RESEARCH DOSSIER

Corporate Profit Shifting

Abstract. This Research Dossier examines the tax-sensitive geographical allocation of multinational corporate income. It integrates the economics of intra-group transactions, corporate boundaries, tax competition and public finance with a critical assessment of firm-level regressions, transaction-linked studies, foreign-affiliate statistics, country-by-country reporting and international reform. The analysis distinguishes profit redistribution from aggregate base erosion, accounting locations from productive activity, and behavioral semi-elasticities from estimates of shifted income and government revenue losses. Historical and contemporary evidence is interpreted within each source's population, fiscal period and counterfactual. Particular attention is given to developing economies, extraction, intangible assets and the architecture of the global minimum tax through 2026. The original contribution is a disciplined analytical synthesis and illustrative accounting comparisons that explain the compatibility and limits of available estimates. No new empirical estimate or formal theory is claimed.

Keywords: Corporate profit shifting; multinational enterprises; offshore economics; tax base erosion; international taxation; transfer pricing; country-by-country reporting; corporate tax competition; economic substance; global minimum tax; fiscal externalities; empirical identification

Scientific Object and Scope

Corporate profit shifting is the tax-sensitive redistribution of reported multinational corporate profits across fiscal jurisdictions relative to the distribution that would arise under a specified alternative allocation of income. The economic object includes the choices of firms, the institutions that structure those choices, and the fiscal and statistical consequences of their aggregation. A profit that appears in one country's accounts can represent the return to production, financing, customer relationships, entrepreneurial risk, or a contractual claim located elsewhere. These dimensions of location are related but rarely identical.

The analysis therefore starts with the multinational group rather than the individual registered company. A group coordinates production and ownership across national borders while its constituent entities face separate fiscal rules. It can change the allocation of its recorded income through related-party transactions and organizational arrangements without necessarily changing where its employees work, where its goods are produced, or where final customers are located. It can also change its physical operations in response to taxation. The latter decision changes the geographical distribution of production and investment itself. Both responses matter to public finance, but they require different empirical explanations.

This dossier asks how taxes alter the attribution of corporate profits, how economists distinguish reporting responses from changes in production, how the magnitude is measured, and what kinds of intervention change the incentives and distributional outcomes. Its primary disciplinary membership is Offshore Economics. International taxation provides a necessary institutional context; legal validity of a particular transaction requires a separate fact-specific legal analysis. Offshore household wealth refers to personal assets, often measured as stocks, whereas the corporate outcome considered here is a flow of earnings, normally expressed per accounting period. The two may be related by ownership or dividend distribution, but their measures cannot be combined as if they represented independent instances of the same process.

The use of the term offshore concerns jurisdictional intermediation, including relationships within the same regional market. Profit shifting requires neither an island jurisdiction nor an offshore financial center in any geographical sense. Conversely, foreign investment, international financing and the operation of a holding company have legitimate commercial functions. A country's aggregate profitability can signal the existence of an allocation question without identifying a specific transaction, legal offense or taxpayer. This separation preserves the difference between economic behavior, an administrative assessment and a judicial finding.

Historical Development and Institutional Foundations

Modern research developed around the observation that the distribution of cross-border corporate earnings sometimes bears a weak relationship to the distribution of conventional production factors. Hines and Rice (1994), in Fiscal Paradise, documented unusually high reported profitability among American affiliates in tax havens using 1982 evidence. This established a research problem that extends beyond the identification of jurisdictions with low tax rates. Unusually profitable foreign affiliates can become economically significant components of a multinational's reporting structure even if they account for a relatively modest proportion of its tangible operations.

The expansion of cross-border ownership has made the question increasingly complex. An industrial multinational can own suppliers, distributors, research laboratories, financial affiliates and intellectual property rights across several jurisdictions. Historically, domestic production and domestic corporate income taxation could often be analyzed through relatively close economic and legal boundaries. Multinational production separates those boundaries. Jurisdictions still retain distinct powers over taxpayers and tax bases while business groups plan their activities as integrated economic systems. The resulting tension concerns the allocation of jointly produced returns between institutions that exercise fiscal authority separately.

The historical series by Wier and Zucman (2022), historical profit-shifting series, estimates a substantial rise in profit shifting across the period beginning in 1975. This is a reconstructed series rather than a census of tax avoidance transactions. Earlier values depend especially on historical data coverage and extrapolation. The 2022 working-paper version and later revisions should be distinguished because methodological and data changes alter estimated levels. The 2026 updated historical working paper by Gidron, Wier and Zucman extends a common post-2015 method through 2023 and reconstructs earlier years with additional assumptions. The authors estimate that 34% of foreign profits were shifted in 2023, and that lost revenue equaled 8% of collected corporate tax. The authors report a decline from the earlier peak; the series alone cannot isolate the causal contribution of individual reforms. Comparisons with the 2022 version require reconciliation of methods and denominators.

The institutional development of reporting standards and minimum tax coordination creates successive changes in the environment of multinational groups. A historical account should examine the evolution of those environments rather than retrospectively applying a contemporary legal or statistical category to every arrangement. International business can grow rapidly while the proportion of its profit allocated to low-tax jurisdictions remains stable. In that situation, the global amount exposed to allocation practices may rise even in the absence of a greater shifting propensity per dollar of foreign profit. Separating economic growth, internationalization and behavioral intensity is essential to understanding long-run trends.

Corporate Boundaries and the Accounting of Profit

Consider a group with operations in countries A and B. Before any allocation change, the affiliates earn 100 and 20 monetary units respectively, for a combined profit of 120. Let the tax rates be 25% and 5%. Under simplified equal accounting and taxable income definitions, no withholding, no credits, no losses and no minimum tax, tax payments are 25 and 1, for a total of 26. An additional 40-unit internal expense recognized in A and equal income recognized in B changes their booked profits to 60 and 60. Group pre-tax income remains 120, while group tax falls to 18. A loses 10 in tax and B gains 2. The difference between A's revenue loss and the reduction in worldwide tax is the revenue obtained by B. This hypothetical example is an accounting illustration, not a normative recommendation about structuring transactions.

The identity depends on symmetry between the expense and income. Internal payments eliminate on consolidation only to the extent that the same transaction is recognized consistently. Differences in timing, currency, tax base and legal classification can break the identity in tax accounts even though consolidated financial reporting eliminates the intragroup transaction. A payment deductible in one jurisdiction but excluded in another can reduce total taxable income. By contrast, a payment recognized by both may primarily redistribute the base. The economic study must therefore distinguish redistribution of an aggregate base from erosion caused by inconsistent recognition.

