Randomised audits in Denmark, VAT enforcement experiments in Chile and the grocery-store salience experiment reveal mechanisms that surveys alone would miss. Bunching around thresholds can show where reported behaviour changes.

The evidence also has hard limits. Taxpayers who vanish from the base can be difficult to observe. Legal and administrative systems differ, so an estimate from Danish wage reporting or Chilean VAT cannot be carried unchanged to another country. Policy changes often coincide with economic shifts, anticipation and other reforms. Incidence unfolds through prices, wages, returns and location over different periods, making one clean percentage rare. Tax-gap estimates combine data with models and should not be mistaken for direct counts. The strongest conclusions identify a mechanism and its setting, then state how far the evidence can travel.

What People Get Wrong

“The person who sends the money pays the tax”

The bank transfer is compelling evidence because it is visible. An employer sends payroll contributions, a retailer sends VAT and a company sends corporation tax, so each appears to bear the charge. The payment record proves remittance. It does not prove economic incidence.

Prices, wages, returns, employment and asset values can adjust. A retailer may pass much of a broad consumption tax into prices, absorb part through margins or press suppliers for lower prices. Employer charges can affect labour costs and, over time, wages or hiring. Corporation tax can reach shareholders, workers, customers or landowners through several channels. The split depends on alternatives, bargaining and time, not the heading on a return. A landlord may receive a property-tax bill yet be unable to raise rent in a weak market; the burden then falls on the owner. Where supply and demand permit, rent may rise when leases reset. If the tax was anticipated before purchase, part may already be embedded in the price paid for the property. The same legal bill can travel differently across place and time.

This correction matters because politics can move a legal payment without moving the economic burden. Renaming an employee charge as an employer charge may change administration and visibility while leaving the labour wedge close to where it was. Follow the adjustment after the cheque.

“A higher bracket taxes all your income at the higher rate”

The mistake survives because tax tables print one rate beside one income range. Readers treat the range as a label for the whole person. In a normal graduated schedule, each rate applies only to the slice within its band. Crossing a threshold leaves earlier slices under their earlier rates. If the first £30,000 is taxed at 20 per cent and the next slice at 40 per cent, earning £30,001 adds 40 pence of tax on the final pound, not a 40 per cent charge on all £30,001. Take-home pay still rises by 60 pence before any other deductions or withdrawals.

A true cliff can still exist. A benefit, allowance or legal status may disappear when income crosses a line. Then one extra pound can trigger a larger loss. Effective marginal rates can also become high when income tax, social contributions and benefit withdrawal act together. Those are real problems, but they are not how ordinary brackets work.

Confusing a marginal rate with an average rate makes people misread pay rises, compare countries badly and overlook the sharper problem of cliffs. Calculate the total bill, then divide by the relevant income if you want the average. Look at the next pound if you want the marginal rate.

“Corporation tax is borne only by companies”

Companies owe and remit corporation tax. That is correct in law. The mistake is treating the company account as the end of the economic story. The immediate accounting effect is a smaller pool of post-tax profit for distribution or reinvestment. Economic incidence asks which people ultimately lose real income or wealth as markets adjust prices, pay, jobs, returns and asset values.

The initial effect commonly falls on shareholders through lower after-tax returns. Longer-run effects can include lower investment, wages or employment, higher prices, and changes in land or business values. The burden differs between location-specific economic rent and returns that must remain high enough to attract mobile investment. Research supports several channels and does not produce one universal division.

The correction matters in both directions. “Workers pay it all” is no more reliable than “shareholders pay it all”. A corporate tax can reach foreign owners, capture economic rent and act as a backstop to personal taxation. It can also discourage investment or shift reported profit. Judge the base, profit type, market and time horizon rather than treating the company logo as the final bearer.

“Consumption taxes are always regressive”

Measured against annual income, broad consumption taxes often take a larger share from low-income households because those households spend more of current income. That is a legitimate result, not the whole result.

Annual income can be temporarily low while spending is financed by savings, borrowing or family support. When households are ranked by current expenditure, the pattern can change, partly because spending is often smoother than annual income. That does not turn expenditure into a direct measure of lifetime resources. Exemptions and reduced rates matter, as do cash benefits and direct taxes in the same package. Reduced rates on necessities can help poorer households as a share of income, yet richer households may receive more cash benefit because they spend more in total. Direct compensation can be better targeted, provided the benefit system reaches the intended people and does not create damaging withdrawal rates. A uniform VAT combined with targeted transfers can distribute differently from a VAT full of reduced rates that also subsidise high-spending households.

