This chapter provides guidance for compilers to disaggregate foreign direct investment (FDI) financial transactions, positions, and income by partner economy and by industry. It is recommended that disaggregated statistics be presented according to the directional principle. This Benchmark Definition recommends that compilers disseminate a geographic disaggregation of inward and outward FDI statistics according to the economy of the immediate investing economy and inward positions according to the ultimate investing economy on a standard basis. It also recommends several supplementary presentations, including inward FDI income by ultimate investing economy, outward FDI position by ultimate host economy, and outward statistics by the residency of the ultimate controlling parent. For FDI by industry, it recommends that inward statistics be disaggregated by the primary economic activity of the resident direct investment enterprise. For outward statistics, it puts first priority on the statistics disaggregated by the primary economic activity of the non-resident direct investment enterprise.
OECD Benchmark Definition of Foreign Direct Investment (Fifth Edition)
Abstract
494. The data disaggregated by partner economy and by industry are the two major dimensions required in presenting more detailed analyses of foreign direct investment (FDI) statistics under the directional principle. These detailed directional data are complementary with the FDI asset/liability aggregates also recommended by this Benchmark Definition and published as a part of the balance of payments and international investment position statistics. This chapter describes the recommended methodologies for disaggregating direct investment statistics by partner economy and by industry under the directional principle.
495. For a reporting economy, the following series disaggregated by geography are recommended. Inward and outward FDI statistics – financial transactions, income, and positions – should be compiled by immediate partner economy using the debtor/creditor principle with resident special purpose entities (SPEs) separately identified.1 In addition, inward FDI position statistics should be compiled on an ultimate investing economy (UIE) basis. It is also encouraged to compile three additional sets of supplementary statistics: 1) inward FDI income by UIE; 2) outward FDI positions by ultimate host economy (UHE); and 3) outward FDI statistics according to the residency of the ultimate controlling parent.
496. FDI statistics disaggregated by industrial activity should be compiled on a similar basis to that outlined for partner economy above with respect to SPE reporting. Each direct investor/direct investment enterprise (DIE) should be classified to a single industry, according to the main activity of the enterprise. As a standard requirement, industry classification of positions and financial and income transactions should be compiled according to the activity of the DIE (and on a supplementary basis according to the activity of the direct investor) for both inward and outward investment.
497. This chapter will first describe the methodologies for producing the statistics by partner economy, starting with the immediate partner economy, followed by the UIE and finishing with the UHE. The final section will discuss the disaggregation by industry. Additional details are provided in Annex 8.A and Annex 8.B.
8.2.1. Identifying the immediate partner economy
498. Direct investment statistics, i.e., financial transactions and positions and associated income transactions, by partner economy should be compiled under the directional principle described in Chapters 2, 3, and 4 and Annex 2.B. The statistics should be compiled for all partner economies individually as well as for major geographical regions or economic zones and currency or monetary unions. Individual economy and regional statistics are of prime importance to users of FDI statistics. In addition to supporting analysis, disaggregated bilateral data provide valuable information to compilers to identify and to examine at a more detailed level potential quality concerns when large differences arise in bilateral comparisons.
499. Notwithstanding the general guidance above for compiling FDI statistics, compilers have obligations to protect the confidentiality of individual enterprise data. Their obligations to protect data may at times limit their ability to disseminate transactions/positions at the level of an individual economy. When necessary, compilers are encouraged to combine individual economy data into higher aggregates that allow the release of as much geographical detail as their confidentiality constraints allow.
500. There are two principles that may serve as the basis for geographic disaggregation of direct investment transactions/positions: the debtor/creditor principle, which is recommended by this Benchmark Definition, and the transactor principle (which is not recommended). According to the debtor/creditor principle, geographic allocation is made according to the economy of residence of the DIE or direct investor, even if the funds are paid to or received from another economy. Under the debtor/creditor basis, changes in financial claims of the reporting economy are properly allocated to the economy of residence of the non-resident debtor, and changes in liabilities are properly allocated to the economy of residence of the non-resident creditor. According to the transactor principle, which is not recommended here, direct investment transactions should be allocated to the economy to which the funds are payable or from which the funds are receivable, even if this is not the economy of the DIE or the direct investor. Under the transactor basis, changes in the claims and liabilities are allocated to the economy of residence of the non-resident party involved in the settlement of the transaction (the transactor). Therefore, this Benchmark Definition recommends that the immediate investing economy and immediate host economy be identified based on the debtor/creditor principle.