Four profit concepts are indispensable. Economic profit is a return above the opportunity cost of inputs. Accounting profit is the residual specified by accounting recognition and measurement. Taxable profit follows fiscal definitions, including deductions, exemptions, depreciation schedules and carryforwards. Consolidated group income eliminates intragroup transactions for financial reporting. A large accounting profit in one entity may coexist with little current taxable profit because of a loss carryforward or an incentive. Tax paid in cash may lag the period when income is generated. A jurisdictional accounting figure divided into taxes paid creates a backward-looking ratio; it is not automatically the marginal rate on the next dollar of activity.

Affiliates may also report dividends received from other group members. If an analyst adds the underlying earnings in one place and the distributing company's dividend income in another, the same income is counted twice. The same problem arises when upstream ownership chains include multiple holding entities whose accounts record returns on participation. Data harmonization must decide whether the unit is an unconsolidated entity, a jurisdictional subgroup, a controlled affiliate population, or the consolidated enterprise. A correction to overlap is a statistical operation and should be documented before interpreting a profitability gap as a tax response.

Profit is a residual flow. Its location can change without a directly observable movement of currency, tangible assets or securities. Conversely, a large cross-border capital transfer can occur without any corresponding current-year profit. An offshore stock of assets and an annual flow of shifted earnings answer different questions. This matters particularly in a research program that also studies holding companies, direct investment and household wealth: common organizations may appear in several datasets without representing additive measures of an offshore economy.

Incentives, Costs, and Corporate Responses

An elementary economic description begins with the tax saving on a unit of profit allocated away from a higher-tax location toward a lower-tax location. The saving depends on taxes actually relevant to the incremental unit, including the consequences of parent-country rules, withholding, credits and any top-up liability. It is the net relative treatment of income across the feasible locations that matters. A single headline statutory rate provides useful information but can be a poor complete measure of the tax incentive.

Suppose a group considers a marginal change in the reporting allocation. The expected benefit increases with the relevant tax differential. Offsetting costs include personnel and systems required for cross-border operations, negotiation and administration, expected adjustments and penalties when applicable, uncertainty, litigation, reputational costs and the opportunity cost of organizational complexity. These costs differ between firms, transactions and jurisdictions. Expected after-tax returns can therefore change at different rates even when groups confront identical statutory tax rates.

Huizinga and Laeven (2008), in their multi-country perspective, develop the insight that the incentive faced in any individual country depends on tax differences across the group's network of jurisdictions. This is important for interpreting a corporate rate change: the affected country is embedded in a firm's ownership and transactional network, and reallocations can occur toward more than one destination. The weight assigned to each alternative depends on the structure that the empirical method adopts. A bilateral difference between two observed entities can omit important alternatives.

The extensive margin concerns whether a group maintains a presence or arrangement in a particular jurisdiction. The intensive margin concerns the amount of profits reported within a given structure. These margins react on different horizons. A reorganization requires changes in ownership, contractual relationships or control functions; a change in reported margins may be associated with a much shorter accounting period. A tax reform can depress one margin while stimulating the other. The first-year response accordingly should not be treated as a stable long-run elasticity.

Organizational costs can also produce a relationship between firm size and shifting capacity. Fixed administrative costs have a smaller average burden on a large amount of income, but larger groups can face more extensive scrutiny and more complicated coordination. Small international businesses may retain local tax advantages without establishing elaborate structures. The sign and magnitude of size effects remain empirical questions. Any inference from large-group data to all multinational corporations must address coverage and composition.

The conventional comparison holding real activity constant has analytical value because it isolates an income-allocation response. It does not imply that real and accounting decisions are independent across years. Where a jurisdiction changes the cost of locating a function, the business may genuinely transfer employees, managerial responsibility or productive capital. A long-term reform assessment must account for the joint evolution of the production base and the reporting base.

Commercial Transactions, Intangibles, and Finance

International groups use commercial transfer prices to account for exchanges of merchandise, services, rights and financial claims between associated entities. Every such price determines both a seller's revenue and a purchaser's expense or asset cost. Commercially supportable prices make it possible to attribute transactions to separately accounted business units even when the group has coordinated the underlying activity. The economic concern arises when tax incentives influence the distribution of the resulting margins independently of underlying commercial factors.

The study by Liu, Schmidt-Eisenlohr and Guo (2020), based on linked UK trade-tax data, connects export transactions with corporate tax records. The authors find tax-sensitive patterns in related-party export pricing and evidence that a shift in the United Kingdom's taxation of foreign earnings altered the incentive to report income abroad. The study addresses a defined country, period and class of transaction. It does not measure every form of profit shifting through goods or services.

A transaction-price comparison is strongest where independent and related-party transactions share the same product specifications, market, transport conditions, contractual quantities, timing and currency treatment. Apparent unit-price differences can instead reflect quality, warranty, volume discounts, location risk or the composition of deliveries. Customs classifications often cover heterogeneous goods. A researcher should demonstrate comparable observations and inspect sensitivity to alternative comparison groups. The operation of an associated distributor may also differ systematically from that of an independent customer. This creates a selection problem distinct from mismeasurement of a price.

Intangible assets intensify the attribution problem. A patent, brand, software system or accumulated know-how may create returns in many markets without being consumed in the way a physical intermediate good is consumed. Legal title, the teams that develop the asset, the group entities that manage it, and the locations where customers use it can differ. Because development is risky and many intangibles are unique, observed market transactions may provide few close comparisons. Even genuine enterprise-specific rents need a method for attribution. A substantial return to intangible property is not itself proof that the amount was improperly assigned.

Internal finance presents a parallel problem. A group may allocate funding to support genuine investments, centralize treasury functions or manage currency and maturity exposures. Interest deductibility and differences in creditor-country treatment create an incentive over the allocation of financing costs. The OECD (2016) Action 4 update recommends constraining net interest deductions in relation to taxable earnings. Such rules seek a verifiable link to activity without requiring administrators to estimate a correct group financing arrangement for every payment. They can also restrict deductions for genuine leveraged business operations, which explains the need to assess the interaction with earnings volatility and actual financing needs.