None of this makes every consumption-tax rise fair. It makes “regressive” depend on the denominator, time horizon and compensation. State which measure you used. Otherwise one word is carrying a distributional analysis it has not performed.

“Taxing the same money twice is automatically wrong”

Money has no memory. The same pound can be involved in several distinct taxable events: earned as wages, spent on a purchase, received as profit, distributed as a dividend, used to buy land and later transferred by inheritance. Calling this double taxation can identify a genuine overlap, but it does not decide whether the overlap is incoherent.

What matters is the tax base and the economic activity. Taxing corporate profit and then taxing a shareholder’s dividend may create a combined burden on one return, which systems can integrate or relieve. Taxing earnings and then consumption taxes two different uses of resources. A property transaction tax layered on recurrent property tax can create a costly wedge even though the labels differ.

The useful question is not how many times a banknote appears. It is whether combined taxes treat comparable choices consistently, create an unintended distortion or exceed the intended burden.

“Lower tax rates always raise more revenue”

At a tax rate of zero, revenue is zero. At a sufficiently extreme rate, avoidance, evasion and collapse of the base could make revenue lower than at a less extreme rate. That is the insight represented by the Laffer curve. It is logically sound and routinely abused.

The curve does not tell you where the current rate sits, how quickly the base responds or whether a cut pays for itself. Those are empirical questions. The answer differs for tobacco, labour earnings, capital gains, company profit and land. A rate cut can broaden reported activity and still reduce revenue because the response is too small. A rate rise can raise revenue while causing a behavioural cost. Revenue maximisation is also not the same as good policy.

Claims that a tax change will fund itself need evidence about the relevant base, starting rate and time period. A sketch of a curve supplies none of them.

“Complexity is caused by incompetent drafting”

Bad drafting exists. So do obsolete rules, political gimmicks and provisions nobody would choose from a blank page. They explain some complexity. They do not explain why mature tax systems tend to keep producing more.

Tax needs definitions. Fairness creates distinctions between low and high incomes, business and private costs, need and preference. Economic policy adds reliefs. Administration adds thresholds and reporting rules. Taxpayers adjust, so anti-avoidance law guards the boundaries. Old systems cannot be replaced without transition rules because assets, contracts and expectations were built under them. Different governments and levels of state retain their own taxes. Political durability adds another force. A relief creates identifiable beneficiaries who notice its removal, while the cost is spread thinly across everyone financing the narrower base. Addition is often easier than subtraction. Complexity accumulates even when each decision had a defensible purpose at the time.

A one-page tax code is easy if it has one base, one rate, no reliefs, no international activity, no hardship cases and no concern for avoidance. The public rarely accepts that system once the bill arrives. The harder correction is that complexity is partly the price of competing objectives. Reform should remove needless differences and align the remaining ones. It cannot keep every promise while deleting every rule.

Use It

Name the base before arguing about the rate

When somebody proposes “a tax on wealth” or “lower taxes for business”, refuse to let the noun do the work. Ask what is measured, on which date, after which deductions and at which legal entity. Is wealth gross assets or assets minus debt? Does it include pensions, private companies, farms and homes? Is business tax corporation tax, payroll contributions, property charges, VAT compliance or the tax paid by owners?

This is not pedantry. Different bases reach different people and produce different responses. A tax on land value, a tax on property transactions and a tax on rental income can all be described as property taxation while doing unlike things. A broad label allows supporters and opponents to argue against different policies without noticing.

The rate matters after the base exists. Until then, the proposal has no denominator and no operating meaning.

Separate liability, remittance and incidence

Draw three boxes. The first contains the person legally liable. The second contains the person who calculates and sends the payment. The third contains the people whose real income or wealth may fall after prices, wages, returns and asset values adjust.

Sometimes the same person belongs in all three. A homeowner facing an unexpected annual charge may remit it and bear it. Often the boxes differ. Employers withhold employee tax. Retailers remit VAT collected in prices. Companies remit tax whose burden reaches owners, workers or customers. A platform can report and collect tax arising from a seller’s activity.

Then add time. Who bears the cost this month, after contracts renew and after investment or location changes? This simple map prevents one of the most common tricks in tax politics: moving the collection point and announcing that the burden has moved with it.

Translate the headline rate

Every percentage needs a label. Ask whether it is statutory, marginal, average or effective. Identify the numerator and denominator. Check whether employer charges, employee charges, benefits, indirect taxes or inflation are included.

