RESEARCH DOSSIER
Ultimate and Immediate Investor Statistics
Abstract. Foreign direct investment statistics assign financial positions to economies by immediate counterparty and, in increasingly important complementary presentations, by ultimate investor. This monograph establishes the economic meaning of the two geographic views, examines the OECD fifth Benchmark Definition, differentiates winner-takes-all and proportional-ownership allocation, and analyzes the effect of corporate control chains, special purpose entities, reverse investment, pass-through capital, and round-tripping. Comparative evidence from the United States, Canada, the European Union, the United Kingdom, Japan, and Italy demonstrates the importance of matching statistical populations, instrument coverage, valuation, accounting basis, and reference periods. Original worked examples establish the conservation of a fixed position total under complete geographic reallocation, the non-identification of individual investment paths from marginal totals, and the limits of inferring funding and production from control. The study distinguishes official observations from model-based estimates and identifies the appropriate scope of interpretation for financial geography, multinational production, and policy. It concludes that the two investor views provide complementary evidence about the financial and organizational architecture of international investment.
Keywords: Foreign direct investment; immediate investing economy; ultimate investing economy; direct investment positions; corporate control; directional principle; winner takes all; proportional ownership; special purpose entities; pass-through capital; round-tripping; offshore economics; external sector statistics. JEL: F21, F23, C82
Introduction
International investment is simultaneously a financial relationship and an expression of organizational structure. A company located in one economy can hold an equity position in an enterprise located in another while its own voting rights are controlled in a third. The economy in which the direct claim is booked then differs from the economy that exercises authority over its holder. Both geographic facts have economic content. The first describes the external balance sheet and the institutional location of the relationship; the second describes an aspect of the organization of multinational enterprise.
The research object of this dossier is the geographic attribution of foreign direct investment (FDI), principally inward positions, when the immediate investing economy (IIE) and the ultimate investing economy (UIE) diverge. Four questions organize the inquiry. What is being measured before a position is assigned to an economy? Which ownership relationships identify an ultimate investor? What changes mathematically when the attribution changes? Which conclusions about international integration survive the distinction between financial claims, corporate control, production, and the ultimate supply of funds?
The subject belongs to Offshore Economics because international corporate networks distribute their financing and holding functions across jurisdictions. The economic geography of these functions deserves measurement in its own right. A foreign affiliate that intermediates a claim may perform important legal, treasury, or organizational tasks, even when it employs few people locally. Reassigning its investment to a controlling parent answers a distinct question while preserving the significance of the recorded immediate relationship.
The central conclusion developed here is that investor geography is a property of a precisely defined statistical relationship. A country ranking has meaning only together with its population, accounting presentation, instruments, reference date, valuation, and allocation rule. A change in one of these dimensions can change the ranking without any new capital movement. The empirical task is therefore to distinguish reattribution of an existing amount from changes in the amount of investment itself.
Scope, Sources, and Research Method
This research uses a comparative analysis of official standards and published country evidence, followed by a formal examination of attribution and independently calculated illustrative examples. The normative foundation is the fifth OECD Benchmark Definition (2025). Official national publications are treated as observations for their own defined years and populations. Published research on conduit capital and the multinational production problem is used to investigate the consequences of competing measurement decisions. The international macroeconomic context is the IMF's March 2025 release of BPM7, the seventh edition of its external-sector statistical manual.
The evidence design distinguishes four levels. Standards define a measurement procedure; official statistical releases provide observations under a specified procedure; research papers propose or test methods of inference; mathematical examples establish consequences under stated assumptions. Their evidential roles remain explicit. A model estimate cannot be converted into a national official observation merely because its result is reported by an international organization. Abstract-only access to research is used for its stated purpose and summary findings, not for undocumented claims about estimation details.
The reference frame for statistical positions is the end of the stated calendar year. Monetary examples use artificial units unless a currency is explicitly identified. The empirical comparisons below use published selected figures rather than a new microdata linkage or harmonized world matrix. The scope is international and comparative, with examples from the United States, Canada, the European Union, the United Kingdom, Japan, and Italy. The analysis is methodological rather than a census of every national implementing agency. The selected source records are documented in the bibliography and evidence ledger. Claims about contemporary standards distinguish adoption of recommendations from the actual timing of national implementation.
The authorial contribution is an integrated account of allocation, identification, and interpretation, with worked examples and a reproducible set of descriptive comparisons. This is a scientific synthesis of existing measurement concepts. It does not claim discovery of an independent empirical dataset or a new named theoretical system.
The Statistical Object of Direct Investment
The unit of observation is a financial relationship between institutional units associated through direct investment. Under the OECD's direct investment relationship framework (2025, chapter 2), a cross-border holding of at least 10 percent of voting power is the threshold identifying a direct investment relationship, including qualifying indirect relationships. Majority control is conventionally associated with more than 50 percent of voting power. The framework also encompasses qualifying positions between fellow enterprises with a common investor. The population of direct investment relationships is consequently wider than the population of majority-controlled enterprises.
These thresholds refer to voting power. The market value of ordinary shares, the proportion of dividends paid to an investor, the number of board members it nominates, and its voting rights need not be identical. A minority share can establish a direct investment relationship while another shareholder exercises control. A debt claim between fellow enterprises can belong to the same statistical category even though the lender did not subscribe to voting equity in the borrower. The financial amount recorded and the identity of the controlling unit are separately determined.
An investment position measures the outstanding value of a relationship at a specified time; a financial transaction measures activity during an interval; investment income measures earnings or interest attributable to the financial relationship. Changes in the reported stock cannot be read directly as changes in physical capacity. The acquisition of existing shares, the retained earnings of an affiliate, changes in the reporting currency, and intercompany borrowing can each affect the figures through different accounting channels.
Consider a resident factory financed by equity and debt supplied through a foreign holding company. Its operating plant may be physically unchanged when that company's parent is acquired by a group resident in another economy. Yet the factory's statistical investor geography can change. Conversely, its equity can appreciate because the factory becomes more profitable while the countries in both ownership presentations stay the same. Separating financial measurement from organizational classification is essential for interpreting both episodes.