Beer, de Mooij and Liu (2018) organize evidence in their review of channels and magnitudes around several channels, including transaction pricing, debt and institutional arrangements. The evidence indicates that no single accounting category captures the complete reporting response. An intervention directed only at interest may alter other margins if group organizations can adjust. A fall in related-party loans therefore provides insufficient evidence about the global effect of an interest limitation in the absence of a broader analysis of profit and tax accounts.

Other mechanisms arise from the interaction of fiscal rules governing the same entity or payment across jurisdictions, including differences in residency, treatment of dividend distributions, characterization of debt and equity, or timing of tax recognition. These can create base erosion without a symmetric profit gain elsewhere. Structural differences among legal systems are an explanatory variable in economic studies. The legal assessment of a specific arrangement depends on the law applicable at the relevant time and the underlying facts. The present analysis describes mechanisms for measurement and policy evaluation, not instructions for exploiting them.

Economic Substance and the Allocation Counterfactual

The central measurement difficulty is the construction of an alternative allocation against which observed profit can be evaluated. There is no universal physical unit of profit. A wage payment has an identifiable contractual recipient, a factory has an identifiable site, but a firm's residual return can depend simultaneously on workers, specialized capital, past research, market demand, management and exclusive rights.

A conventional production-based benchmark relates profits to observable inputs and possibly industry conditions, productivity or demand. A benchmark based on comparable independent firms estimates returns consistent with similar market transactions. A division of multinational profit by employment, assets and sales assigns income according to explicitly chosen weights. Each method answers a different question. Their results diverge when production is jointly organized, intangibles are economically important or returns exceed normal competition.

Consider two otherwise identical factories that utilize the same software, one owned by its developer and another obtaining it through an independent supplier. Their observable physical assets and payroll can be equal, yet their legitimate costs and financial returns need not be equal. An allocation proportional to payroll assigns the same gross profit to both; a market-price method recognizes a cost for the independent license and must determine the appropriate return for the integrated enterprise. The difference illustrates why a low ratio of employees to profits can be a risk indicator without being self-sufficient evidence of shifting.

The taxpayer-level allocation considered by the OECD (2022) Transfer Pricing Guidelines uses the widely established arm's-length approach. As an institutional benchmark, its function is to determine tax treatment according to an accepted comparison of commercial and financial relations, while managing risks of economic double taxation. Economic research can assess whether actual profit distributions appear sensitive to incentives and whether the application of a benchmark changes those distributions. It cannot substitute an aggregate regression for examination of a taxpayer's facts.

Economic substance is also distinct from the existence of a legal entity. The economic question concerns the activities, decision rights, resources and risks that explain returns. A finance company may have few employees and nevertheless conduct economically meaningful specialized operations. A legal entity with a large payroll may in contrast report income attributable to transactions outside its main business. A sound interpretation asks which return is being measured, the period to which it relates and whether its apparent concentration survives appropriate comparisons.

A benchmark should be stated before the estimate is interpreted. When a study estimates a deviation from a normal profitability relationship, its residual is a model-dependent deviation. When it reconstructs the influence of a tax differential, it estimates a behavioral response. When it allocates a fixed amount of group profit using a formula, it describes a distribution under a rule. These are legitimate analytical exercises, but they cannot be treated as interchangeable measurements of a unique unobserved location of profit.

Firm-Level Econometric Identification

The standard microeconomic approach estimates the relationship between reported affiliate profit and a measure of tax incentives, after accounting for observed inputs and other determinants. A stylized estimating equation is log P = α + βD + γ log K + δ log L + controls + ε, where P denotes positive reported pre-tax profit, K and L represent measured capital and labor, and D is a tax differential defined in the direction of increasing incentive to locate profit in the affiliate. The estimated coefficient β quantifies a conditional association; a causal interpretation requires a defensible source of variation and sufficient controls.

If D is measured as a fraction rather than percentage points, an estimated semi-elasticity of 0.8 implies that a one-percentage-point increase in the incentive is associated, locally, with approximately 0.8% higher reported profit. A ten-point change corresponds to an approximate 8% increase using the linear approximation; using the log specification yields exp(0.08) minus one, about 8.33%. The difference becomes important for larger scenarios. Reversing the definition of D reverses the sign of the coefficient, not its underlying economic meaning.

Heckemeyer and Overesch (2017), in their meta-analysis of profit responses, report a predicted semi-elasticity near 0.8 across the studies they synthesize. Beer and colleagues' later review, cited above, reports an average response closer to one percent for a one-percentage-point rate difference. These are estimates conditioned on study design and evidence composition. Neither should be mechanically interpreted as a universal global tax revenue loss. The average of estimates is affected by variation in samples, identification strategies, functional forms and the kinds of profit observed.

A major identification challenge is endogeneity. Governments may lower corporate tax rates in response to actual or expected base erosion. A firm may enter a low-tax jurisdiction because it has valuable commercial opportunities or expects high profitability. A tax rate can be correlated with market access, public infrastructure, industry composition and institutional quality. Controlling for observed capital and employment cannot guarantee that the remaining correlation represents the reporting channel. Unobserved productivity and reverse causation remain central concerns.

A second challenge concerns losses. Log profit is undefined for zero and negative profit. Dropping unprofitable affiliates changes the sample and may exclude exactly those firms whose reported results are most sensitive to tax incentives. A researcher may estimate the probability of positive profit separately, consider a transformation capable of retaining losses or use suitable level-based methods. Every choice changes the meaning of the estimand. A near-zero reported profit can also result from genuine business conditions or accounting policies. It should not automatically be classified as tax-motivated.

Johannesen, Tørsløv and Wier (2020), in their cross-country micro-data study, improve identification by emphasizing the rates confronting foreign affiliates rather than relying solely on domestic rates and by examining a near-zero-profit indicator. They report stronger profit responses in less developed economies in their sample. Their approach illustrates the value of variation that changes a firm's international reporting incentive without directly changing the tax rate on its own local production. Network location choices and the representativeness of available corporate accounts still warrant examination.

Fixed effects can absorb persistent differences between affiliates, years, industries or multinational groups. They can also remove useful variation when statutory rates change little within the observation period, leaving identification dependent on narrow events. Event studies around tax changes require checks for anticipation, differential pre-trends, simultaneous regulation and changes in the composition of firms. A group whose income becomes taxable at a new threshold may adjust before the legal date. A credible study must identify who was exposed to the change and distinguish changes in eligibility from changes in behavioral intensity.

