RESEARCH DOSSIER

Treaty Shopping and Investment Flows

Abstract. This monograph examines the economic consequences of tax treaty shopping for the direction, ownership, measurement, and fiscal incidence of foreign direct investment. It distinguishes additional productive investment from financial reorganization and separates treaty-induced changes in immediate-counterparty statistics from changes in ultimate control. The research integrates official tax treaty and direct investment frameworks with published studies of treaty networks, Dutch special purpose entities, investment responses, source-country revenue, and a historical Indian treaty reform. It evaluates the distinct channels of tax relief, legal certainty, ownership routing and investment protection, and the relationship between anti-abuse provisions and minimum taxation. The analytical contribution is a reconciliation of apparently disparate empirical findings through explicit treatment, outcome, counterfactual and evidence distinctions. Worked examples illustrate how changes in ownership and fiscal incidence can occur without an equivalent increase in physical capital. The study concludes that treaty policy should be assessed against productive outcomes, fiscal receipts, legal predictability and social value rather than recorded bilateral FDI alone. The evidence is documentary and comparative; the manuscript does not report a newly estimated causal effect or a new named theoretical system.

Keywords: Tax treaties; treaty shopping; foreign direct investment; withholding tax; offshore economics; treaty networks; financial intermediation; ultimate investor; tax revenue; anti-abuse; international public finance

Introduction and research object

Foreign direct investment is an economic relationship expressed through institutional ownership, financial claims, and international production. A treaty changes the environment in which those relationships are organized. It may alter the prospective return on a new production facility, the cost of operating one already in existence, the legal location of a shareholder, and the tax treatment of earnings sent across a border. These effects need separate measurement. The central question of this monograph is how access to treaty benefits changes the allocation, composition, recorded origin, and fiscal consequences of international investment.

For the purposes of the economic inquiry, treaty shopping means the pursuit of a benefit under an international tax treaty through an arrangement involving an intermediary or other connection to a contracting jurisdiction where the benefit would otherwise be unavailable. Its economic implications depend on the counterfactual arrangement and the nature of the benefit. A third-country holding company may perform substantial management and financing functions, may hold securities without substantial operating activities, or may combine both functions. Such configurations require evidence of their activities and tax treatment before they can be classified. Legal entitlement to any actual benefit is a separate question.

Double taxation treaties coordinate jurisdiction over income and capital. International investment agreements primarily address investment relations and investor protection. Their separate but interacting objectives are explained in UNCTAD's 2024 investment-policy guide. This distinction is indispensable because protection from adverse state conduct and relief from overlapping taxation have different economic mechanisms, legal instruments, and measurable outcomes. Where this work refers to treaty shopping without a qualifier, it concerns tax treaty access.

The research adopts the perspective of Offshore Economics. Its object is the relation between the international organization of ownership, capital deployment, and fiscal incidence. It draws on international public finance and international economics and treats legal instruments as the independently documented institutional environment. It does not adjudicate entitlement, attribute motives to named taxpayers, or establish anyone's tax liability.

A successful treaty can improve the functioning of a bilateral economic relationship even when the immediate FDI statistics show little movement. Conversely, a treaty can be statistically associated with a large recorded inflow while leaving local production nearly unchanged. Policy evaluation should therefore proceed from separately observed financial, operational, and fiscal outcomes. This proposition is pursued through institutional analysis, evidence appraisal, worked accounting examples, and a synthesis of published research.

Treaties as economic institutions

Tax treaties arise in an environment of intersecting tax claims. A source jurisdiction may tax income generated within its territory; a residence jurisdiction may also tax the recipient or the enterprise receiving that income. Differences in tax bases, entity classifications, timing, and relief can create burdens beyond either jurisdiction's isolated tax schedule. Bilateral coordination can allocate taxing rights, limit withholding, and facilitate the resolution of disputes. The economic value of coordination depends on what would occur under domestic law if the treaty were absent.

The relevant benchmark is therefore a complete counterfactual tax and compliance environment. Consider two economies with comparable investment opportunities. In one, domestic relief already eliminates a large part of juridical double taxation. A treaty may change only a narrow class of transactions or reduce uncertainty. In another, overlapping taxation may make a marginal investment unprofitable before treaty relief. Equal treaty language can generate different changes in after-tax returns. Comparing reductions in statutory maxima without considering their practical incidence will overstate some treatments and understate others.

Tax treaties also affect uncertainty. A firm evaluates an investment in the presence of risks concerning the characterization of income, permanent establishment exposure, valuation, and conflicting assessments. A clearer division of jurisdiction and access to dispute resolution can improve the expected value of a project independently of a reduction in a dividend withholding rate. Investment responses may arise gradually as firms learn whether a treaty is administered predictably. This is one reason to distinguish the date of signature, legal entry into force, the date a provision begins to apply, and the date the affected behavior is observed.

The institutional significance of anti-abuse rules follows from these legitimate purposes. OECD's treaty abuse guidance records the development of beneficial ownership language in the 1977 Model and the later Action 6 minimum standard. The treaty network historically developed through numerous bilateral bargains; the economic consequences of overlapping agreements created a need for coordinated safeguards.

A state may value direct trade and productive investment, revenue stability, reciprocity, legal predictability, and the treatment of its own residents abroad. Concessions often affect more than one class of income and the parties may value them differently. A capital importer evaluating a reduction in source taxation may expect additional production or improved financing conditions, while a capital exporter values relief from foreign tax and protection of its residents. The relative weight of these interests changes with bilateral and third-country ownership patterns.