For a personal tax schedule, calculate the bill slice by slice. For labour, compare gross wage, net pay and total employer cost. For a company, inspect taxable profit and the timing of deductions rather than multiplying the accounts profit by the headline rate. For a capital gain, ask whether the denominator is nominal or adjusted for inflation, and whether tax is deferred until sale.

Watch thresholds. A bracket usually creates a kink; an allowance withdrawal or lost benefit can create a much sharper effective rate. Frozen thresholds can raise the burden without changing a printed percentage. The translation turns “the rate is 40 per cent” from a conclusion into the beginning of a calculation.

Draw the whole tax-and-benefit system

Distributional arguments fail when they isolate the convenient instrument. Put direct taxes, social contributions, cash benefits and major consumption or property taxes on the same page. State whether the unit is an individual or household, whether the comparison uses annual income, current expenditure or an explicitly modelled longer-run measure, and whether public services are valued.

This does not require pretending every item can be measured perfectly. It requires exposing omissions. A consumption-tax increase with targeted compensation is a different policy from the same increase alone. A pension relief may be progressive or regressive depending on who receives it, which rate applies and how later withdrawals are taxed. An inheritance tax cannot be judged solely by the estate that writes the cheque if the question is opportunity among heirs.

Once the package is visible, disagreement usually becomes more honest. People may still choose different balances between equality, insurance, incentives and revenue. At least they are choosing between complete packages.

Find the margin that can move

Ask what the taxpayer can change. The answer may be hours worked, participation, spending, saving, investment, sale timing, legal form, residence, reported value or compliance. Do not assume that every response destroys real activity. Delaying a sale, shifting income between years and relabelling salary as dividends can shrink the base with little change in production. A pollution tax can shrink its base because the policy worked.

Look for control. Employees whose wages are reported by employers have less room to alter the reported figure than self-employed people. Listed securities have observable prices; a private company does not. Land cannot cross a border; intellectual property can be assigned through legal entities. The same rate can therefore produce different responses on different bases.

A forecast should name the expected margin, evidence and time horizon. “People will leave” and “nobody changes behaviour” are competing slogans until those details appear.

Test the information chain

For any proposed tax, ask who knows the taxable fact and who can verify it. An employer knows wages. A bank knows interest paid. A land registry knows a transfer. A household knows some private use and mixed expenses that no third party sees. The gap between those information positions predicts much of the administrative difficulty.

Then trace the process: registration, record, return, payment, match, enquiry, correction and appeal. Estimate the burden on compliant taxpayers as well as the revenue authority. A rule that raises little from thousands of small payers while requiring elaborate records may fail even if its economic target is defensible. A threshold can reduce that burden and create an incentive to stay below it.

Digital systems can pre-populate, cross-check and collect in near real time. They can also reproduce wrong data at scale. Ask how a person sees the evidence, corrects it and reaches an independent decision-maker. Collection without a workable correction route is efficient only from the collector’s chair.

The limits

Tax analysis can reveal mechanisms and still leave the central political choice open. Evidence can estimate how a base responds, who is likely to bear a burden and what administration costs. It cannot decide how much inequality is acceptable, whether inherited wealth deserves different treatment from earnings, how much privacy should be traded for enforcement or which generation should finance a public investment.

Incidence estimates are often ranges shaped by assumptions. Long-run adjustment can differ from short-run experience. A result from one country may depend on its labour institutions, informality, reporting systems or access to substitutes. People can respond at margins the data do not capture. The absence of measured movement is not proof that the burden stayed where law placed it.

Simplicity also has limits. Removing a relief may broaden the base and create a hardship the relief was designed to prevent. Integrating two taxes can improve coherence while forcing a difficult transition on contracts and assets formed under the old rules. The cleanest system on paper may demand information the state cannot obtain fairly.

This book gives a method for seeing those trade-offs. It does not turn them into one correct rate schedule.

The one thing to keep

Keep the chain.

Start with the base: what fact or event has law chosen to tax? Move to the rate: which percentage applies to which slice and denominator? Identify liability and remittance, then refuse to stop there. Follow the burden through prices, wages, returns and asset values. Look for the margin that can move. Finish with the information trail that makes the liability collectible and contestable.

That chain changes the way a tax claim sounds. “Tax the company” becomes a question about profit, rent, investment and people. “The rate is 45 per cent” becomes a question about the next pound, the average bill and other withdrawals. “VAT hurts the poor” becomes a question about annual income, spending, exemptions and compensation. “Close the loophole” becomes a question about why the boundary exists and what new boundary will replace it.