Additional analyses should compare related and unrelated firms, examine outcomes less likely to react to pure reporting incentives, and test alternative measures of tax differences. If a rate change is followed by a large rise in employment and capital, a simultaneous rise in reported profit may contain a production effect. If profits change while measurable operations remain stable and a closely matched control population shows no similar response, the reporting interpretation becomes more plausible. Neither pattern alone resolves all possible confounding.

The scientific value of the semi-elasticity literature lies in its identification of tax sensitivity, heterogeneity and the relative importance of different channels. Extrapolation to a global dollar amount requires a model of the set of firms, their initial profit distribution, available destinations, counterfactual tax differences and the degree to which regression estimates capture reallocation rather than real changes.

Macroeconomic Measurement and Bilateral Allocation

Macro studies approach the question from jurisdictional totals. They compare earnings attributable to foreign affiliates with indicators of productive activity, such as wages, or examine international accounts recording cross-border income and services. Their coverage can extend beyond commercial micro-databases, in which accounts from some low-tax jurisdictions are incomplete. The tradeoff is that aggregation obscures sectoral and organizational differences.

Tørsløv, Wier and Zucman (2023), in The Missing Profits of Nations, employ foreign-affiliate statistics and bilateral international accounts. Their preferred estimate puts the fraction of foreign multinational earnings shifted to tax havens in 2015 at 36%. The relevant denominator is profit earned outside the country of the multinational parent, not all worldwide corporate profit. The author manuscript dates from April 2022; the final journal issue is 2023. Preserving that distinction avoids treating a publication year as the observation year.

The underlying allocation logic compares foreign and domestic profitability in haven jurisdictions and uses bilateral information to attribute shifted earnings to source economies. When high foreign-affiliate profits are systematically disconnected from local factor payments, the data support a residual allocation estimate under a stated benchmark. The method does not directly observe each group's counterfactual transaction prices. The benchmark faces the possibility that foreign and domestic firms operate in genuinely different sectors or earn different economic rents. Attribution to countries of origin adds another layer of assumptions concerning routes, ownership and recorded payments.

An elementary example illustrates the consequences. Suppose three countries report affiliate profits of 80, 60 and 100. A researcher estimates that the last country's production-related benchmark is 30, producing excess profit of 70. If the two higher-tax countries account for 60% and 40% of a defensible bilateral attribution measure, their reconstructed profits become 122 and 88, and the third country's becomes 30. The world total stays 240. If the last country's benchmark were instead 50, excess profit would fall to 50, and the reconstructed additions become 30 and 20. The initial benchmark choice affects the estimates even when the bilateral allocation weights do not change.

If the two source economies have statutory tax rates of 25% and 20%, the first scenario would imply gross origin-country revenue effects of 10.5 and 5.6 units in a simplified fixed-base calculation. At a 5% receiving-country rate, the receiving jurisdiction collects 3.5 on the assigned excess. Net worldwide revenue reduction is 12.6 units under the assumptions. Alternative deductions and fiscal rules can change this calculation materially. A bilateral allocation weight is also not automatically a measure of the original place of production; it may reflect how the available international accounts resolve the counterparty.

In the national accounts, changes in the recording of intellectual property receipts or multinational income may influence territorial output, trade balances, operating surplus and international investment income. The residence of corporations separates domestic production from income accruing to residents. A profit reported in a country may ultimately accrue to a foreign owner. Hence changes in gross domestic product, gross national income and tax revenue are analytically separate. Adjusted distributions should preserve accounting identities and specify precisely which measure is changed.

Macro residual methods provide a global view that firm-level research alone may struggle to obtain. The two approaches are complementary because a macro estimate depends on an allocation benchmark while a micro estimate relies on variation across firms or over time. Agreement in the direction of results can strengthen the evidence for tax-sensitive reporting, but apparent numerical discrepancies may be explained by different populations, denominators, time periods and estimands. Comparing them requires a crosswalk rather than a mechanical averaging procedure.

Country-by-Country Reporting and Data Quality

Country-by-country reports aggregate selected financial and operational information of a multinational enterprise by tax jurisdiction. They make it possible to compare profit, revenue, employees, assets, and tax outcomes across locations within the reporting population. Administrative access to individual reports supports risk assessment; publicly available aggregate statistics support broad economic analysis without exposing the identities of the underlying groups.

The OECD (2026) 2026 country-by-country statistics chapter covers fiscal year 2023. Sixty jurisdictions supplied statistics for more than 9,400 multinational groups. The chapter reports that investment hubs continue to display comparatively elevated indicators of profit and activity misalignment, while some high-level indicators increased modestly relative to recent years and remained below earlier peaks. The report cautions that the economic environment of 2023, including high inflation, can affect interpretation. The report year and fiscal year must appear together whenever findings are cited.

Consider the ratio of positive pre-tax profit to employees. The numerator may include high returns from capital, financial activities or intellectual property, while the denominator does not measure hours, skills, outsourced labor or wage levels. A holding company with a small workforce produces a high ratio by construction if it earns dividends or participation income. Comparing the ratio across sectors without separating these activities confounds business composition with a possible reporting response. Changes over time can similarly result from entry and exit, changes in reporting coverage or a different mix of groups.

The relevant fiscal outcome must also be defined. Cash taxes paid during the year can include settlement of earlier liabilities; accrued taxes relate more closely to the reporting period but may be affected by accounting treatment. Ratios that divide by profits of both loss-making and profitable entities can become unstable when net profit is small. Aggregating taxes and positive profits separately may change the result relative to averaging group-level effective rates. Each statistic carries information about a different object.

Confidentiality suppression and aggregation limit the researcher’s ability to determine which entities generated an aggregate result. A missing cell does not necessarily equal zero. Different jurisdictions may report different subsets of tables, use different definitions or have periods that do not end on the same date. A published series of totals may consequently have changes in composition. A balanced-jurisdiction analysis, checks for double counting and documentation of missing observations are appropriate prerequisites to interpreting a trend.

A practical reproducibility protocol for an empirical extension would preserve the exact downloaded table and metadata, document the observation year and reporting population, specify filters and corrections, separate loss-making subgroups where feasible, compare alternative measures of substance, and publish the sensitivity of results to coverage choices. Such work would require the underlying tables and a reproducible computational pipeline. This dossier does not claim to have downloaded or recomputed those statistics. The directly inspected report chapter establishes their scope and official interpretation.