A treaty is consequently a multidimensional institutional change. The same legal text may simultaneously affect the marginal cost of capital, the distribution of fiscal rights, compliance certainty, information sharing, and the attractiveness of operating through a particular ownership structure. Treating a treaty as a binary condition in an economic regression approximates only part of this change.

Investment statistics and the economic object

The statistical concept of direct investment is anchored in a cross-border relationship of influence over an enterprise. The OECD's 2025 direct investment framework retains the threshold of at least 10 percent of voting power for establishing a direct investment relationship; chains of control and influence extend the relevant enterprise relationships. A position records a financial stake at a point in time. A transaction records a flow during a period. These quantities answer different questions from employment, output, capital formation, and infrastructure expenditure.

An acquisition of an existing domestic company by a foreign enterprise can qualify as FDI without immediately increasing the country's stock of produced assets. Conversely, an affiliate can expand production by retaining earnings, borrowing locally, or using existing financial resources. Its capital expenditure may differ materially from financial inflows from the foreign shareholder. A financing transaction can be important for the company even if the funds are not applied to construction or machinery in the same period.

Direct investment positions change through financial transactions and other changes, including valuation adjustments. Let P(t) denote a position, F(t) the financial transactions consistent with its accounting presentation during a period, V(t) valuation effects, and O(t) other changes. The accounting relationship is P(t) = P(t−1) + F(t) + V(t) + O(t). Observing a change in P permits no unique inference about F, much less about physical investment. Currency appreciation, enterprise valuations, debt restructuring, and reorganization can affect the measured position. A researcher must obtain reconciling information or acknowledge that the change remains unexplained.

The geographical attribution of a position introduces a further distinction. According to the OECD guidance on geographic disaggregation (2025b), ultimate investing economy statistics can allocate inward positions through controlling relationships. The winner-takes-all and proportional ownership approaches may yield different allocations. Immediate counterpart information remains valuable for assessing bilateral exposures and financial administration, while ultimate attribution helps locate control. Neither geography alone establishes where the initial financial resources were raised, who bears all the risks, or who benefits from each particular payment.

For a simple ownership chain, imagine that enterprise A in economy A controls holding company B in economy B and operating company C in economy C is wholly owned by B. If A finances B with 100 and B acquires a newly issued equity stake of 100 in C, the two cross-border positions can each be 100 under the assumed simplified equity accounting. C's inward position is 100 attributed immediately to B. Under ultimate attribution it is assigned to A in this simple case. A gross sum of inward positions across the two recipient economies reaches 200 even though only 100 enters the final operating affiliate. The sum reflects two financial relationships. The change in the affiliate's productive capacity remains a separate observation.

This example is a statistical illustration with no treaty rates, real taxpayer, or legal conclusion. Its significance is that an additional layer in an ownership chain can create measurable cross-border financial relations without a one-for-one increase in productive assets. The nationality of a direct investor in aggregate data cannot by itself identify the investor that initiated the economic decision. Knowledge of the chain improves geographical interpretation but cannot replace observations of the enterprise's physical activity.

Channels between treaty access and investment

Treaty access affects an enterprise through the expected value of its activities. A prospective project has expected operating receipts and costs, investment expenditure, financing obligations, tax payments, and a distribution of future states of the world. The investment decision is sensitive to changes in after-tax value around the margin of acceptance. Treaty provisions that lower genuine double taxation or reduce uncertainty can raise that value and permit otherwise unattractive investments to proceed.

There are at least three empirically different responses. In the first, a treaty influences the establishment or expansion of productive activity. In the second, it changes the investor or financing route associated with an existing activity. In the third, it changes the fiscal treatment of income earned from an existing activity while the ownership and production structure remain relatively stable. Responses can coexist within a firm and within an aggregate series. The analyst's objective is to separate the consequences with evidence appropriate to each.

Investment location is also affected by market access, local skills, infrastructure, exchange rates, the cost of energy and finance, supplier relationships, and regulatory conditions. An unusually favorable treatment of cross-border income may influence financing or ownership without overcoming an unfavorable production environment. A capital-intensive plant with market-specific assets may be substantially less mobile than the financial claims held against it. The enterprise's holding arrangements can change more quickly than the physical production network.

Some treaty effects arise exclusively through the division of tax collections. A reduction in source tax can increase the amount received by an investor, be offset by tax in another country, or result in a mixture depending on applicable relief, tax rates, and other circumstances. The source government can record less revenue even where the overall group tax burden changes little. An economic assessment of treaty access must follow fiscal incidence across the relevant jurisdictions, identifying which government collects the tax and which private participants ultimately benefit.

The administrative cost of obtaining and maintaining treaty relief also affects expected net returns. Documentation requirements, professional expenses, uncertainty about eligibility, and possible dispute costs are resources consumed by the enterprise or the administration. The economic availability of a benefit is not synonymous with the statutory wording. A network study that represents an agreement as a favorable numerical rate therefore measures a potential incentive; actual entitlement and net returns depend on additional information.

The adjustment process may be asymmetric. Firms with a mature international organization have established foreign affiliates, compliance professionals, and internal capital markets. A smaller enterprise entering a new market may lack these resources. Identical treaty rules can create different effective costs for each group. Administrative simplification and predictable interpretation may have relatively pronounced effects for some entrants, while established firms can more readily restructure existing financial relations.

These considerations explain why the effect of a treaty on the amount of investment cannot be obtained from a description of its tax provisions alone. A change in expected returns is a mechanism. Its empirical magnitude depends on how many projects or transactions lie sufficiently close to a decision margin, what alternative structures already exist, and how the reform is implemented.