The chain does not make politics disappear. It removes the hiding places where arithmetic, law and economics are blurred together. Once the steps are separate, a person can support a tax while admitting its behavioural cost, oppose a relief while recognising its beneficiaries, or demand more revenue without pretending the legal payer is the final one.

Tax is difficult because governments are trying to convert a changing economy and contested ideas of fairness into rules that can be measured, paid and enforced. The complication is not a fog surrounding the subject. It is the subject made visible.

Follow the chain, and the argument has to show its workings.

Terms

Tax base. The income, value, transaction, asset, quantity or event to which a tax applies. The base determines precisely what is inside the charge before the rate is used, and which nearby activity sits outside it.

Tax liability. The amount a person or entity legally owes under the rules. Liability may differ from cash paid now because of withholding, instalments, credits, refunds, deferral or a dispute under appeal.

Remittance. The act of calculating and sending tax to the authority. The remitter may be an employer, retailer or company rather than the person bearing the economic cost.

Statutory incidence. The legal placement of a tax: who is named as liable or required to remit it. Statutory incidence is visible in law and on returns, but it does not settle the final economic burden.

Economic incidence. The distribution of reduced real income or wealth after prices, wages, returns, employment and asset values adjust. It asks who is worse off because the tax exists.

Marginal tax rate. The tax applying to the next unit of the base, such as the next pound earned. It matters for decisions at the margin and differs from the rate on total income.

Average tax rate. Total tax divided by total income or another stated base. It describes the overall burden for that denominator, not the treatment of the next pound.

Effective tax rate. Tax paid or expected relative to an economic measure after deductions, credits and timing. The term needs a stated denominator, period and calculation method to be meaningful or comparable.

Tax wedge. The difference between total employer labour cost and take-home pay, usually counting income tax and social contributions on both sides of payroll.

Allowance. An amount of income or another base that can be received before tax applies. Allowances can protect low amounts, simplify administration, recognise costs or pursue a policy preference.

Deduction. An amount subtracted from the tax base before the rate is applied. Its value often rises with the taxpayer’s marginal rate unless the deduction is capped.

Tax credit. An amount subtracted from the tax bill after rates have been applied. Credits can deliver equal cash value across rate bands more readily than deductions.

Refundable credit. A credit that can exceed the taxpayer’s pre-credit liability and produce a payment. It makes the boundary between tax relief and cash benefit thin.

Exemption. A person, activity, receipt or asset removed from a tax that would otherwise apply. Exemptions narrow the base and usually require qualifying rules, evidence and boundary policing.

Threshold. A point at which liability, registration, a rate or a relief changes. Thresholds reduce filing or payment burdens for some taxpayers and create incentives around the line.

Tax bracket. A range of the base taxed at a stated marginal rate. In a graduated schedule, entering a higher bracket does not re-tax the lower slices.

Progressive tax. A tax whose burden rises as a share of the chosen ability-to-pay measure. The label depends on the denominator, tax unit, comparison group and period used.

Proportional tax. A tax taking the same share of the stated base at every level. A flat statutory rate can still produce different effective burdens through allowances and credits.

Regressive tax. A tax whose burden falls as a share of the chosen measure as that measure rises. Annual income and current expenditure can give different rankings; neither alone measures lifetime resources.

Residence and source. Connecting rules that allocate taxing claims. Residence links a taxpayer to a jurisdiction; source links income to where work, property or business activity occurs.

Transfer pricing. Rules for pricing dealings between related entities, commonly by reference to terms independent parties would use. Unique assets and integrated functions make comparison difficult.

Withholding. Tax collected at source before income reaches the recipient, commonly through payroll or payments to non-residents. It improves payment timing, reduces arrears and supplies third-party information for matching.

VAT or GST. A broad consumption tax collected through registered firms, which charge tax on sales and usually receive credit for tax paid on business inputs.

Excise duty. A tax on a selected product or activity, often based on quantity, strength, emissions or value. It can seek revenue, behavioural change or both.

Corporation tax. A legal charge on company profit as defined by tax law. The company owes and remits it; the direct effect is lower after-tax profit, while longer-run economic incidence can reach shareholders, workers, customers and owners of less mobile assets.

Capital gains tax. A tax on increases in asset value, commonly charged when the gain is realised by sale or another event. Deferral and valuation dominate its design.