Estimates, Denominators, and Comparability

The literature yields several headline quantities. Correct comparison begins by identifying the numerator and denominator, reference period and counterfactual, as well as whether the figure is a directly observed accounting quantity or a modeled estimate. The OECD (2015) Action 11 report estimated worldwide corporate income tax losses of USD 100–240 billion per year at 2014 levels and emphasized serious data limitations. That range is a fiscal estimate, while the 36% figure from the foreign-affiliate approach described earlier is a proportion of multinational foreign earnings in 2015. One is expressed in money per year; the other is a fraction of income.

The historical UNU-WIDER working-paper version identified above describes a relatively stable haven share of foreign multinational profits near 37% across 2015–2019 and estimates an increase in the share of corporate tax revenue lost over that interval from around 9% to 10%. Those figures are attributed to that version, not presented as present-day levels. The revised 2026 working paper reports a lower shifted-foreign-profit share in 2023 than in 2015. A trend within a consistently constructed vintage differs from a comparison of estimates across vintages. Neither vintage alone establishes the causal effect of individual reforms.

The following distinctions help organize comparison:

Quantity Unit Question answered
Reported profit in an investment hub Currency per period How much income is recorded there?
Estimated shifted profit Currency per period How much differs from the study's allocation benchmark?
Share of foreign multinational profit shifted Percent How large is that estimate relative to the defined multinational profit base?
Tax semi-elasticity Percent profit response per rate-point change How sensitive is reported profit to the modeled incentive?
Gross origin-country revenue loss Currency per period How much tax might the origin collect under the alternative?
Net worldwide revenue loss Currency per period How much global tax changes after accounting for gains elsewhere?

Suppose one study gives a semi-elasticity near one and another estimates a shifted-profit share of 36%. The first coefficient describes a local response to a change in the tax incentive; the second is a level under a counterfactual. Neither proves nor refutes the other. A stock adjustment accumulated over decades can be large even if the immediate response is modest. Conversely, a high short-run coefficient in a limited population may coexist with a small aggregate amount when the firm's base is small.

A further difficulty is the use of tax rates to convert the shifted-profit estimate to revenue loss. Gross domestic statutory rates may exceed the actual tax that the reallocated income would attract because of credits, deductions and losses. A correct calculation also distinguishes taxes collected by the recipient country, the parent country and any jurisdictions entitled to top-up tax. If a counterfactual changes business activity, the tax base itself may shrink or expand. A static estimate can be a useful first-order approximation, but it is not a forecast of receipts from enforcement or policy reform.

Analysts should publish a denominator dictionary with any comparative table, including how foreign profits and total corporate profits are distinguished, whether the profits include losses, how parent-country and domestic subsidiaries are treated, and whether the loss share divides by actual or counterfactual revenue. Uncertainty includes not only sampling and standard errors but also competing allocation conventions and data revisions. A precise numerical standard error does not remove uncertainty about the underlying counterfactual. Separate reporting of assumptions, estimated quantities and sensitivity ranges is more informative than a single aggregate number asserted without qualification.

Fiscal Externalities, Developing Economies, and Sectoral Exposure

International profit shifting reallocates the fiscal capacity associated with the corporate sector. A country can lose tax base when the resident affiliates of multinational groups report a greater share of their profits elsewhere. A recipient jurisdiction can gain income and associated revenue. The interests of countries are therefore heterogeneous. The distribution of losses need not correspond to the distribution of absolute economic output or population.

Crivelli, de Mooij and Keen (2015) investigate developing-country spillovers using panel data and distinguish foreign tax spillovers that operate through real decisions from effects attributed to avoidance. The analytical distinction is especially important for developing economies. A change in a foreign country's tax policy can affect the location of a factory or investment, and it can alter the accounting returns assigned to existing local activity. The two mechanisms have different implications for growth and tax revenue.

An economy that relies materially on corporate income taxation from a narrow set of large firms can be exposed to changes in multinational reporting. The same absolute revenue loss has different significance depending on the government's tax base, alternative revenue instruments, administrative capacity and financing constraints. The evidence from developing-country firm-level research discussed previously suggests stronger tax responses in some less developed settings, but the magnitude cannot be universally generalized across all low- and middle-income countries.

Resource extraction illustrates a particularly important distinction between the location of rents and the contractual allocation of earnings. A mineral deposit cannot relocate in response to a tax change. Yet its financing, procurement, marketing services, processing arrangements and legal ownership may be internationally distributed. Resource projects also face volatile prices, long payback periods and substantial sunk costs. The interpretation of reported losses and earnings therefore requires attention to project lifecycles, commodity prices and capital allowance systems.

Beer and Devlin (2021), in the extractive industries study, report that historically sectoral profits in their sample fell by slightly more than 3% in association with a one-percentage-point increase in the domestic corporate rate, with stronger responses for mining than hydrocarbons. The study finds evidence consistent with interest limitation rules reducing observable avoidance and weaker evidence concerning the presence of transfer pricing legislation. These results apply to the relevant sample and method; the presence of a law does not measure its enforcement, nor does a rate-profit coefficient exclusively identify one channel.

Administrative capacity should be treated as an economic input. Specialist knowledge, access to comparable transactions, information exchange, competent dispute resolution and reliable data processing affect expected enforcement costs. Introducing rules without staff, guidance and procedures may increase formal compliance costs while producing little improvement in revenue collection. Conversely, targeted risk analysis can concentrate scarce investigative resources on financially consequential discrepancies. Because an adjustment can produce double taxation, procedural cooperation with other jurisdictions is a component of an effective fiscal system.

The distributional consequences extend beyond governments. Corporate taxes affect after-tax returns and can influence investment and market competition. The statutory liability is formally borne by the corporation, while economic incidence may be distributed among shareholders, employees, consumers and other agents. A jurisdiction that recovers revenue can finance public services or alter other taxes; a recipient jurisdiction may lose a source of government revenue. An assertion about distributive welfare requires a model of these responses, not merely the corporate tax accounting identities.

Market Competition, Investment, and Macroeconomic Interpretation

Tax-sensitive allocation can affect the relative competitiveness of firms with different access to cross-border structures. A group that can lower its aggregate tax burden may retain more internal funds or distribute more income to owners. Domestic businesses with similar activities but fewer foreign affiliates may face different after-tax returns. The magnitude of the competitive effect depends on market structure, financing constraints, the treatment of losses and the extent to which tax savings influence prices or investment rather than distributions.