Immediate Investor and Residence
The immediate investing economy is the resident economy of the relevant direct-investment counterparty. Under the OECD geographic allocation rules (2025, paragraph 500), the counterparty relationship anchors bilateral positions even when settlement uses an institution elsewhere. The location of a wire transfer, the currency of a bank account, and the jurisdiction of the securities custodian are insufficient evidence of the direct investor's residence.
Suppose a parent in economy A controls a holding company resident in B, which owns shares issued by an operating enterprise resident in C. The position in C attributed to B is a genuine cross-border financial relationship. B's holding company may have few employees, but its shares and liabilities are part of its balance sheet. C's immediate-investor statistics answer a contractual and residence-based question about the ownership of its obligations and equity. That answer can inform a study of balance sheet linkages, distributions of income, and the exposure to institutional arrangements in B.
Residence also differs from nationality in ordinary language. A corporation can be resident in the economy where it is statistically recognized as an institutional unit while its investors, executives, or employees reside in several other economies. An individual's residence is a statistical economic concept rather than a passport classification. The term "ultimate investing economy" therefore describes the residence of the designated ultimate entity, rather than citizenship of its directors or shareholders.
For a researcher examining financial intermediation, a large immediate position in a financial center constitutes evidence of the center's role in the observed financial structure. For a researcher examining authority over a manufacturing affiliate, the same position may be an incomplete geographic descriptor. The scientific value of the immediate measure becomes clearest when the research question is stated first.
Ultimate Investor and Corporate Control
A control chain is a succession of ownership relationships through which superior units can exercise authority over subordinate ones. In a simple hierarchy, a compiler moves upward from the relevant entity through controlling links until reaching a unit without a further controlling parent. The resulting residence is the ultimate investing economy under the chosen attribution method. The statistical meaning of "ultimate" is consequently dependent on the designated starting point and the rule governing continuation of the search.
A simple chain can preserve control while greatly reducing the ultimate parent's proportional cash-flow entitlement. If A owns 60 percent of B and B owns 60 percent of C, A controls B and through B controls C. The product of these two participation shares is 36 percent, under the simplifying assumption of identical voting and cash-flow rights and no other holdings. It follows that assigning an investment position to A as a controlling economy does not establish that A receives 100 percent of C's profits. Nor does the 36 percent product establish the amount of financial capital originating in A.
This distinction has consequences for the measurement of national ownership. The top company of a multinational enterprise may itself have a broad investor base. Its largest shareholder may possess a substantial economic stake without controlling it. Moving the chain above the corporate parent until every natural-person shareholder is identified answers a different question and may lack a unique answer. The statistical endpoint can be a corporation, household, or other relevant unit; the endpoint must be meaningful for the measurement purpose and recorded with its residence.
The geographic attribution is thus an organizational classification of measured financial relationships. An analyst interested in the ultimate saver, creditor, or recipient of cash distributions needs additional information about funding and contractual rights. A classification of a parent company's location cannot substitute for that evidence.
Winner Takes All and Proportional Ownership
The OECD's fifth Benchmark Definition recognizes two allocation methods (2025, chapter 8): winner takes all and proportional ownership. Winner takes all (WTA) assigns the inward position of a direct investment enterprise (DIE) to the economy of its ultimate controlling parent. Proportional ownership (PO) assigns each immediate investor's measured position to the economy of that investor's ultimate controlling parent. Here proportional allocation refers to keeping the financial amount of each qualifying investor relationship, rather than multiplying the shares along every corporate ownership chain.
Consider an operating company in C. A holding company in B owns 70 percent of its voting equity and is itself controlled in A. An independent company in D owns the remaining 30 percent. Assume an inward FDI position of 200 units consisting of a 120-unit position held by B and an 80-unit position held by D. These are assigned illustrative position values; they need not be proportional to votes because the positions could contain different amounts of debt.
| Partner economy | Immediate | PO ultimate | WTA ultimate |
|---|---|---|---|
| A | 0 | 120 | 200 |
| B | 120 | 0 | 0 |
| D | 80 | 80 | 0 |
| Total | 200 | 200 | 200 |
WTA identifies A as the controlling economy of the enterprise and attributes the full 200 units to it. PO identifies A as the controlling economy of B's investment and D as the controlling economy of D's own investment. The balance sheet total remains the same. The two columns address different aspects of organization and financial participation. A user concerned with the likely locus of strategic operating decisions may prefer WTA; a user studying the distribution of financial stakes among investors needs the information retained under PO.
The contrast is especially important in joint ventures and enterprises subject to limits on foreign control. Imagine that a domestic parent owns 60 percent of the local enterprise and a foreign investor holds a 40-unit position. WTA can attribute that inward position to the domestic controlling economy. Under PO the foreign investor's position remains attributed to its own ultimate controlling economy. Without the method label, a reader could wrongly interpret a domestic WTA value as evidence that all recorded foreign funding was routed abroad by domestic investors.
When every immediate relationship falls under the same ultimate controller, both methods give the same allocation. Their agreement is a consequence of the structure of the observation, not a universal accounting identity. The presence of more than one independently controlled investor is a sufficient reason to test the sensitivity of the results to the allocation method.
Reallocation Arithmetic and Conservation
The arithmetic of geographic reassignment can be expressed without assuming a behavioral theory. Fix a reporting economy h, reference date, coverage, instrument, accounting basis, and valuation. Let p(e,j) be the recorded inward position of enterprise e associated with immediate investor j. Let r(j) denote the residence of that investor and u(j) the residence of its ultimate controlling parent. A representative PO table sums these positions according to r(j) for immediate geography and u(j) for ultimate geography. Under WTA it first sums within e and assigns the sum according to the residence of the ultimate controller of e.