Recurrent property tax. A repeated charge on land, buildings, rental value, market value or a proxy. The valuation date and treatment of improvements shape its effects.

Tax expenditure. A relief embedded in the tax system and measured as revenue forgone against a selected policy benchmark. Its estimated cost depends on what counts as normal taxation.

Tax avoidance. Lawful arrangement of affairs to reduce tax. It ranges from intended use of reliefs to transactions that satisfy legal form while frustrating the rule’s purpose.

Tax evasion. Illegal concealment, misstatement or non-payment of tax. Evasion must be distinguished from avoidance, error, inability to pay and legitimate disagreement over uncertain law.

Go Deeper

The history

Michael Keen and Joel Slemrod, Rebellion, Rascals, and Revenue: Tax Follies and Wisdom through the Ages (Princeton University Press, 2021). This is the inviting next book: a global history written by two economists who understand administration as well as theory. It moves from ancient collection devices to modern avoidance and keeps returning to the questions used here: what can be observed, how people respond and why apparently foolish taxes survived. The range is an advantage and a warning. Read it for recurring problems, failed devices and memorable cases rather than a full treatment of any one country. It is especially good on the way administrative limits shape systems long before economists arrive to redesign them.

The citizen’s guide

Joel Slemrod and Jon Bakija, Taxing Ourselves: A Citizen’s Guide to the Debate over Taxes, fifth edition (MIT Press, 2017). The title is accurate. It explains how to evaluate tax proposals without pretending that economics removes political values. It is strongest on incidence, fairness, behavioural response and the evidence behind competing claims. Much of the institutional detail is American, and rates have changed since publication. The analytical questions travel well even when the schedules do not. Use it to test claims about growth, distribution and reform, then replace its dated figures with current material from the jurisdiction under discussion.

The design

James Mirrlees and others, Tax by Design: The Mirrlees Review (Oxford University Press for the Institute for Fiscal Studies, 2011). This is the major integrated treatment. Instead of reforming one tax at a time, it asks how earnings, saving, consumption, companies, land and transfers should fit together. It is written for the United Kingdom but explicitly seeks principles for open developed economies. It is long, technical in places and unusually clear about trade-offs. Read the introduction, the economic approach and the conclusion before choosing specialist chapters. Its central discipline is coherence: judge each rule by the system it joins, rather than by whether it looks attractive alone.

The evidence

OECD, Revenue Statistics 2025 (OECD Publishing, 2025). Use this as the antidote to claims built from one country or one headline tax. It provides comparable revenue data, definitions and breakdowns across OECD members, with final data through 2023 and provisional figures for 2024. The tables are more useful than the prose once you understand the classifications. Keep the methodological notes beside you: a contribution counted as tax in one framework may have a different political name at home, and composition matters as much as the total. Pair it with the OECD’s Taxing Wages series when the question concerns labour costs, household type or the tax-benefit wedge.

Notes and Sources

Scope and terminology

This book describes mechanisms that recur across modern tax systems rather than the current liabilities of any one reader. Legal labels, filing units, rate schedules and reliefs differ by jurisdiction and change frequently. The distinctions among tax base, statutory incidence, remittance, economic incidence, marginal rate, average rate and effective rate follow standard public-finance usage as presented in Joel Slemrod and Jon Bakija, Taxing Ourselves, and James Mirrlees and others, Tax by Design. The word taxpayer sometimes means the person legally liable and sometimes the person bearing an economic burden; the text separates those meanings wherever the distinction matters.

Revenue mix and the tax wedge

The OECD comparison in Why You Should Care uses Revenue Statistics 2025. For 2023, the latest year with final data for all OECD countries in that edition, social security contributions averaged 25.5 per cent of tax revenue, personal income taxes 23.7 per cent, VAT 20.5 per cent and other consumption taxes 10.8 per cent. These are shares of tax revenue, not shares of national income, and the OECD applies a common classification that may differ from domestic political labels.

The 35.1 per cent labour-tax wedge comes from OECD, Taxing Wages 2026. It is the 2025 OECD average for a single worker without children earning the average national wage. The measure relates personal income tax, employee and employer social contributions and relevant cash benefits to total labour costs. It does not describe every worker, every tax paid by that worker or the value of public services received.

Salience and the grocery experiment

The Californian grocery-store result comes from Raj Chetty, Adam Looney and Kory Kroft, “Salience and Taxation: Theory and Evidence”, American Economic Review 99, no. 4 (2009), pages 1145-1177. In the field experiment, posting tax-inclusive shelf labels for selected products reduced demand by 8 per cent. The paper also studied alcohol taxes and developed a model of tax salience. The text uses the grocery result only to show that presentation can alter response while the final price is unchanged. It does not assume that every hidden tax produces the same effect.