Profit shifting can also affect tax competition. A government that reduces the cost of locating profits may attract a tax base larger than its own domestic production would otherwise generate. A country with a higher statutory rate may lose part of its base to that strategy and respond by changing general rates, targeted incentives or enforcement. Governments have their own objectives, including stable revenue, employment and development. The fiscal externality arises because an individual policy can affect the bases and welfare of other countries.

The distinction between mobile accounting profit and mobile productive capital matters for these incentives. A low tax rate that attracts a reporting entity without substantial local activity can yield revenue while changing little in local employment. An incentive that genuinely attracts a productive facility changes economic activity and may generate spillovers beyond direct corporate tax receipts. Tax and investment policy evaluation should compare these benefits with fiscal costs and effects on the allocation of resources. Even a jurisdiction that combines both kinds of attraction should not treat them as identical sources of growth.

Measurement of foreign direct investment is likewise affected by the multinational group's financial organization. A holding entity can receive an investment from a parent and invest the same funds in an operating affiliate elsewhere. Bilateral investment positions register the immediate relationships under their statistical definitions. Summing the gross links as if each represented a separate newly created factory would overstate the economic activity attributable to the initial capital. Corporate profit shifting is related to these networks without being measured by their outstanding financial positions.

A corrected allocation of profits may change estimates of domestic operating surplus or the capital share, as the foreign-affiliate literature suggests. It can also affect the interpretation of productivity measured as value added per worker. A rise in recorded value added caused by income attribution can make output per worker increase without an equivalent improvement in local technology or production processes. Researchers must therefore distinguish published national statistics, proposed analytical adjustments, and independently identified changes in real productivity.

An equally important distinction concerns the direction of causation. The existence of profitable enterprises attracts real investment and may influence policy choices; tax incentives affect both the reported profitability and location of those enterprises. A correlation between low tax rates and high profitability is therefore an empirical starting point rather than a sufficient causal conclusion. The economic interpretation gains strength through complementary information about the firm's activities and through changes in incentives that can be isolated from other determinants.

Policy Instruments and International Coordination

Measures directed at profit shifting act through at least three economic mechanisms: reducing the gain from jurisdictional differences, increasing the expected cost of unsupported allocations, and improving the information available to assess those allocations. These mechanisms can be implemented through different institutional designs and may be complementary. Their success depends on the net behavioral effect rather than on the number of regulations enacted.

Transfer pricing rules target attribution at the transaction level. Their value depends on whether comparative evidence can be obtained and interpreted effectively. Interest limitations target a category of deductions through a rule linked to earnings or leverage. Rules addressing foreign controlled entities and inconsistent cross-border classification operate through other elements of the tax base. Disclosure can increase the probability that an allocation will be scrutinized without itself determining an additional tax liability. The policy mix should be evaluated for interactions rather than treating isolated rule effects as additive.

International cooperation serves both enforcement and protection against overlapping taxation. If one country increases the taxable income of a local company but the foreign counterparty receives no corresponding adjustment, the same group income may be taxed twice. The fiscal gain for one country and the compliance burden for a group therefore depend partly on the procedures available to resolve conflicting claims. Coordination creates institutions for information sharing and dispute management while leaving national tax authorities responsible for implementing their legal powers.

Formula-based allocation offers another policy approach. A fixed group base may be divided across jurisdictions using factors such as sales, employment or assets. It reduces reliance on individual internal transaction prices within the scope of the formula but introduces incentives associated with its factors. A payroll-weighted rule can affect labor location; a tangible-asset factor changes the tax implications of investment; a sales factor requires a reliable destination definition. The political economy of factor choice is consequential because the formula distributes fiscal rights differently from transaction-based allocation.

A destination-based tax arrangement emphasizes customer location through a different conception of the fiscal base. Its implications for investment, revenue and reporting depend on the exact design, transition treatment and interactions with other taxes. A structural reform may reduce some profit-shifting incentives while creating different boundary questions regarding the eligible scope and taxpayer. This dossier confines itself to the economic tradeoffs rather than asserting one method as a complete replacement for existing international tax rules.

A useful reform evaluation therefore reports multiple outcomes. Revenue changes must be assessed alongside changes in the allocation of profits, real activities, taxes paid to different countries, compliance and administrative costs, and unresolved disputes. A decline in one accounting indicator may be offset by an increase elsewhere. A rise in cash collections can also result from unrelated rate changes, settlements or changes in the corporate profit cycle. Transparent assessment should document both substitution and simultaneous shocks.

Global Minimum Tax and Evidence through 2026

A coordinated minimum tax changes the marginal advantage of allocating income to a jurisdiction where the covered effective tax burden is below the required level. In the OECD's global minimum tax framework as described in 2026, the headline minimum rate is 15% for in-scope groups. The computation uses jurisdictionally aggregated covered taxes and qualifying income; a substance-based exclusion reduces the amount of income to which the top-up percentage applies. Primary collection rights can belong to a qualified domestic minimum top-up tax, followed by mechanisms at the parent or other implementing jurisdictions under the applicable rule order.

This architecture changes the monetary incentive in the elementary two-country example. Holding the 40-unit shift and all simplified assumptions constant, suppose the 5% tax on the receiving profit is raised to an effective 15% by an additional 10% top-up, with no exclusion or other adjustment. The marginal tax saving relative to a 25% source rate falls from 20 to 10 percentage points. The reduced incentive follows directly from the arithmetic. The example cannot establish actual liability because GloBE computations use specified income and tax definitions, exclusions and detailed administrative rules.

Who collects the top-up is an independent question. A domestic top-up tax permits the receiving jurisdiction to collect additional revenue in the simplified example. An equivalent collection by a parent jurisdiction changes the geographic allocation of revenue even if group tax is unchanged. Hence the tax minimum's effect on worldwide tax differs from its revenue distribution among countries. A relatively low tax jurisdiction can reassess its policies if the alternative is for another jurisdiction to collect the difference.

In January 2026, the Inclusive Framework agreed on a Side-by-Side package including simplified calculations, transitional relief, treatment of certain substance-linked incentives and safe harbors for qualifying parent jurisdictions. This is dated institutional context, not evidence that every country's legislation has identical effective dates and rules. The economic effect depends on coverage and domestic implementation as well as the formal coordination framework. Thresholds also leave some firms outside the relevant scope. Responses should be assessed for groups exposed to the rules and separately for others.