For a fixed population, define I(i) as the aggregate attributed immediately to economy i and U(i) as the aggregate attributed ultimately to economy i. Include explicitly any unknown or unallocated economy as an additional category. Complete reassignment implies that the sum of I(i) over all categories equals the sum of U(i). Define D(i) = U(i) - I(i). The sum of D(i) is zero. This is an accounting implication of a change in labels while keeping the financial positions fixed.
A positive D(i) indicates that more of the selected aggregate is attributed to i after the change. A negative D(i) indicates that the immediate geographic representation assigned a greater amount to i. Either observation is compatible with many combinations of underlying chains. Difference tables reveal net shifts in measured attribution, not the individual origin-destination paths responsible for the shift.
A transition matrix adds information. Let M(i,k) be the sum of the original positions allocated from immediate economy i to ultimate economy k. Its row sums are the immediate totals and its column sums are the ultimate totals. The off-diagonal entries identify the gross value reassigned across economies. Two offsetting movements can produce identical marginal totals, so D(i) = 0 does not imply that no reassignment occurred in economy i.
For example, in a nonnegative portfolio of 100 units, 20 units might move from A to B, with 20 units moving from B to A. Both net changes are zero, but 40 units change geographic category. The total variation between marginal distributions is therefore a lower bound on the gross amount reassigned, where all amounts are nonnegative and denominators consistent. This lower-bound interpretation breaks down for signed net positions. The same row and column sum identity still holds, but cancellations and negative values preclude interpreting the resulting differences as ordinary shares of a nonnegative mass.
If an aggregate changes when the geographic reassignment is made, the researcher first tests whether coverage, valuation, signs, SPE exclusion, or missing categories also changed. Geographic reassignment alone conserves the observed reporting-economy total. This fact provides a practical quality check for published tables and for research estimates.
Position, Transaction, Income, and Valuation
A direct investment stock is a snapshot of a balance sheet relationship after accumulated events. The OECD's standard and supplementary series (2025, paragraphs 485–489) include a reconciliation of opening and closing positions through financial transactions, revaluation, and changes in volume. A methodological transition creates another reason to examine how a change in the published geography was produced.
Let a stock equal 100 units at the beginning of a period. During the period transactions add 15, price movements contribute 8, exchange-rate movements subtract 3, and other volume changes subtract 5. Its closing stock is 115 units. A pure reclassification across investor economies can then shift the 115 between country cells without changing the total. An ultimate-controller change may coincide with an actual acquisition involving other parties; that acquisition and the classification effects have to be investigated separately.
The transaction concept has a further complication. An acquisition of shares by an intermediate affiliate may be funded through another group company or through external borrowing. A classification based on the affiliate's controlling parent follows the organization of the relationship, not necessarily the immediate cash origin of the transaction. Historical point-in-time ultimate-owner information should be matched with the date of the transaction before a transaction-level geographic reconstruction is attempted. Applying a controller observed in 2025 retroactively to a 2015 transaction can create spurious historical movements.
Investment income also requires care. Retained earnings and distributed dividends have different implications for the balance sheet and cash payments. Interest may flow through group treasury arrangements. An ultimate attribution of income offers a useful group-level description but cannot reconstruct the payment received by every shareholder or creditor. Dividing a current-year income amount by a year-end position produces a descriptive ratio rather than an exact return on invested capital; calculating the latter would require a more suitable denominator and timing information.
Comparisons between economies must respect valuation methods. A historical-cost amount and a market-value amount may differ substantially even with identical ownership and underlying assets. Ratios and ranks can be affected where the mix of old and newly acquired investments differs. A comparison on a common currency basis also requires harmonized dates and conversion methods. These issues are independent of ultimate-investor classification.
Directional and Asset–Liability Presentations
The directional presentation groups financial relationships according to the direction of influence associated with direct investment. The asset–liability presentation groups cross-border claims and obligations on the resident balance sheet. The two presentations are related through the handling of reverse investment and fellow-enterprise relationships. This difference must be resolved before drawing comparisons across two tables.
Consider a resident parent in H with a loan of 90 units to one controlled affiliate in F and a borrowing of 20 from another affiliate in G. Assume both instruments qualify as FDI debt. The balance sheet presentation gives 90 of external assets and 20 of external liabilities. The directional outward figure is 70, consisting of +90 against F and -20 against G. A reader who treats the negative G position as a negative quantity of physical capital misreads the accounting presentation. The negative sign represents an offset in the net directional relationship.
The Bank of Japan recording principles (2021) explicitly distinguish these presentations and specify the directional basis for Japan's regional inward positions by ultimate investor. This example demonstrates why one cannot combine an immediate liability series from an IIP table with a directional ultimate inward series and treat the difference as conduit investment. A discrepancy can exist even when every parent is directly resident in the economy of its recorded counterparty.
The treatment of fellow enterprises is also consequential because the residence of their controlling parent helps determine the direction assigned to their qualifying financial relationships. A centralized treasury may lend across sister companies with no direct shareholding between them. If the analyst reconstructs inward ultimate positions from equity holdings alone, those debt relationships may disappear or be assigned inconsistently. The appropriate data collection includes the type of relationship and instrument alongside the ownership hierarchy.
A defensible paired comparison therefore uses the same accounting presentation on both sides and documents any deviations from the standard. Where published tables do not supply a bridge, the limits of the comparison belong in the result itself.
Special Purpose Entities and Pass-Through Capital
A special purpose entity (SPE) is an institutional arrangement whose economic functions can include holding equity, raising funds, or intermediate financing within a group. Its classification requires the applicable statistical definition, not simply the observation that it is registered in a financial center. Pass-through capital describes financial positions or movements transmitted through intermediate entities. The two phenomena overlap but are analytically separable: an operating business can also perform treasury or conduit functions.