Bases, rates and system design

The treatment of bases, deductions, timing, tax expenditures, integrated reform and the interaction of taxes and benefits draws chiefly on Tax by Design and the Institute for Fiscal Studies report Tax and Public Finances: The Fundamentals. The examples using round rates and amounts are illustrative calculations, not descriptions of a current national schedule. The distinctions among statutory, marginal, average and effective rates depend on the denominator and period stated. Effective tax rates on investments can also vary with financing, depreciation, inflation and the timing of deductions.

Tax expenditures are departures from a selected benchmark system that reduce revenue through exemptions, deductions, credits, reduced rates or deferral. The benchmark is not discovered automatically, so estimates involve judgement. Alessandro Turrini and colleagues, Tax Expenditures in the EU: Recent Trends and New Policy Challenges, reviews the concept and measurement across personal income tax, VAT and corporate taxation in the European Union.

Economic incidence

The central incidence claim is deliberately conditional. Law identifies who owes or remits a tax; economic incidence follows adjustment through prices, wages, employment, returns, rents and asset values. The division depends on supply and demand responses, market structure, bargaining institutions, mobility and time. The corporation-tax discussion follows the Institute for Fiscal Studies explanation that shareholders are not necessarily the sole bearers and that investment and location responses can affect workers and customers. No universal percentage split is asserted.

The land-value discussion uses the standard result that a tax on the pure site value of fixed land cannot reduce the physical supply of land. That does not make every property tax neutral. Buildings, maintenance, development, valuation, liquidity and transition rules remain responsive. Tax by Design supplies the integrated treatment of recurrent property taxes, transaction taxes and land-value taxation used here.

Progressivity, consumption taxes and time

Progressive, proportional and regressive are defined relative to a stated measure and period. The discussion of VAT against annual income and expenditure draws on the “Broadening the VAT Base” chapter of Tax by Design. Annual income can be temporarily low while expenditure is financed from savings or borrowing, so ranking households by current income and by current expenditure can produce different patterns. Expenditure may be smoother than annual income, but it is not a direct measure of lifetime resources. This does not establish that a VAT rise is fair. It shows why the unit, denominator, time horizon, exemptions and compensating transfers must be stated.

Behaviour and thresholds

The bunching result comes from Emmanuel Saez, “Do Taxpayers Bunch at Kink Points?”, American Economic Journal: Economic Policy 2, no. 3 (2010), pages 180-212. Saez found clear bunching at the first kink of the United States Earned Income Tax Credit, concentrated among self-employed filers, and at the threshold where income-tax liability begins. The text does not turn that result into a universal elasticity. It uses the contrast between self-reported and employer-reported income to show that control over the reported base matters.

The discussion of the Laffer curve states only its logical shape: revenue is zero at a zero rate, and an extreme enough rate can shrink the base sufficiently to reduce revenue. Whether any actual rate lies beyond the revenue-maximising point is an empirical question. The book therefore rejects claims that a rate cut funds itself unless evidence is supplied for the specific base, starting rate, response margin and time period.

Withholding and PAYE

The UK Parliament’s history of wartime taxation records that Britain introduced Pay As You Earn in 1944, when millions of workers were paying income tax, and that weekly or monthly deduction by employers allowed more efficient collection than annual or twice-yearly payment. The text makes no claim that withholding began in Britain or in 1944 worldwide.

Third-party reporting and Danish audits

The Danish evidence comes from Henrik Jacobsen Kleven, Martin B. Knudsen, Claus Thustrup Kreiner, Søren Pedersen and Emmanuel Saez, “Unwilling or Unable to Cheat? Evidence from a Tax Audit Experiment in Denmark”, Econometrica 79, no. 3 (2011), pages 651-692. The study used a representative sample of more than 40,000 individual filers, random audits and later threat-of-audit letters. Evasion was close to zero for income subject to third-party reporting and substantial for self-reported income. Denmark’s extensive reporting and institutional setting limit direct numerical transfer to other countries; the information mechanism is the retained finding.

VAT and the invoice chain

The VAT calculation in the operating section is illustrative. At a 20 per cent rate, tax collected at successive stages, net of input credits, sums to 20 per cent of the final pre-tax sale when every firm is registered and fully compliant. Exemption and zero rating differ because an exempt supplier generally cannot recover input tax while a zero-rated supplier can, subject to the jurisdiction’s rules.