The network character of bilateral taxation

A collection of bilateral tax treaties constitutes a network because companies may have relationships extending across several contracting jurisdictions. The same concession can have different consequences depending on the alternative routes or ownership arrangements already available. An evaluation confined to the attributes of a single country pair can misclassify the economic relevance of a new treaty.

In Optimal tax routing (2018), van 't Riet and Lejour represent corporate income taxation, withholding, double taxation agreements, and relief within a network. Their abstract reports a potential reduction in dividend repatriation taxation of about six percentage points on average and associations between bilateral positions and network centrality. The result describes a modeled opportunity and empirical relationship, with the assumptions and historical inputs of the study. It should not be read as a realized average tax saving for all multinational firms.

Hong's network approach to FDI (2018) studies a 70-country network and reports an average difference of 3.66 percentage points between direct and tax-minimizing modeled routes. The study also reports a positive association between the existence of a favorable direct route and the inward FDI stock. These findings use a different network implementation and outcome from the preceding study. Their quantitative estimates cannot be pooled without harmonizing the tax measure, samples, and weights.

Network relevance concerns marginal changes in the system. Petkova, Stasio, and Zagler (2020), in On the relevance of double tax treaties, distinguish treaties that improve direct tax distance relative to domestic law and the existing network. Their abstract reports an estimated increase in FDI of approximately 18 percent for relevant treaties. This estimate is specific to the authors' definition and econometric analysis; it is neither a universal treaty premium nor proof of an equivalent increase in production.

These studies make a shared methodological point: an agreement's economic treatment can depend on third-country alternatives. A new bilateral concession can change the relative attractiveness of existing direct and indirect relationships; changes elsewhere in the network can alter the value of an unchanged bilateral agreement. Empirical comparisons should therefore specify a network version at each date, including applicable treaty and domestic provisions, rather than applying a contemporary network retrospectively.

Network exposure also has distributional implications. Jurisdictions possessing extensive commercially useful treaty relations can serve as sites for corporate administration and cross-border intermediation. Their legal and financial service sectors may benefit from this role. At the same time, the presence of a favorable tax network can generate large balance-sheet positions whose size exceeds the locally employed labor and capital associated with them. These are different economic quantities. Aggregate centrality cannot identify the commercial or fiscal purpose of an individual entity.

The author's interpretation is that bilateralism changes the marginal value of other bilateral relations. Consequently, the economically relevant unit of analysis is sometimes a particular host-investor pair within a larger system of potential relationships, rather than the isolated treaty. This interpretation does not add a named theory to the existing literature; it organizes the evidential implications of the cited studies.

Financial routing, special purpose entities, and phantom FDI

The Netherlands provides an empirical setting in which tax treaty relationships and financial intermediation can be examined. Weyzig's study of Dutch FDI diversion (2013) uses microdata on special purpose entities. The author reports that tax treaty connections, including reductions in dividend withholding tax, help explain the geography of routed investment. The data and regressions concern the study's historical Dutch SPE population; the result cannot determine the motives or treaty entitlement of all entities located in the jurisdiction.

Special purpose entities are statistical and institutional objects; their activities vary and definitions should track the data compiler. A holding or financing entity may employ a small number of people while holding substantial financial assets or liabilities. These accounts can record genuine legal claims. Their relationship with local production is an empirical matter. A holding company's assets can support real investment in a third economy even if the intermediary's local employment is modest.

Damgaard, Elkjaer, and Johannesen (2019) estimate a global network in What Is Real and What Is Not in the Global FDI Network?. Their historical reconstruction separates positions associated with shell entities from positions linked to more substantial economic activity and reallocates the latter toward ultimate investors. The authors estimate that phantom investment may represent almost 40 percent of global FDI positions in their studied historical setting. This is a modeled global stock estimate, not a count of illicit transactions, an estimate of treaty shopping alone, or a current percentage for 2026.

Removing intermediary positions and identifying ultimate locations also address different measurement issues. An analyst can exclude a domestic SPE from one inward series while retaining the corresponding operating affiliate's position in another country's accounts. Reallocation by ultimate investor instead changes the geographic label of a position without necessarily changing the host total. Applying both transformations requires a consistent definition, a known coverage denominator, and transparent treatment of unmatched entities.

Consider an operating affiliate with an existing equity position of 100. Assume its foreign owner transfers the entire ownership interest to another foreign entity under the same ultimate controlling parent and its plants, employees, equity value and outstanding financial claims remain unchanged. After the transfer, the immediate partner may change from A to B. The host's ultimate attribution may remain A. Production and the host's total inward equity position remain unchanged. A statistical compiler would need transaction and other-change information to determine the appropriate entries for the actual transfer. A mere comparison of partner-country positions cannot settle that classification.

The economic distinction has implications for research on the offshore system. Large financial positions can reflect corporate organizational scale and cross-border contracts. Their geographic distribution can be influenced by treaties and other institutional features. A country's productive contribution, financial exposure, and tax receipts must each be investigated on their own evidence. Labeling the whole quantity as either productive investment or abuse obscures the underlying relationships.

The empirical literature and heterogeneous findings

An empirical study of treaty effects must identify the relevant outcome. Bilateral direct investment positions, financial flows, affiliate sales, new-establishment counts, tangible assets and tax receipts differ in their responses to treaty formation. A coefficient on one outcome is not directly transferable to another. The presence of different estimates in the literature can reflect differences in treatment and outcome definitions rather than a substantive contradiction.