An illustrative chain makes the separation clear. A parent in A provides 100 units to a holding company in B, which provides 100 to an operating company in C. With equal values at each stage, two cross-border financial relationships of 100 each exist. Inward positions across B and C sum to 200; the productive affiliate receives an investment position of 100. A study of the recorded external claims of all resident units counts both links. A study of financing directed toward the operating company may focus on the second link, attributed ultimately to A.
Simply reallocating C's inward amount from B to A leaves B's inward position of 100 in the system. A global aggregate of all resident enterprises remains 200. Excluding B's SPE from a particular analytical population can reduce that aggregate to 100. These operations have different mathematical effects. The first shifts a country's attribution within a total; the second changes the set of measured enterprises.
This distinction explains why a corrected network in empirical research may be both smaller in total and different in geographic composition. The OECD includes separate identification of resident SPEs in standard FDI statistical series (2025). A compiler can publish both the inclusive series and an analytical comparison excluding them. A user then separates the influence of the SPE population from the influence of the ultimate allocation.
The structure itself supplies no finding about legal compliance or the presence of evasion. Relevant questions include balance-sheet size, employment, control, financial function, value added, funding, and applicable law. Evidence of a large reattribution can justify further study of a financial network; it cannot establish the motives of the enterprise or the legality of its arrangements.
Round-Tripping and Domestic Ultimate Investors
A domestic ultimate investor can be associated with an inward foreign direct investment position because the immediate investor is resident abroad. A common organizational pattern consists of a domestic parent controlling a foreign affiliate that invests in an enterprise back in the home economy. The recorded foreign-investor link is consistent with a domestic source of corporate control. A PO attribution can reveal this relationship when the controlling parent of the foreign immediate investor is resident in the compiling economy.
The strongest economic claim supported by this information concerns the origin of control under the allocation rule. A domestic ultimate parent may use capital borrowed from foreign banks, issue bonds purchased by foreign investors, or retain earnings from overseas business. Consequently, a round-tripping position does not quantify the amount of savings originally generated domestically. If the research question is the additional foreign funding entering the country, the chain must be investigated alongside debt issuance, capital raising, and earnings.
Imagine a domestic parent that owns a foreign affiliate. The affiliate obtains 30 units of outside borrowing and 70 units from the parent, and then invests 100 units in the home country. In the simplified example, the full inward position has a domestic ultimate controller, while 30 units of its funding are external to the group's original home-country financing. A control-based statistic cannot identify the funding split without additional records. The same conclusion holds if the group uses its overseas retained earnings.
Evidence of domestic ultimate investment is useful for studying the organizational reasons that capital crosses borders and later returns. Possible explanations include historical group structure, financial intermediation, corporate restructuring, investment law, or tax incentives. Discriminating between them requires separate evidence. The causal interpretation should never be inferred from the statistical category alone.
Under WTA, a locally controlled enterprise with a genuine foreign minority investor can generate an apparently domestic allocation despite the foreign investor's independent control. The difference between the two methods is thus not confined to the numerical size of the category; it affects the economic phenomenon that the analyst can validly infer.
Compilation from Enterprise-Level Information
Compilation begins with a defined register of resident enterprises and their financial relationships. For each relationship, the statistical authority requires a reference date, reporting enterprise identifier, counterparty identifier and residence, financial instrument, direction, position value, valuation convention, and any relevant reporting classification. Ownership data should record voting power separately from capital participation and identify changes in control over time.
A practical sequence starts by reconciling the direct-investment population with the external accounts and obtaining counterparty information. Ownership chains are then linked from the relevant starting units. The compiler determines the controlling parent according to the chosen method, records the evidence for that designation, and assigns the existing financial amount. The process should retain the original country variable as well as the assigned UIE code. Comparisons with business registers and information reported by enterprise groups can resolve inconsistent names or legal entity identifiers.
Complex structures call for explicit quality controls. Two independent groups may own a joint venture equally, shares may be dispersed, a control record may be outdated, and ownership may be nested in multiple layers. A legal entity that appears several times in a chain should be distinguished from several similarly named entities. If a loop is detected, the simple upward-tracing algorithm should stop and flag the case rather than infer a foreign parent by arbitrary termination. Documented standard and national rules govern difficult ownership cases.
Unknown information is a substantive statistical category. Missing parent residence should be retained as unknown or clearly identified as estimated. Assigning every unknown observation to the immediate investor would implicitly assume no geographic reassignment for precisely the observations with missing chain information. Such a choice can bias the allocation toward jurisdictions with low ownership transparency.
Publication also requires protecting respondent confidentiality. Detailed country-by-industry data may contain sparse cells dominated by an individual enterprise. Complementary suppression, rounding, or ranges can affect the precision of tabulations. An analyst must not reconstruct suppressed company information from published marginal totals. Instead, aggregate comparisons should acknowledge the available resolution and retain distinctions between true zero, missing, suppressed, and unallocated values.
Aggregate Inference and Identification Limits
A complete enterprise-level linkage is not always available, especially across multiple jurisdictions with different registration systems. Research has therefore explored ways to estimate ultimate-investor distributions from available macroeconomic observations and samples of observed ultimate relationships.
Casella's (2019) probabilistic approach to conduit FDI uses absorbing Markov chains to infer ultimate origins. At a conceptual level, a probabilistic traversal treats certain states as intermediaries and allocates investment across terminal states according to estimated transition behavior. This offers a way to infer a geographic distribution in the absence of a fully observed corporate graph. The inferred result is conditional on the design of the model and the quality of its inputs.
The identification limit can be illustrated with an artificial two-origin system. Suppose B has inward investment of 100 from A and 100 from D, and its companies hold 100 in C. These aggregate bilateral numbers are consistent with all 100 in C being controlled by A, all 100 being controlled by D, or any mixture. The balance sheet totals alone contain no enterprise-level matching between B's inward and outward links. A proportional assignment of 50 to A and 50 to D is a possible assumption, not an identified result.