The OECD's July 2026 announcement of the 2026 economic impact assessment reports a projected decrease in global profit shifting of 22.6–44.6% under the assessed framework. This is a modeled range, presented alongside preliminary evidence from the first implementation year; it should not be treated as an observed global reduction of that size. The announcement also points to an expected narrowing of jurisdictional effective-rate differences. The underlying assessment was not fully retrieved for this dossier, so its model assumptions and estimates remain subject to further audit.

A credible causal evaluation must distinguish implementation dates, transitional arrangements, affected groups and the baseline policy environment. A comparison just above and below a revenue threshold can be informative when firms cannot precisely manipulate the determining variable and other policies do not change discontinuously at the same threshold. Anticipation, changes in reporting and multinational reorganization can compromise such comparisons. Short-run findings also may differ from long-run investment effects. Evaluating the global minimum tax requires specialized analysis beyond the treatment here, which focuses on how it changes the incentives and measurability of corporate profit allocation.

The new rules do not create a uniform tax on every unit of global corporate profit. Income definitions, substance-based exclusions, scope and continued rate differences above the minimum can preserve variations in marginal returns. Income allocation can also matter after a minimum tax where jurisdictional ownership of the tax receipt changes. The effect on the profit-shifting margin is thus a matter of degree and institutional operation, rather than an assumption that allocation incentives simply disappear.

Critical Synthesis and Research Design

The evidence supports an economic interpretation of profit shifting as a consequence of the relationship between integrated multinational business and territorially differentiated fiscal institutions. The multinational firm can change the geographical attribution of its earnings because its commercial activities generate returns jointly while accounting and fiscal systems recognize them in specific entities and territories. Observed tax sensitivity across firm-level methodologies and the distribution of profitability in macroeconomic evidence provide complementary support for the existence of this response.

A rigorous synthesis separates three levels of inference. The first is accounting: a change in the recognition of related-party income or expenses alters local profit and can leave group pre-tax income constant. The second is behavioral: variation in relative tax treatment can cause firms to alter what they report, conditional on the design of the study. The third is fiscal and macroeconomic: reconstructing alternative country distributions requires an allocation benchmark, a population and detailed assumptions about taxation and aggregate adjustments. These are analytical distinctions used to organize existing knowledge, not a new theoretical system.

Some evidence is stronger for particular questions than for a comprehensive global measure. Transaction-linked data can isolate a relation between observed prices and tax differences; they do not cover the entire multinational corporate universe. Affiliate-level econometrics can identify profit responsiveness; incomplete accounting coverage affects external validity. Foreign-affiliate totals can capture activity in otherwise underrepresented jurisdictions, but allocating a residual to specific source countries requires additional assumptions. Administrative reporting aggregates add useful breadth, although data suppression and changing coverage constrain longitudinal inference.

Consequently, an apparent conflict among empirical studies should first be diagnosed through their estimands. A regression coefficient with a different sign convention may describe the same behavioral direction. A lower estimated fraction of shifted foreign profit may follow from a broader denominator. A higher worldwide fiscal loss can occur when the amount of foreign profits increases with shifting intensity unchanged. Evidence about the formation of new entities may show a prolonged response even when existing affiliates adjust their reported income quickly. Such differences are scientifically productive when they are made explicit.

For future empirical research, the first step is to establish the precise observation unit: entity, enterprise group, jurisdictional subgroup, reporting country, or bilateral position. The second is the construction of the relevant income and tax concepts, including losses and intra-group income. The third is to specify and defend a counterfactual, with explicit treatment of sector, production factors and economic rents. The fourth is to identify variation in tax incentives and examine alternative explanations. The fifth is numerical reconciliation: the sum of inward and outward reallocations should match for a pure transfer, whereas a measured erosion of aggregate taxable income should be explained by tax or accounting asymmetries. The final step is fiscal interpretation and uncertainty analysis.

Data quality can be evaluated by examining the coverage of firms and jurisdictions, the consistency of reporting periods, the ability to eliminate duplicate income, and the treatment of unobserved destinations. Sensitivity analysis should vary the assumptions most consequential for the estimate. A macro reconstruction may vary normal profitability and destination weights. A micro analysis may vary its sample, tax differential, treatment of losses and control variables. A reform study should test alternatives for its comparison group, anticipation period and simultaneous policy changes.

The policy implication is that success cannot be measured exclusively by a decrease in profits recorded in an individual low-tax jurisdiction. Such a decrease can result from rerouting, a changed sectoral mix, losses, or a real contraction. More compelling evidence would demonstrate coordinated changes in the profit distribution, the tax burden, real activities and relevant administrative outcomes, while showing that competing explanations do not account for the results. It should also distinguish revenue accrued to source, recipient and parent jurisdictions.

The economic argument has an institutional implication. Corporate groups have access to a system of jurisdictions rather than merely individual tax rates. Policies affecting one link in the network can change the use of others, while country-specific enforcement and coordination affect the set of feasible arrangements. In evaluating reform, the proper unit of analysis is therefore both the national fiscal outcome and the multinational group's full distribution of income. This conclusion follows from the combined evidence and ordinary accounting reasoning; no named authorial doctrine or independently estimated structural model is asserted.

Limitations and Evidence Assessment

This research uses a structured critical literature synthesis and official-document review. It includes journal articles, working-paper records, an author-hosted research manuscript and OECD institutional documents. It is not a systematic-review protocol with exhaustive screening of every publication, and it does not replicate regressions or construct a new international panel dataset. The examples are hypothetical calculations chosen to clarify accounting and interpretation; none should be read as a country or firm estimate.

Source verification is claim-specific. Directly inspected materials include the April 2022 author manuscript of the foreign-affiliate study, the OECD country-by-country chapter, institutional report abstracts, the published article representation on developing-country micro-data and official minimum-tax documentation. Some journal landing pages returned access errors and were checked through indexed abstracts or bibliographic records. The 2026 historical working-paper text and the extractive-sector working-paper text were subsequently inspected at claim and method level; the full underlying July 2026 OECD economic assessment remains incomplete at document depth. No strong assertion depending on unexamined methods is made, and the unresolved access limitations are retained in the evidence record.

The literature covers multiple periods and populations. Its estimates should not be combined into an unsupported global average or presented as the current observed quantity. The 2026 OECD reporting statistics describe fiscal year 2023, while the 2026 policy assessment contains modeled future effects and early implementation evidence. The analysis is bounded by sources available on the research date; subsequent law, data revisions or publication changes may require reassessment. No external peer review is represented.