Blonigen and Davies (2002), in Do bilateral tax treaties promote foreign direct investment?, examine OECD-country data for 1982–1992. The working-paper abstract reports that recent treaty formation does not promote new investment in their analysis and that some specifications suggest negative effects. This early evidence is particularly relevant to claims that treaty signing itself necessarily creates FDI. It belongs to a specific sample and vintage of institutions.

Using firm-level data from the US Bureau of Economic Analysis, Blonigen, Oldenski, and Sly (2014) report a different result in The differential effects of bilateral tax treaties. Firms using differentiated inputs show a larger positive investment response in their study. The authors relate that heterogeneity to the complexity of establishing arm's-length values and the benefits of access to government assistance when tax disputes arise. They report effects involving existing affiliate sales as well as entry. The implication is that institutional certainty can have observable value alongside tax-rate changes.

The network studies discussed above examine a further dimension: the importance of the available direct and indirect tax relationships. Their outcomes include measured bilateral positions, and their results inform the geography of investment claims. Where those studies report greater bilateral investment associated with a favorable route, researchers still require operational evidence to identify how much of the response represents additional production.

Several conditions reconcile these results. A treaty may change tax certainty substantially for a subset of firms but produce limited variation in aggregate investment. A treaty may alter the ownership route without changing total investment in the host. A treaty's relevant effect may depend on the wider network and existing domestic law. Time periods also differ in tax enforcement, reporting standards, and the scale of multinational structures.

The comparison is a qualitative synthesis, not a statistical meta-analysis. The primary research was consulted at different access levels, with much of the peer-reviewed literature limited to original abstracts and bibliographic details. Numerical findings are therefore reported only at the scope explicitly supported by those records. No independent coefficient reconstruction, standard-error extraction, sample harmonization, or publication-bias correction has been performed for the DTT literature in this monograph.

An additional source of disagreement is treatment selection. Governments are likely to negotiate treaties with existing trading partners or with countries anticipated to become important investors. Firms' investment expectations can therefore influence the likelihood of a treaty. A positive correlation need not be wholly attributable to the agreement. Likewise, a negative correlation can reflect a policy response to an existing deterioration in the relationship or efforts to reduce undesirable income routing. Identification requires carefully specified counterfactual behavior.

Empirical research should recognize differences among three questions: whether a treaty promotes aggregate foreign investment in the host economy; whether the treaty redirects existing or planned positions through the signatory's entities; and whether treaty access improves economic welfare after accounting for taxes and costs. The observed coefficient relevant to one question is evidence about the others only when further relationships are established.

Developing economies and the public revenue trade-off

The distribution of taxing rights is particularly significant when the recipients of foreign investment depend on revenue generated within their territories. An investment treaty partner's residence status may be a poor predictor of the ultimate control of firms earning income in the host. Negotiation based on direct bilateral investment patterns can then overlook the exposure of source revenue to third-country ownership relationships.

Beer and Loeprick (2021), in Too high a price?, study 41 African economies across 1985–2015. Their published abstract reports that agreements with investment hubs are not associated with additional investment in their sample, while revenue reductions are observed; they also provide indirect evidence of treaty shopping through the responses of nominal investment. The result concerns a regional historical sample and the authors' estimation strategy. It does not establish that all agreements with investment hubs reduce welfare or that the estimated losses are attributable to a single legal provision.

Revenue policy requires separating the extensive margin of investment from concessions on an existing tax base. A tax reduction on future profits may induce some projects that would otherwise not proceed. It may also reduce tax collected on payments associated with investments which would have occurred without the concession. The additional projects and the unaffected existing investments require different counterfactuals.

A deliberately simplified fiscal example illustrates the distinction. Suppose a host initially collects 10 on existing cross-border income and the investment policy reduces those receipts to 6. Suppose additional activity attributable to the policy creates 3 in other net tax revenue. The observed revenue change is −1 before administration costs and any other changes. If new activity instead yields 5, the revenue change is +1. This arithmetic is independent of the identity of particular jurisdictions or the provisions supporting any treaty claim. It illustrates the need to measure the existing-base concession and incremental-base response separately.

The example also makes clear that a revenue-neutral outcome is not necessarily the economic optimum. Additional investment can produce wages, consumer benefits, productivity spillovers, environmental costs or congestion. The incidence of these consequences depends on labor market conditions, public services, domestic ownership, and the use of fiscal revenues. A state may place different weights on public revenue, employment, and distributional outcomes. Policy assessment must identify those objectives and avoid treating a higher FDI stock as a complete welfare measure.

For administrations with limited resources, a complex and uncertain tax environment can make enforcement costly. A broad treaty concession may be easy to administer at the point of payment but costly in foregone revenue; a more targeted regime may retain revenue while requiring more documentation. The evaluation must encompass both private compliance and public administrative capacity. Empirical work should report actual revenue collected and applicable tax measures separately from modeled potential tax losses.

The fiscal exposure of a country is not determined solely by its average statutory rate or the number of treaties it has concluded. It depends on the composition of income payments, ownership chains, relative treatment under alternative agreements, treaty enforcement, and the concentration of multinational production. A single globally central agreement can have a different marginal effect from numerous agreements with little relevant activity.

India and Mauritius as a historical reform episode

A historical reform of the India–Mauritius tax treaty illustrates the importance of dates, income category, and transitional protection. The Indian Ministry of Finance's August 2016 protocol notification records signature on May 10, entry into force on July 19, and official notification in August. The announcement describes a shift toward source taxation of capital gains on shares in Indian resident companies acquired on or after April 1, 2017, grandfathering of prior investments, and a transition period ending March 31, 2019 in which a reduced rate was subject to eligibility conditions.