In a simple probabilistic example, an intermediate state has a 0.5 probability of reaching terminal A, 0.3 of reaching terminal D, and 0.2 of remaining in the intermediate state. Repeated traversal under unchanged probabilities gives eventual absorption probabilities of 0.5 divided by 0.8 = 0.625 and 0.3 divided by 0.8 = 0.375. This is elementary illustration of conditional normalization under eventual absorption; it is not a reproduction of any published empirical estimator.
The researcher should distinguish measured links from inferred ones and run sensitivity analysis over alternative assumptions about intermediary status, missing countries, negative positions, and the probability of terminating a chain. A matrix built from estimates can be highly valuable for comparative analysis while conveying less evidential certainty than a directly observed investor classification. Model validation against independently observed ultimate distributions is the central test of its usefulness.
International Standards and Data Architecture
The development of FDI standards reflects the growing complexity of multinational structures. The OECD 2025 Benchmark Definition expands the standard presentation of inward positions by UIE, while UHE breakdowns of outward positions and UIE breakdowns of inward income have supplementary roles. The distinction between a recommended standard series and an actual national data release remains important during implementation. New recommendations do not retrospectively transform earlier data compiled under previous manuals.
The IMF's Direct Investment Positions dataset (formerly the Coordinated Direct Investment Survey) provides international bilateral immediate-partner positions, commonly distinguished by equity and debt. Mirror information derived from counterparts can assist where a reporter has missing observations. This architecture is especially important for the study of immediate financial networks, but the dataset description is not a statement that a complete global directly observed ultimate-investor matrix exists.
A sound empirical workflow starts with the appropriate series. The IMF bilateral immediate matrix can establish the financial connections recorded by reporters; OECD or country-specific ultimate series can identify reattribution for countries and years in which comparable tables exist. National methodological notes supply the allocation rule, SPE coverage, instrument definitions, valuation basis, and confidentiality rules. A world comparison also requires common country coding and a record of which territories are separate statistical economies.
Immediate outward positions of one country and immediate inward positions of its partner are conceptually related, but they can differ in practice because of timing, valuation, reporting coverage, and data asymmetry. A UIE inward position adds another asymmetry because it has been assigned to an ultimate origin. It is not a conventional mirror of that origin's outward immediate position. A researcher needs a consistently constructed outward destination-side measure to pose a different mirror question.
The distinction is particularly important for the Global Offshore System program. Its geographic evidence should retain the full architecture of the underlying series. Decompositions of an official immediate matrix into analytically adjusted networks must carry the source coverage, method, and estimation uncertainty with them.
United States: Foreign Parent and Ultimate Beneficial Owner
The U.S. Bureau of Economic Analysis publishes inward direct investment by country of foreign parent and by country of ultimate beneficial owner (UBO). Its 2025 direct investment release (2026) provides a recent comparison on a historical-cost basis. Japan is the leading foreign-parent economy with 751.8 billion), Canada (738.3 billion). In the UBO ranking, Japan remains first at $827.1 billion; Canada follows at $819.8 billion and Germany ranks third at $706.2 billion.
Using the rounded published amounts, the paired difference for Japan is $50.8 billion, about 6.5 percent of its foreign-parent position. Canada's paired difference is $72.5 billion, about 9.7 percent of its foreign-parent position. These country differences are net increases in attribution. They cannot reveal the precise intermediaries through which the corresponding investments are held without a transition matrix or linked ownership records. The ranking also changes because the location of a controlling group can differ from that of its immediate U.S. investor.
BEA uses its own terminology. The word UBO in these tables denotes the entity at the top of a group ownership chain as defined for direct investment statistics. Its institutional context must be distinguished from anti-money-laundering terminology. The BEA release is evidence of an operational national statistical series, rather than a license to treat UBO definitions across regulatory domains as interchangeable.
The release revises the total historical-cost inward position for 2024 from $5,708 billion to $5,598 billion. That $110 billion downward revision illustrates the relevance of data vintages. A difference between two successive publications can contain a revision even for the same reference year. The release also notes use of coarsening for confidentiality, so displayed precision and country coverage deserve attention when calculating derived indicators.
Canada: A Paired Country Comparison
Statistics Canada's paired 2025 investor-country chart (2026) supplies selected country positions in billions of Canadian dollars. Unlike a ranking comparison involving different sets of countries, this publication directly presents paired figures for the same selection. The table below uses six of its ten published rows; the difference is calculated as ultimate minus immediate. The amounts retain the release's one-decimal precision.
| Investing economy | Immediate | Ultimate | Difference |
|---|---|---|---|
| United States | 737.3 | 816.8 | +79.5 |
| China | 24.6 | 43.4 | +18.8 |
| Germany | 26.2 | 40.5 | +14.3 |
| France | 30.4 | 41.9 | +11.5 |
| Luxembourg | 76.2 | 28.3 | -47.9 |
| Netherlands | 182.9 | 23.4 | -159.5 |
The United States' ultimate position exceeds its immediate position by about 10.8 percent relative to the immediate baseline. The reported Netherlands amount falls by approximately 87.2 percent relative to that baseline when it is compared with the corresponding ultimate amount. Luxembourg's decline relative to its immediate amount is approximately 62.9 percent. These changes describe the difference between two country allocations, not the fraction of individual investors in those economies engaged in pass-through activities.
The six selected differences sum to -83.3 billion. This is neither a loss of Canadian investment nor a contradiction of conservation because the six countries are a subset of the full geographic distribution. Net changes in omitted categories can balance the result. The selection is purposefully illustrative of contrasting positive and negative differences and should not be mistaken for a representative sample of all foreign investor economies.
This case highlights why a dataset with a shared Canada-specific reporting population and two partner classifications is preferable to comparisons assembled from unrelated national totals. A further research step would require the complete tables, including all categories, and a breakdown of the transitions connecting immediate to ultimate geography. The present comparisons use published displayed values rather than a downloaded underlying database.