Further empirical work should inspect full source versions, reproduce selected macro allocations with archived inputs, and examine the distribution of tax receipts and activity before and after relevant reforms. Legal assessment of individual transactions and detailed jurisdiction-specific implementation belong to related research objects. The present study establishes the economic questions and compares the kinds of evidence available to address them.

Conclusion

Corporate profit shifting redistributes reported multinational earnings across jurisdictions in response to differentiated fiscal incentives and the organizational capacity of international enterprises. Its significance arises from the interaction between an integrated private production system and separately administered tax bases. The same organizational flexibility that supports international trade and investment can alter the geography of corporate income recognition.

Firm-level evidence identifies tax-sensitive profit responses; transaction-level studies observe specific channels; macroeconomic reconstructions estimate the geographical distribution of income under explicit benchmarks; administrative reporting describes the alignment of profits, taxes and observable activities. Their results support a coherent account of the phenomenon while addressing different questions. A scientific assessment preserves the distinctions among reported profit, shifted profit, taxable income, economic activity and fiscal revenue.

The author's analytical contribution is the integration of those measurement perspectives with a consistent accounting treatment of jurisdictional transfers and fiscal effects. The hypothetical examples establish why revenue losses in origin economies differ from global revenue losses and why a change in the assumed counterfactual alters the estimate. The comparison of empirical methods establishes why differences in denominators, coverage and identification must be resolved before estimating changes in the intensity of shifting.

International measures can alter the expected benefits and costs of profit allocation. Minimum-tax coordination narrows some differential incentives, while reporting, transaction attribution and interest limitation rules target other aspects of the problem. Evaluation requires distinguishing real investment responses, substitution across allocation channels, fiscal distribution and compliance burdens. An effective economic analysis proceeds from the observed outcomes to a transparent counterfactual and distinguishes projected changes from demonstrated effects.

The research dossier locates corporate profit shifting within Offshore Economics as a specific, independently identifiable scientific object: the distribution of multinational earnings under separate fiscal jurisdictions. It provides a basis for further research on intellectual property, internal finance, tax competition, fiscal capacity and the evolving architecture of international taxation while preserving each subject's distinct disciplinary scope.

References

Beer, Sebastian, and Dan Devlin. 2021. Is There Money on the Table? Evidence on the Magnitude of Profit Shifting in the Extractive Industries. IMF Working Paper 21/9. Extractive Industries Working Paper.

Beer, Sebastian, Ruud A. de Mooij, and Li Liu. 2018. International Corporate Tax Avoidance: A Review of the Channels, Magnitudes, and Blind Spots. IMF Working Paper 18/168. DOI: 10.5089/9781484363997.001. Journal version: Journal of Economic Surveys 34(3): 660–688 (2020). DOI: 10.1111/joes.12305. International Corporate Tax Avoidance.

Crivelli, Ernesto, Ruud A. de Mooij, and Michael Keen. 2015. Base Erosion, Profit Shifting and Developing Countries. IMF Working Paper 15/118. Base Erosion and Developing Countries.

Gidron, Idann, Ludvig Wier, and Gabriel Zucman. 2026. Global Profit Shifting, 1975–2023. Skatteforsk Working Paper 38; NBER Working Paper 30673 revised July 2026. Global Profit Shifting, 1975–2023.

Heckemeyer, Jost H., and Michael Overesch. 2017. Multinationals’ Profit Response to Tax Differentials: Effect Size and Shifting Channels. Canadian Journal of Economics 50(4): 965–994. DOI: 10.1111/caje.12283. Multinationals’ Profit Response.

Hines, James R., Jr., and Eric M. Rice. 1994. Fiscal Paradise: Foreign Tax Havens and American Business. Quarterly Journal of Economics 109(1): 149–182. DOI: 10.2307/2118431. Fiscal Paradise.

Huizinga, Harry, and Luc Laeven. 2008. International Profit Shifting within Multinationals: A Multi-Country Perspective. Journal of Public Economics 92(5–6): 1164–1182. International Profit Shifting.

Johannesen, Niels, Thomas Tørsløv, and Ludvig Wier. 2020. Are Less Developed Countries More Exposed to Multinational Tax Avoidance? Method and Evidence from Micro-Data. World Bank Economic Review 34(3): 790–809. DOI: 10.1093/wber/lhz002. Developing-Country Micro-Data.

Liu, Li, Tim Schmidt-Eisenlohr, and Dongxian Guo. 2020. International Transfer Pricing and Tax Avoidance: Evidence from Linked Trade-Tax Statistics in the United Kingdom. Review of Economics and Statistics 102(4): 766–778. Earlier version: Federal Reserve International Finance Discussion Papers, 2017. International Transfer Pricing.

OECD. 2015. Measuring and Monitoring BEPS, Action 11: 2015 Final Report. Paris: OECD Publishing. DOI: 10.1787/9789264241343-en. Measuring and Monitoring BEPS.

OECD. 2016. Limiting Base Erosion Involving Interest Deductions and Other Financial Payments, Action 4: 2016 Update. Paris: OECD Publishing. DOI: 10.1787/9789264268333-en. BEPS Action 4.

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

OECD. 2026. Corporate Tax Statistics 2026: Country-by-Country Reporting Statistics. Paris: OECD Publishing. Country-by-Country Reporting Statistics.

OECD. 2026. Global Minimum Tax. OECD topic overview, including January 2026 Side-by-Side package. Accessed November 18, 2026. Global Minimum Tax.

OECD. 2026. OECD Publishes New Analysis on the Economic Impacts of the Global Minimum Tax. Announcement, July 15, 2026. Economic Impacts of the Global Minimum Tax.

Tørsløv, Thomas, Ludvig Wier, and Gabriel Zucman. 2023. The Missing Profits of Nations. Review of Economic Studies 90(3): 1499–1534. DOI: 10.1093/restud/rdac049. Author manuscript dated April 8, 2022. The Missing Profits of Nations.

Wier, Ludvig, and Gabriel Zucman. 2022. Global Profit Shifting, 1975–2019. WIDER Working Paper 121/2022. Helsinki: UNU-WIDER. DOI: 10.35188/UNU-WIDER/2022/254-6. Global Profit Shifting, 1975–2019.