This episode concerns a specified category of capital gains and affected acquisition cohorts. It should not be summarized as a uniform withdrawal of every tax treaty benefit on a single date. The reform changed the economic environment for investments acquired in different periods. Investors could anticipate the change after its announcement, while legacy holdings continued under transitional or grandfathered arrangements. Thus the relevant treatment may depend on whether an investment was already held, newly made, or subsequently disposed of.

The episode provides a useful prospective empirical design. A researcher could construct acquisition-cohort series and compare affected holdings with suitable unaffected investments or outcomes before and after the announcement and effective dates. Data on immediate and ultimate shareholders would improve interpretation. Evidence on establishments, acquisitions, divestitures, and capital formation would help distinguish changes in portfolio composition, ownership residence, and real investment. The design would also need to account for other contemporaneous regulatory and economic changes.

Anticipation is especially important. Transactions can be accelerated or delayed around a known legal date. A temporary increase in gross recorded FDI before a reform is not necessarily evidence of a lasting increase in production. A reduction in one source country's recorded shares after the reform may reflect substitution toward other immediate investor jurisdictions. An appropriate analysis would examine both the host's total investment and the distribution across counterparty jurisdictions, then test whether the country composition changes without similar shifts in establishment or output measures.

The historical government release is sufficient to verify the chronology and stated policy design used here. It does not establish the size of any behavioral response. No post-reform causal estimate is presented in the absence of matched transaction data. The subsequent legal development of the treaty and domestic legislation has not been comprehensively updated through 2026; the episode is treated as a bounded historical change, not a current investment or tax recommendation.

Anti-abuse rules, uncertainty, and compliance

The institutional response to treaty shopping changes the economic value of treaty access by changing the set of eligible arrangements and the certainty of entitlement. The OECD/G20 project on base erosion and profit shifting placed treaty abuse within a minimum standard involving a common treaty purpose and alternative substantive anti-abuse mechanisms.

Article 7 of the 2016 Multilateral Convention provides a principal purposes test under which a benefit is denied when obtaining it was among the principal purposes of an arrangement, as assessed in light of the relevant facts, unless granting relief is consistent with the purpose of the applicable treaty provisions. The precise text, covered agreement, reservations, notifications and effective dates are essential to establishing applicability. Ratification of the multilateral instrument by one country does not by itself modify every bilateral treaty.

Economically, rules based on entitlement and commercial circumstances change the relationship between the headline treaty network and the effective network accessible to a given firm. A model based only on statutory withholding maxima will tend to miss this change. A researcher investigating anti-abuse reform should therefore record both the rates and the enforced eligibility rules and distinguish treaty amendments from domestic implementation and administrative practice.

An anti-abuse rule may influence existing organizational arrangements, planned investments and compliance expenses. Some previously observed financial relationships may disappear or move; commercially substantial investment may continue. A change in measured foreign investment following an anti-abuse reform consequently has several plausible channels. Identifying them requires a decomposition of financial links, enterprise activity and fiscal receipts rather than an assumption that all formerly routed capital was artificial or that all recorded reductions are productive losses.

Legal clarity is economically valuable. If eligibility becomes more predictable, an enterprise can estimate prospective costs with greater accuracy. If uncertainty rises, some businesses may incur additional compliance costs or defer a project until treatment is clear. These effects differ across firms and across administrative systems, and can coexist with a reduction in opportunities for inappropriate treaty claims.

Public authorities also face competing demands. They need to evaluate the factual basis of claims, protect legitimate treaty rights, prevent unjustified concessions and maintain coherent administration. Reforms designed solely to maximize denied claims or minimize measured financial flows may produce poor policy metrics. The appropriate objective includes the integrity of the tax base, fair and predictable treatment, and the welfare contribution of actual economic activity.

For empirical purposes, implementation intensity is not identical to the text of a treaty. An analyst should seek observable administrative information, where lawful and available, such as aggregate disputes, withholding relief granted, treaty claims processed, the timing of regulatory changes and compliance costs. Confidential case information is not necessary for aggregate evaluation and should be handled under appropriate disclosure safeguards.

Investment protection and nationality restructuring

International investment agreements provide a separate route through which the organization of foreign ownership can influence investment incentives. Protection against certain forms of adverse state conduct and access to applicable dispute settlement can alter perceived political risks and the expected distribution of investment returns. The relevant definitions and the existence of consent determine the legal protections; tax treaty residence and investment treaty nationality must not be assumed to coincide.

For the economic analyst, expected protection may affect a risk premium and the willingness to commit capital. It may also alter the legal structure of foreign ownership around existing assets. Changes of nationality in the corporate chain can change the treaty framework associated with an investment even when the operating enterprise's assets remain in the same country. Evidence about investment treaty protection is thus relevant to the general question of routing, but remains distinct from evidence about tax treaty concessions.

Brada, Drabek, and Iwasaki (2021) present a meta-analysis of investor protection based on 2,107 estimates from 74 studies. Their abstract describes an average effect small enough to be regarded as negligible and reports publication-selection concerns, while recognizing that research designs may lack power to detect underlying effects and that origin-country differences exist. These results apply to investment-protection agreements; they cannot be substituted for estimates of the effects of double taxation treaties.

A common methodological challenge is the attribution of observed changes to the new treaty rather than prior investment expectations. Investor protection and tax coordination can also change in the same period as trade arrangements, investment promotion policies, and domestic reforms. Classifying agreements and outcomes independently helps avoid combining distinct mechanisms into a single coefficient.