European Union: Matched Reporting Populations
Eurostat's 2024 ultimate investor comparison (2025) reports inward positions for 13 EU member states that supplied the relevant ultimate-investor data for that reference year. Both the immediate and ultimate presentations total EUR 3,892 billion within the matched reporting population. On the ultimate basis, the United States accounts for 14 percent and Germany for 12 percent. On the immediate basis, Luxembourg accounts for 17 percent and the Netherlands for 15 percent. These are rounded shares of the same reported aggregate.
The result directly demonstrates a change in geographic emphasis when investor attribution changes while the total is held constant. It also has an important limitation: the 13-country cohort is not the whole EU. Germany can be a prominent source of ultimate investment in this sample even though Germany's own inward positions were not available in that comparison. Reporter status and partner status are logically independent.
The 17 and 15 percent immediate values jointly represent about 32 percent of the aggregate, based on rounded figures. They cannot be interpreted as the combined amount of phantom investment because the ultimate values of those two economies and the underlying enterprise characteristics are separate questions. It would also be inappropriate to combine the sample's ultimate ranking with a full-EU aggregate to derive country amounts.
Comparisons over successive Eurostat releases must hold the cohort fixed or explicitly quantify the effect of adding and removing reporting economies. A newly participating economy can alter the aggregate shares even if the investment structures of all continuing reporters remain unchanged. Method labels, reporter lists, and reference periods are therefore integral to the result.
Regional aggregation introduces a further choice about whether intra-regional financial relationships are included or eliminated. Aggregating national inward data does not automatically reproduce an external-investment statistic for a consolidated regional economy. The distinction is particularly pertinent for transnational ownership chains passing through member states.
United Kingdom, Japan, and Italy
The United Kingdom Office for National Statistics published ultimate controlling economy analysis (2023) for reference years 2014–2021, comparing inward positions and earnings on a directional basis. It reports that earnings associated with UK round-tripping were approximately GBP 0.3 billion in 2021 and less than 0.5 percent of total inward FDI earnings that year. The example demonstrates the feasibility of an ultimate-geographic income analysis. It does not establish the corresponding percentage of the stock, because earnings and positions measure different quantities.
Japan's presentation, discussed above, shows that an official statistical system may maintain regional data on both accounting bases. Before comparing its ultimate inward position with another series, the analyst selects the corresponding directional immediate series. The choice of measure is determined by its documentation rather than the title of a downloaded table.
An Italian study by Accoto and Oddo (2023), Looking Beyond the Curtain, uses survey information collected for balance of payments compilation in 2013–2020 to investigate pass-through and round-tripping across countries and sectors. Its stated methodology illustrates the value of direct firm-level ultimate-investor information for separating those phenomena. The study's design supports the methodological point; this dossier does not quote its empirical magnitudes because those findings were not independently reconstructed from the full research data.
These three cases illustrate distinct dimensions of statistical capability: extending ultimate attribution to earnings, maintaining transparency about accounting presentations, and using microdata to investigate the economic structure behind aggregate financial positions. Each national institution operates within its own collection system. Cross-country comparisons should state whether the published ultimate mechanism follows the same starting unit, valuation, and population as the other national series.
The United States, Canada, and the European example supply directly comparable published facts within their respective releases. The additional cases show the diversity of procedures and available measures. Together they justify methodological differentiation, rather than the construction of a single pooled ranking from incomparable figures.
Research Literature on Global FDI Networks
Damgaard and Elkjaer (2017) analyze bilateral asymmetry, SPEs, and ultimate geography in The Global FDI Network. Their publisher summary reports that the adjusted global network has an inward investment total approximately one-third smaller after removing SPEs and reallocating the remaining investment. The reduction relates to the change in analytical coverage; the change in partner distribution relates to ownership attribution.
Damgaard, Elkjaer, and Johannesen (2019) extend the inquiry in What Is Real and What Is Not. Their published summary estimates that investments in empty corporate shells could represent nearly 40 percent of global FDI and reports higher explanatory power of conventional gravity variables when phantom investment is separated and real investment is attributed ultimately. The estimate is a result of the authors' empirical procedure for their studied data, rather than a constant applicable to every year, a measured proportion of illegal capital, or a substitute for a contemporary official statistic.
The key analytical lesson is that two sources of discrepancy can coexist. Global positions may be enlarged by multiple recorded cross-border relationships within one corporate structure, while the immediate partner field may conceal the group controlling a downstream investment. Adjusting only one can leave the other unresolved. Furthermore, a regression estimate using a modified dependent variable cannot by itself identify a causal effect of offshore intermediation. The model design, competing covariates, and changes to the sample need independent investigation.
Casella, Borga, and Wacker (2023) address the wider question in Measuring Multinational Production with FDI. Their published summary revisits the objections to using FDI as an indicator of multinational production and considers conditions for appropriate use. The availability of FDI observations is an advantage for international research, but measurement of the geographic location of financial claims has to be linked carefully to the location of productive operations.
These contributions support an evidence-sensitive approach to adjusted FDI matrices. They address overlapping but distinct research tasks, ranging from the measurement of financial intermediation to the distribution of ultimate ownership and the representation of multinational production. No research estimate is treated in this dossier as an independently verified cell of a newly compiled global matrix.
Connections to Production, Securities, and Banking Statistics
The corporate group that ultimately controls a plant and the country in which its workers and machinery operate describe different dimensions of economic activity. Foreign affiliate statistics (FATS) can record employment, turnover, and production-related measures for controlled enterprises. Eurostat's foreign affiliate data guidance explains this population and its measures. The OECD's links with other statistics (2025) outlines the alignment of controlled-enterprise geography with ultimate ownership while preserving differences between the statistical domains.
FDI includes qualifying minority relationships, whereas inward FATS organizes information about foreign-controlled enterprises. Even when WTA aligns investor geography with the controlling parent, the population and measured variables remain different. A researcher who combines the two sources should align industry classifications, group identifiers, coverage, and reference dates, and retain separate variables for finance and operations.