The enforcement evidence comes from Dina Pomeranz, “No Taxation without Information: Deterrence and Self-Enforcement in the Value Added Tax”, American Economic Review 105, no. 8 (2015), pages 2539-2569. Two randomised experiments among more than 400,000 Chilean firms found preventive deterrence where transactions already generated a paper trail and enforcement spillovers along the VAT chain. False invoices, missing traders and fraudulent refunds show that the mechanism is not complete self-enforcement.

Tax gaps

The United Kingdom figures come from HM Revenue and Customs, Measuring Tax Gaps 2026 Edition. The provisional estimate for 2024 to 2025 was 6.4 per cent of total theoretical tax liabilities, or £59.2 billion, against estimated liabilities of £924.4 billion. HMRC describes the two most recent years as projections because of data lags, publishes uncertainty ratings and revises earlier estimates as data and methods improve. The text therefore calls the tax gap a modelled management estimate, not a direct count of evasion or money known to be missing.

Company profit, saving and transfers

The book distinguishes accounting profit, taxable profit and cash. Capital allowances, loss relief, interest deductions and group rules can alter the timing and present value of corporation tax without changing the headline rate. OECD, Corporate Tax Statistics 2026, and Tax by Design were used to check terminology and the distinction between statutory and effective corporate tax rates. The book does not quote a single cross-country effective rate because the result depends on the modelled investment and financing assumptions.

The saving section follows the standard distinction between taxing contributions, returns and withdrawals. Pension accounts can exempt contributions and returns before taxing withdrawals, or tax contributions while exempting later returns. Similar-looking annual deductions can therefore imply unlike lifetime treatment, and opposite-looking cash flows can produce comparable treatment. Capital gains are described as creating valuation, liquidity, inflation and realisation problems. Estate and gift taxes are treated as transfer systems requiring aggregation and valuation, not as a recommendation for a particular rate.

The annual net wealth-tax paragraph is qualitative. It distinguishes visible financial assets from hard-to-value private and non-financial assets, and notes valuation, ownership, liquidity, exemption and mobility problems without asserting one universal behavioural effect. The treatment was checked against OECD, The Role and Design of Net Wealth Taxes in the OECD.

Cross-border taxation

The distinction among residence, source and destination is common to international tax law and public finance. The account of treaties and permanent establishments was checked against the OECD Model Tax Convention and its 2025 update. Transfer pricing and the arm’s-length principle were checked against the OECD Transfer Pricing Guidelines for Multinational Enterprises and Tax Administrations 2022. The text notes that prices for unique intangibles and highly integrated functions may be difficult to observe; it does not imply that all internal prices are arbitrary.

The Pillar Two description was current on 3 September 2026. The OECD/G20 Global Anti-Base Erosion rules establish a coordinated system of top-up taxation when the jurisdictional effective rate for an in-scope multinational group falls below 15 per cent. Scope tests, income calculations, covered taxes, substance-based exclusions, safe harbours and ordering rules make the full regime more qualified than the one-sentence explanation. The OECD published consolidated commentary on 28 May 2026 incorporating agreed administrative guidance released through May 2026, alongside further implementation material. The book explains the operating model rather than jurisdiction-specific commencement dates.

Administration, digital systems and appeals

OECD, Tax Administration 2025, compares registration, filing, payment, assessment, compliance management, dispute handling and digital transformation across 58 advanced and emerging economies, mainly using 2023 administrative data. It supports the general account of pre-filled returns, matching, risk selection and digital administration. The text adds the necessary counterweight: automated collection can scale incorrect data or opaque decisions, so correction, reasons and independent appeal belong inside administrative quality.

Complexity

The final model draws on Keen and Slemrod’s historical survey, Slemrod and Bakija’s policy framework, the Mirrlees Review and the empirical work above. Complexity can come from poor drafting, but it also follows from multiple bases, distributional distinctions, behavioural response, information limits, international overlap, policy reliefs, political durability and transition from old rules. The claim is not that complexity is desirable. It is that a short code cannot preserve every distinction, relief, anti-avoidance safeguard and transition while removing the rules that express them.

Data date and uncertainty

Current institutional material was checked on 3 September 2026. Publication dates are not data dates: Revenue Statistics 2025 gives final comparable revenue composition through 2023, Taxing Wages 2026 uses labour-tax data through 2025, Tax Administration 2025 mainly reports 2023 administration data, and HMRC’s 2026 release estimates the 2024 to 2025 tax year. Cross-country statistics remain sensitive to classifications, household assumptions and measurement methods. The text uses them as bounded anchors rather than timeless constants.