The economic comparison supports a broader governance point. Policymakers need a consistent view of the fiscal, investment-protection, and regulatory environment within which firms take decisions. One treaty may have effects on public policy already constrained or enabled by other instruments. Detailed analysis of investor nationality doctrines, tribunal jurisdiction, abuse of rights and the legal boundaries of restructuring is assigned to the corresponding legal research. This monograph restricts its conclusion to the expected-return, risk, ownership and measurement channels.

Identification, evidence, and research design

An empirical assessment begins by specifying what effect is to be estimated. Suppose Y is a host economy's productive outcome. Possible outcomes include employment by foreign affiliates, gross fixed capital formation in foreign-controlled establishments, establishment counts, and value added. A separate measure M can describe immediate bilateral financial positions, and a separate measure T can describe source tax received. A single causal estimand cannot stand in for changes in all three.

A panel study can estimate associations between treaty events and these outcomes with country-pair or enterprise controls, date effects and measured economic conditions. Stable differences between countries can be addressed with fixed effects. Common shocks can be addressed using appropriately specified time effects. Nevertheless, the assumption that the remaining variation in treaty adoption is exogenous requires substantive defense. Evidence of pre-existing differential trends, policy anticipation, concurrent reforms and changes in data coverage should be disclosed and tested.

An event study around a dated treaty change can display how outcomes evolve before and after the intervention. A causal interpretation depends on a credible comparison group, the treatment timing, the absence or accounting of other relevant shocks and appropriate statistical inference. Multiple treated cohorts may require methods allowing heterogeneous effects. Researchers should avoid conditioning on a post-treatment ownership structure as though it were an unaffected control; the structure itself can be a mechanism of the treatment.

Network spillovers challenge the use of unaffected country pairs as controls. A reform in one treaty may influence entities in other jurisdictions and substitute flows toward alternative partners. Control observations in the same network can therefore be indirectly exposed. A study should state its exposure mapping and distinguish a direct effect on the treated relationship from substitution within the wider system. Aggregate host outcomes can provide an additional check on whether bilateral redistribution corresponds to a total increase.

FDI flows may include negative values and observations at zero, for example in the presence of divestiture and repayment. Mechanical logarithmic transformation can discard informative observations. Any chosen estimation method must be appropriate to the signed outcome and its distribution. Researchers should document exchange-rate conversion, valuation bases, changes in statistical coverage, missing data, and whether figures use a directional or asset-liability presentation.

Microdata offer opportunities to link an affiliate's ownership change to its investment expenditure, employment or sales. Cross-border matching introduces entity-identifier changes, name ambiguity, incomplete coverage and confidentiality restrictions. Economic membership in a corporate group should be determined using a stable and documented rule. Parent ownership can change during the observation period; assigning all historical observations to a contemporary parent misstates the geography of earlier decisions.

A practical evidence design would maintain four linked but distinguishable datasets: treaty instruments and their applicability dates; ownership and financial position records; operating-enterprise measures; and detailed aggregate tax receipts by income category. Where individual data cannot be linked, matching aggregate strata by industry, cohort and size can support more limited inference. Reporting missingness and coverage is crucial.

The original analysis in this monograph does not pretend that such a joined dataset has been obtained. Its conclusions are based on documented legal and institutional mechanisms, statistical concepts, comparative research findings and internally consistent accounting illustrations. This supports qualitative judgments about interpretation and research design. Quantitatively attributing observed global changes to treaty shopping remains a separate empirical undertaking.

Interpreting outcomes and fiscal incidence

A reform can lead to a variety of observable patterns. If the immediate investor's recorded position changes but host total positions and production measures remain stable, the evidence is consistent with geographical reallocation of ownership. Further verification of transaction classification and control is needed. If production measures rise after a reform while both the immediate and ultimate investor's positions rise, the evidence is consistent with expansion; causal attribution still depends on a counterfactual. If tax receipts change while investment and production remain stable, the reform may primarily have a distributional fiscal effect.

Observed change Plausible economic interpretation Additional evidence required
Immediate partner changes; ultimate parent stable Ownership or financing reorganization Enterprise-level ownership and transaction reconciliation
Host inward position increases; employment stable Acquisition, financing, valuation, or capital-intensive activity Transaction type, capital expenditure, prices and output
New establishments and operating expenditure rise Potential real investment response Credible treatment comparison and operating histories
Withholding collected changes; payment base stable Shift in effective tax incidence Applicable rules, exemptions and collection data
One bilateral position falls; others rise Geographic substitution Complete host totals, ultimate investor and network exposure

Each row states a possible interpretation rather than a definitive diagnosis. The evidence requirements prevent the inference of an unobserved mechanism from the sign of a single statistic. For instance, stable employment is consistent with capital-intensive productive investment and cannot independently establish a lack of substance.

A parallel fiscal accounting exercise separates the immediate collection effect from induced behavior. The change in host revenue can be written as the change in taxes on the prior base plus the change arising from altered activity and its composition plus relevant administration and timing adjustments. When these components interact, the quantitative decomposition must define an ordering or explicitly account for interactions. Numerical estimates from static tax-rate models need not capture the dynamic component.

The distribution of gains from a treaty concession is an incidence problem. The immediate recipient of relief may be the legal taxpayer, while the economic gain can be shared among shareholders, borrowers, lenders, customers and workers through changes in returns and prices. The sharing depends on market structures, bargaining and elasticities. A host government may forego a unit of tax without producing an equal increase in domestic wages or investment. Measuring those outcomes requires evidence beyond the statutory rate.