A neighboring literature examines portfolio securities issued through offshore affiliates. Coppola, Maggiori, Neiman, and Schreger (2021), Redrawing the Map of Global Capital Flows, reassign traded securities to the issuer's ultimate parent in their study of international portfolio holdings. The shared insight is that a legal issuer's residence may diverge from the group location relevant to certain analytical questions. Their security-level method addresses a different population and does not supply a direct-investment UIE matrix.
Banking statistics provide another distinction. The BIS consolidated banking statistics describe bank groups' claims on immediate counterparties and on a guarantor basis after relevant contractual risk transfers. A guarantee can reallocate country risk to an entity different from the company that controls the immediate obligor. Equity control alone is therefore insufficient to reconstruct the ultimate contractual bearer of a credit loss.
A coherent study of international finance uses these measures in complementary ways. Immediate FDI identifies resident financial relationships; ultimate FDI classifies the location of specified corporate control; foreign affiliate statistics measure operations; securities reallocation links investments to issuing groups; banking exposure statistics address creditor claims and risk transfer. Combining them without matching their concepts would create a hybrid figure with no clear unit of observation.
Legal Beneficial Ownership and Economic Interpretation
The expression "beneficial owner" carries a domain-specific meaning. The FATF glossary on beneficial ownership defines the ultimate beneficial owner for its anti-money-laundering framework in terms of natural persons. The statistical controlling parent can be a corporate institutional unit. A multinational group with dispersed ownership may accordingly have a statistical ultimate parent even when the AML investigation concerns a different set of natural persons or control relationships.
This distinction is essential when economic research uses company registries as evidence. A public registry designed for legal transparency may contain different thresholds, legal tests, exemptions, or update dates from the direct-investment definitions. The compiler must evaluate whether a field called "beneficial owner" actually identifies the statistical entity required for UIE attribution. Information useful for one inquiry can inform another only after the relevant definitions have been reconciled.
A further distinction concerns the economic ownership of assets and the ultimate incidence of risk. A statistical position assigned to a parent economy does not measure the share of losses absorbed by creditors, equity holders, governments, or insurers resident there. Risk is distributed through balance sheets and contracts. A business may issue liabilities to institutional investors around the world, and a minority investor may bear material equity risk independently of corporate control.
Consequently, the geographic location of the ultimate investor is not a comprehensive attribution of private wealth to persons. It is a classification within a defined statistical system. The implications for taxation, investment screening, beneficial ownership disclosure, and legal responsibility require evidence and analysis specific to those fields.
For the offshore research program, this boundary is productive. Economics can analyze the amounts and network relations established through statistical methods while Law can investigate the legal rights and obligations of entities and natural persons. A claim about the ultimate controller's location need not carry an allegation about behavior or compliance.
Comparative Inference, Policy, and Verification
The preferred measure depends on the policy question. In studies of external financial interdependence, immediate positions preserve direct jurisdictional claims. In studies of strategic corporate control, WTA attribution can identify the controlling economy of a direct investment enterprise. In financial-ownership analysis involving minority investors, PO preserves the amounts attached to separate investor relationships. In development research, the position must be compared with productive activity, investment expenditure, and sector-level employment. A country's high position relative to GDP can reflect both financial intermediation and substantial operating activity; the ratio alone cannot separate them.
Applications to gravity models require an explicit dependent variable. An immediate position and an ultimate position are linked measurements but represent different economic relationships. Geographic distance, common language, trade ties, and institutional characteristics may explain operating relationships differently from intermediary locations. Changing the dependent variable, removing SPEs, and dropping incomplete observations can each affect coefficient estimates. A defensible empirical design would vary these decisions independently and report their consequences.
For time-series studies, the analyst should hold a consistent standard and data vintage wherever possible. If a compiler introduces a UIE series or switches the method from PO to WTA, the resulting break may be statistical rather than organizational. If a group restructures its ownership with no change in its underlying plant or financial amount, immediate positions can change while ultimate attribution remains stable. The opposite pattern occurs when control changes while the immediate affiliate remains the same. The two views jointly assist interpretation of these episodes.
A comparative verification sequence begins with source identity and the exact observation labels, checks reference date and currency, confirms matching reporting populations, checks whether values are positive or signed, records valuation and SPE treatment, and identifies whether the ultimate starting point is the enterprise or the immediate investor. Where a complete table is available, totals and difference sums provide useful arithmetic checks. Where only a subset is available, no conservation assertion should be made about that subset. Suppressed values remain suppressed and modelled values remain labelled.
Three broad research risks recur: selection bias from uneven country reporting; measurement error from incomplete ownership chains; and conceptual mismatch when finance is interpreted as production or personal wealth. The practical response is to publish metadata, describe the scope of the evidence, separate observed and estimated values, and perform sensitivity analyses tied to a clearly stated economic question. These are ordinary requirements of sound statistical work rather than a newly proposed theory.
An investment-promotion institution can use ultimate statistics to understand the geographic organization of investment already present, but needs project-level information to evaluate new productive capacity. A financial-stability authority needs instrument, maturity, currency, and guarantee information. A tax authority needs legal arrangements, transactions, and taxable income. Matching each decision to the appropriate measurement prevents the overextension of a useful statistic.
Conclusion
Ultimate and immediate investor statistics establish two complementary geographic descriptions of international direct investment. The immediate presentation is anchored in resident counterparties and preserves the network of recorded financial relationships. Ultimate attribution uses corporate ownership information and a stated allocation method to identify the economy associated with an investment's designated controlling unit. The difference between the two views is a measurable property of multinational organizational structure.
The choice between winner takes all and proportional ownership materially affects enterprises with independently controlled minority investors or domestic majority control. Both methods can be internally consistent, but the starting unit and the attribution rule determine the meaning of the result. The distinction becomes especially important in the measurement of domestic ultimate investment and in comparisons with foreign affiliate statistics.