Bibliography

Empirical research

Chetty, Raj, Adam Looney and Kory Kroft. “Salience and Taxation: Theory and Evidence.” American Economic Review 99, no. 4 (2009): 1145-1177. doi:10.1257/aer.99.4.1145.

Kleven, Henrik Jacobsen, Martin B. Knudsen, Claus Thustrup Kreiner, Søren Pedersen and Emmanuel Saez. “Unwilling or Unable to Cheat? Evidence from a Tax Audit Experiment in Denmark.” Econometrica 79, no. 3 (2011): 651-692. doi:10.3982/ECTA9113.

Pomeranz, Dina. “No Taxation without Information: Deterrence and Self-Enforcement in the Value Added Tax.” American Economic Review 105, no. 8 (2015): 2539-2569. doi:10.1257/aer.20130393.

Saez, Emmanuel. “Do Taxpayers Bunch at Kink Points?” American Economic Journal: Economic Policy 2, no. 3 (2010): 180-212. doi:10.1257/pol.2.3.180.

Official and institutional sources

Delestre, Isaac, and Helen Miller. Tax and Public Finances: The Fundamentals. London: Institute for Fiscal Studies, 2023.

Turrini, Alessandro, Julien Guigue, Áron Kiss, Alexander Leodolter, Kristine Van Herck, Frank Neher, Chrysa Leventi, Andrea Papini, Fidel Picos, Mattia Ricci and Federica Lanterna. Tax Expenditures in the EU: Recent Trends and New Policy Challenges. European Economy Discussion Paper 212. Luxembourg: Publications Office of the European Union, 2024. doi:10.2765/651221.

HM Revenue and Customs. Measuring Tax Gaps 2026 Edition: Tax Gap Estimates for 2024 to 2025. London: HM Revenue and Customs, 2026.

Institute for Fiscal Studies. “Corporation Tax Explained.” IFS Taxlab. Last reviewed 30 November 2022.

OECD. Model Tax Convention on Income and on Capital 2017: Full Version. Paris: OECD Publishing, 2019.

OECD. OECD Transfer Pricing Guidelines for Multinational Enterprises and Tax Administrations 2022. Paris: OECD Publishing, 2022. doi:10.1787/0e655865-en.

OECD. Revenue Statistics 2025: Disentangling Personal Income Tax Revenue in OECD Countries. Paris: OECD Publishing, 2025. doi:10.1787/3a264267-en.

OECD. The Role and Design of Net Wealth Taxes in the OECD. OECD Tax Policy Studies, no. 26. Paris: OECD Publishing, 2018. doi:10.1787/9789264290303-en.

OECD. Tax Administration 2025: Comparative Information on OECD and Other Advanced and Emerging Economies. Paris: OECD Publishing, 2025. doi:10.1787/cc015ce8-en.

OECD. The 2025 Update to the OECD Model Tax Convention. Paris: OECD Publishing, 2025. doi:10.1787/5798080f-en.

OECD. Tax Challenges Arising from the Digitalisation of the Economy: Consolidated Commentary to the Global Anti-Base Erosion Model Rules 2026. Paris: OECD Publishing, 2026. doi:10.1787/4377e89f-en.

OECD. Corporate Tax Statistics 2026. Paris: OECD Publishing, 2026. doi:10.1787/73af6222-en.

OECD. Taxing Wages 2026: The Progressivity of Labour Taxation in OECD Countries. Paris: OECD Publishing, 2026. doi:10.1787/3a5169ef-en.

UK Parliament. “The Cost of War.” Living Heritage: Taxation.

Books and major syntheses

Adam, Stuart, Tim Besley, Richard Blundell, Stephen Bond, Robert Chote, Malcolm Gammie, Paul Johnson, James Mirrlees, Gareth Myles and James M. Poterba. Tax by Design: The Mirrlees Review. Oxford: Oxford University Press for the Institute for Fiscal Studies, 2011.

Keen, Michael, and Joel Slemrod. Rebellion, Rascals, and Revenue: Tax Follies and Wisdom through the Ages. Princeton: Princeton University Press, 2021.

Slemrod, Joel, and Jon Bakija. Taxing Ourselves: A Citizen’s Guide to the Debate over Taxes. 5th ed. Cambridge, MA: MIT Press, 2017.