The social value of investment must also reflect project quality. Domestic spillovers, environmental consequences, local procurement, wage conditions, and infrastructure requirements can change the desirability of projects with identical financial values. Tax revenue may finance public capital and services that influence subsequent investment attractiveness. Accordingly, a fiscal concession should be evaluated in a longer-run policy context rather than as an immediate exchange between a reported investment total and a revenue amount.

These considerations support a comparative standard of analysis: evaluate real activity, financial organization, and public fiscal outcomes in parallel, then explain their consistency or divergence using the institutional evidence. The standard is a methodological conclusion drawn from existing economics and accounting, not a claim to a newly invented theory.

Treaty negotiation and evaluation

Treaty policy should begin with the purpose of the agreement and the conditions under which its benefits will be available. Where the policy objective is to remove a demonstrable obstacle to productive investment, the public evaluation should identify the affected sector, tax mechanism, and plausible extent of the impediment. A concession which primarily changes the intermediary jurisdiction cannot automatically be credited with the same benefits as new productive activity.

Ex ante evaluation can compare existing domestic relief with the contemplated agreement and examine the network exposure of the source tax base. The analysis should document payment composition and ownership geography, and consider sensitivity to different behavioral responses. Estimates based on hypothetical full use of eligible tax relief can provide a stress test; the distinction between modeled exposure and actual projected revenue is critical. A treaty may also bring administrative and legal-certainty benefits which are difficult to express as a percentage of financial inflows.

Ex post evaluation should assess the exact effective changes and compare economic outcomes against an appropriate counterfactual. A transparent dataset should include the time of announcement, signature, entry into force and provisions' application dates, and identify transitional cohorts. Public reporting should state whether reported FDI is a position or flow, whether special purpose entities are separately reported, and how partner-country attribution is performed.

The evaluation should include the effect on the tax base of the host and the location of final collections. Where domestic statistical capacity is limited, a combination of aggregate payment records and affiliate activity indicators can provide a more credible basis for policy than a single total for inward FDI. Data availability can be improved through cooperation between investment agencies, tax administrations, central banks and statistical offices under appropriate privacy safeguards.

International coordination matters because unilateral treaty reform can move some financial relationships into other parts of the system. A policy review should consider substitute treaty relationships in a non-operational, aggregate risk assessment. The objective is accurate treaty design and compliance, not instruction in obtaining particular tax benefits. Aggregate estimates of network exposure can inform negotiations and identify areas in which further legal analysis is necessary.

Policy publication should also distinguish the expected costs from the consequences actually observed. Announced benefits and anticipated investment gains should not be represented as realized returns. Project reviews should be subject to the same standard even if the agreement is politically or institutionally significant. Transparency about implementation, uncertainty and fiscal outcomes permits future decisions to incorporate accumulated evidence.

The central policy conclusion follows from the distinction between additional activity and reallocation: a government should define success in terms of the productive investment, public revenue, legal certainty and social value that the agreement is intended to support. Changes in the aggregate financial geography remain important intermediate observations. They cannot independently establish success or failure.

Limitations and reproducible research agenda

This monograph is a documentary and analytical synthesis. It uses selected official materials, primary research abstracts, selected publisher full-text passages, and historical institutional evidence. It does not claim systematic coverage of every published article, a complete status inventory of treaties through November 2026, independent econometric replication, or access to confidential multinational microdata.

Specific problems remain unresolved at the evidence level. Several empirical studies were available only as original abstracts through bibliographic repositories. Their reported principal results can be accurately identified and compared at that level, but their full specification tables, sample construction, uncertainty estimates and robustness tests could not be independently audited in this session. Selected full-text access was blocked by publisher restrictions. This bounds the confidence of the evidential synthesis, particularly where the literature reports different magnitudes.

Future replication should begin with the precise datasets and historical legal versions used by the cited studies. It should establish the denominator for each measured outcome, reconstruct relevant source and residence provisions, identify exposure to network changes, separate immediate and ultimate investor information, and compare results with enterprise activity. Researchers should preregister the definition of treaty exposure and the primary outcomes, present sensitivity analyses, and use appropriate uncertainty estimates.

A separate comparative project could examine the effect of anti-abuse reforms on the geography of recorded direct investment, productive capacity and actual tax collections. That project would require complete effective-date information across a sample of agreements, consistent reporting by the host economies, and a defensible strategy for spillovers. Its contribution would be empirical. The present research has accomplished a conceptual and evidential reconciliation that defines the questions such a project would need to answer.

The historical India–Mauritius example, the Dutch empirical setting, and the African sample are intentionally bounded cases. They represent different instruments, data and geographic scopes. Their comparison illuminates shared measurement and policy problems without assigning identical effects across countries. The conclusions concerning minimum taxation are restricted to the documentary and scholarly material consulted, rather than asserting that every jurisdiction has implemented a common current regime.

These limits are intrinsic to the evidence used here. They do not imply that each substantive conclusion is unresolved. The accounting distinction between a financial claim and a produced asset is well established; the magnitudes of treaty-induced changes remain conditional on particular evidence. The manuscript makes its strongest claims at the former level and reports uncertain empirical magnitudes within their documented scope.

Conclusion

Treaty shopping is an economically significant phenomenon because international tax treaties alter the returns and obligations associated with cross-border corporate relationships. The resulting incentives can affect where firms operate, where their ownership interests are recorded, how they finance affiliates, and which governments receive tax revenue. These consequences can occur simultaneously and require independent measurement.