The arithmetic of reallocation provides a central finding: where the same positions are covered and all categories are retained, the total remains unchanged, while geographic distribution can shift substantially. Excluding particular entities or instruments changes the aggregate and represents a separate analytical operation. The Canadian, U.S., and matched EU examples illustrate the size and direction of observed differences without requiring assumptions about the legality or precise funding origins of the underlying investments.
Corporate control, ultimate funding, contractual credit risk, personal beneficial ownership, and production location constitute distinct objects of economic measurement. UIE statistics illuminate one of them and strengthen the interpretation of others when the appropriate additional evidence is available. Their value to Offshore Economics lies in revealing the geography of organizational control while preserving the financial network that connects legal institutional units.
An adequate account of international integration therefore retains immediate and ultimate positions, defines the underlying financial population, distinguishes measured and estimated classifications, and explicitly links financial data with operational evidence when production is the subject. The resulting description of offshore financial geography is more precise because the jurisdictional locations of financial claims and decision-making authority are examined together.
The research establishes methodological conclusions supported by comparative standards, observed national releases, published research summaries, and transparent arithmetic. The evidence is adequate for these scoped conclusions; a worldwide enterprise-level reconstruction, full-text review of inaccessible studies, and independent empirical replication remain distinct tasks. No new named theory or quantitative claim of global novelty is advanced.
References
-
Accoto, N., & Oddo, G. (2023, April 21). Looking Beyond the Curtain: Pass-through Capital and Round-tripping in Italy's Foreign Direct Investment. Bank of Italy Occasional Paper 762. Rome: Banca d'Italia. DOI: 10.2139/ssrn.4464149. Bank of Italy Occasional Paper 762.
-
Bank for International Settlements. (n.d.). Consolidated banking statistics: overview. Basel: BIS Data Portal. Accessed November 18, 2026. BIS consolidated banking statistics.
-
Bank of Japan. (2021, June). Recording Principles of Direct Investment. Tokyo: Bank of Japan. Bank of Japan recording principles.
-
Casella, B. (2019). Looking through conduit FDI in search of ultimate investors – a probabilistic approach. Transnational Corporations, 26(1). DOI: 10.18356/8a8b094c-en. Repository version: MPRA Paper 95188, Munich Personal RePEc Archive. MPRA Paper 95188.
-
Casella, B., Borga, M., & Wacker, K. (2023). Measuring Multinational Production with Foreign Direct Investment Statistics: Recent Trends, Challenges, and Developments. IMF Working Paper 23/113. Washington, DC: IMF. DOI: 10.5089/9798400243530.001. IMF Working Paper 23/113.
-
Coppola, A., Maggiori, M., Neiman, B., & Schreger, J. (2021). Redrawing the Map of Global Capital Flows: The Role of Cross-Border Financing and Tax Havens. Quarterly Journal of Economics, 136(3), 1499–1556. DOI: 10.1093/qje/qjab014. Quarterly Journal of Economics article.
-
Damgaard, J., & Elkjaer, T. (2017). The Global FDI Network: Searching for Ultimate Investors. IMF Working Paper 17/258. Washington, DC: IMF. DOI: 10.5089/9781484329658.001. IMF Working Paper 17/258.
-
Damgaard, J., Elkjaer, T., & Johannesen, N. (2019). What Is Real and What Is Not in the Global FDI Network? IMF Working Paper 19/274. Washington, DC: IMF. IMF Working Paper 19/274.
-
Eurostat. (2025, November 11). Top ultimate investing economies: US and Germany. Eurostat News. Luxembourg: Eurostat. Dataset bop_fdi6_pos. Eurostat 2024 comparison.
-
Eurostat. (n.d.). Globalisation of businesses: information on data. Luxembourg: Eurostat. Accessed November 18, 2026. Eurostat data guidance.
-
Financial Action Task Force. (n.d.). FATF Glossary: Beneficial owner. Paris: FATF. Accessed November 18, 2026. FATF glossary.
-
International Monetary Fund. (2025, March 20). IMF and the statistical community release new global standards for macroeconomic statistics. Press release 25/072. Washington, DC: IMF. BPM7 press release.
-
International Monetary Fund. (n.d.). Direct Investment Positions by Counterpart Economy (formerly CDIS). IMF Data Portal, dataset IMF.STA:DIP. Accessed November 18, 2026. IMF DIP dataset.
-
OECD. (2025). Disaggregation by geography and by industry. In OECD Benchmark Definition of Foreign Direct Investment (Fifth Edition), ch. 8. Paris: OECD Publishing. DOI: 10.1787/7f05c0a3-en. Chapter 8: Geographic disaggregation.
-
OECD. (2025). Links between FDI and other statistics. In OECD Benchmark Definition of Foreign Direct Investment (Fifth Edition), ch. 10. Paris: OECD Publishing. DOI: 10.1787/7f05c0a3-en. Chapter 10: Other statistics.
-
OECD. (2025). Main concepts and definitions of foreign direct investment. In OECD Benchmark Definition of Foreign Direct Investment (Fifth Edition), ch. 2. Paris: OECD Publishing. DOI: 10.1787/7f05c0a3-en. Chapter 2: Main concepts.
-
OECD. (2025). Standard and supplementary FDI series. In OECD Benchmark Definition of Foreign Direct Investment (Fifth Edition), ch. 7. Paris: OECD Publishing. DOI: 10.1787/7f05c0a3-en. Chapter 7: Statistical series.
-
Office for National Statistics. (2023, July 25). Foreign direct investment by ultimate controlling economy, UK trends and analysis: 2021. Newport: ONS. ONS ultimate controlling economy analysis.
-
Statistics Canada. (2026, April 29). Foreign direct investment in Canada by select ultimate and immediate investor country, 2025. The Daily, Chart 2; tables 36-10-0008-01 and 36-10-0433-01. Ottawa: Statistics Canada. Statistics Canada chart 2.
-
U.S. Bureau of Economic Analysis. (2026). Direct Investment by Country and Industry, 2025. Washington, DC: U.S. Department of Commerce. BEA 2025 direct investment release.