The international tax treaty network gives individual bilateral agreements economic importance beyond the immediate contracting relationship. Relevant empirical research shows that third-country alternatives matter for investment geography, that certain treaty provisions are associated with the routing of positions, and that treaty effects on investment can vary across firms and country samples. The evidence does not support a universal numerical magnitude for treaty-induced investment or fiscal losses.

The most important distinction is between the creation of additional productive activity and the reorganization of existing financial relationships. An increase in recorded inward FDI can reflect a change in financing, ownership, valuation or asset acquisition as well as a genuine addition to productive capacity. Ultimate investor attribution improves geographic understanding but does not itself quantify the production response or determine the tax legitimacy of an arrangement.

The public-finance consequences must be measured by jurisdiction and tax base. A state can surrender source revenue without necessarily attracting new production, and measures that increase a multinational group's total tax burden need not restore the source government's collections. A treaty may reduce double taxation and improve legal certainty, generating economic benefits even when its effect on a measured bilateral position is modest.

The author's contribution is an integrated interpretation of the empirical disagreement: treatment relevance in a treaty network, firm-specific uncertainty, rerouting of financial claims, and host-level additional investment are separate mechanisms underlying heterogeneous observations. They explain why evidence from positions, operating outcomes and revenue cannot substitute for one another. The proposed evaluation sequence makes those distinctions operational through linked evidence, while leaving econometric quantification to research with sufficient data.

Effective treaty analysis consequently combines institutional specificity, statistical discipline, economic counterfactuals and transparent evidence about distributional outcomes. The appropriate measure of success is the contribution of the treaty to the public objectives for which it was negotiated, examined alongside the actual fiscal and administrative costs. This is the economic basis on which treaty shopping and investment flows can be understood within the global offshore system.

References

  • UNCTAD (2024). Double taxation treaties and their implications for investment: What investment policymakers need to know. UNCTAD/DIAE/PCB/2024/1. United Nations, Geneva. Source record
  • OECD (n.d.). Preventing tax treaty abuse. OECD, Paris. Accessed November 18, 2026. Source record
  • OECD (2016). Multilateral Convention to Implement Tax Treaty Related Measures to Prevent Base Erosion and Profit Shifting. Articles 2, 6 and 7. OECD legal instrument. Source record
  • OECD (2025a). OECD Benchmark Definition of Foreign Direct Investment (Fifth Edition). Chapter 2, Main concepts and definitions of foreign direct investment. OECD Publishing, Paris. DOI: 10.1787/7f05c0a3-en. Source record
  • OECD (2025b). OECD Benchmark Definition of Foreign Direct Investment (Fifth Edition). Disaggregation by geography and by industry. OECD Publishing, Paris. DOI: 10.1787/7f05c0a3-en. Source record
  • Damgaard, J., Elkjaer, T., & Johannesen, N. (2019). What Is Real and What Is Not in the Global FDI Network? IMF Working Paper 19/274. International Monetary Fund, Washington, DC. Source record
  • Weyzig, F. (2013). Tax treaty shopping: Structural determinants of foreign direct investment routed through the Netherlands. International Tax and Public Finance, 20(6), 910–937. DOI: 10.1007/s10797-012-9250-z. Source record
  • van 't Riet, M., & Lejour, A. (2018). Optimal tax routing: Network analysis of FDI diversion. International Tax and Public Finance, 25(5), 1321–1371. DOI: 10.1007/s10797-018-9491-6. Source record
  • Hong, S. (2018). Tax treaties and foreign direct investment: A network approach. International Tax and Public Finance, 25(5), 1277–1320. DOI: 10.1007/s10797-018-9489-0. Source record
  • Petkova, K., Stasio, A., & Zagler, M. (2020). On the relevance of double tax treaties. International Tax and Public Finance, 27(3), 575–605. DOI: 10.1007/s10797-019-09570-9. Source record
  • Blonigen, B. A., & Davies, R. B. (2002). Do bilateral tax treaties promote foreign direct investment? NBER Working Paper 8834. National Bureau of Economic Research, Cambridge, MA. DOI: 10.3386/w8834. Source record
  • Blonigen, B. A., Oldenski, L., & Sly, N. (2014). The differential effects of bilateral tax treaties. American Economic Journal: Economic Policy, 6(2), 1–18. DOI: 10.1257/pol.6.2.1. Source record
  • Beer, S., & Loeprick, J. (2021). Too high a price? Tax treaties with investment hubs in Sub-Saharan Africa. International Tax and Public Finance, 28(1), 113–153. DOI: 10.1007/s10797-020-09615-4. Source record
  • Government of India, Ministry of Finance (2016). Notification of Protocol for amendment of the Convention for the avoidance of double taxation and the prevention of fiscal evasion with respect to taxes on income and capital gains, and for the encouragement of mutual trade and investment between India and Mauritius. Press Information Bureau, August 29, 2016. Source record
  • van 't Riet, M., & Lejour, A. (2025). Developing countries, tax treaty shopping and the global minimum tax. Public Finance Review, 53(6), 671–732. DOI: 10.1177/10911421251334849. Source record
  • OECD (2023). Tax Challenges Arising from the Digitalisation of the Economy – Subject to Tax Rule (Pillar Two). STTR and Commentary. OECD Publishing, Paris. Source record
  • Brada, J. C., Drabek, Z., & Iwasaki, I. (2021). Does investor protection increase foreign direct investment? A meta-analysis. Journal of Economic Surveys, 35(1), 34–70. DOI: 10.1111/joes.12392. Source record