501. Moreover, the standard geographical presentation also requires, on the same basis, that the reporting economy presents separately FDI financial transactions, positions, and associated income transactions of all its resident SPEs. This presentation of the core data facilitates users assessing the impact of SPE activity in individual economies by helping to identify FDI that has a real impact on the host economy by excluding FDI to and from resident SPEs.
502. The example illustrated in Figure 8.1 is included to demonstrate the basic approach towards preparing standard and supplementary disaggregated geographical data under the directional principle (although in reality ownership structures may be much more complex). Each box represents a different enterprise in a different economy and shows the value of the inward and outward position associated with each direct investment relationship; the arrows show the ownership links, including the percentage of ownership. A direct investor (A) in economy 1 (E1) indirectly owns 80% of DIE (E) in economy 5 through a SPE (D) in economy 4 and an operating affiliate (B) in economy 2. A direct investor (C) in economy 3 directly holds the remaining 20% in DIE (E).
Note: All values are in USD.
503. Table 8.1 shows the inward and outward positions that would be recorded for each economy on an immediate partner economy basis for non-SPEs and resident SPEs separately identified.
Note: All values are in USD. ‘n.a.’ refers to ‘not applicable’, as there cannot be any inward or outward FDI with the reporting economy on an immediate partner economy basis.
8.2.2. Identifying the ultimate investing economy
504. There is considerable interest from policymakers in identifying the ultimate investors in enterprises resident in their economies for political, economic, and structural analyses. As can be seen in Figure 8.1 and Table 8.1, the ultimate sources of investment for the DIEs in economies 4 and 5 are not all shown in the statistics on an immediate partner economy basis. Therefore, this Benchmark Definition recommends that compilers disseminate inward position statistics by the UIE. It is recommended that the entire inward equity positions and, if possible, the entire inward debt positions be attributed to the UIE.
505. There are two main methods to attribute inward positions to the UIE. The first method is referred to as the ‘winner takes all’ (WTA) method while the second is referred to as the ‘proportional ownership’ (PO) method. These two methods will be explained, compared and contrasted below. The guidance in the Integrated Balance of Payments and International Investment Position Manual, Seventh Edition (BPM7, (IMF, Forthcoming[1])), identifies the WTA as the preferred method, but the PO is acceptable. Compilers should indicate the method used in their metadata for inward position statistics by UIE. Box 8.1 discusses different concepts that have been applied to identifying the ultimate investor in FDI and activities of multinational enterprises/foreign affiliate statistics (AMNE/FATS). Annex 8.A provides practical guidance for compilers on specific situations they may encounter in compiling inward positions by UIE.
Identifying the ultimate investor in an enterprise is important to better understand the financial interdependencies between economies and how the benefits and risks of globalisation are shared across economies. FDI statistics and activities of multinational enterprises/foreign affiliate statistics (AMNE/FATS) have both developed the concept of ultimate investor/ultimate owner.1 The Manual on Statistics of International Trade in Services 2010 (2010 MSITS, (United Nations et al., 2012[2])) provides the most up-to-date international guidance for the compilation of FATS.2
The fourth edition of the Benchmark Definition (BD4, (OECD, 2009[3])) defined the ultimate investor as the enterprise that has control over the investment decision to have a FDI position in the direct investment enterprise (DIE) (paragraph 610). It recommended identifying the ultimate investor “by proceeding up the immediate direct investor’s ownership chain through the controlling links (ownership of more than 50% of the voting power) until an entity is reached that is not controlled by another enterprise” (paragraph 610).
BD4 discussed the ultimate controlling parent (UCP) in the context of the foreign direct investment relationship, saying that the direct investor could be the UCP or the immediate parent of an enterprise and stating that “there can be only one ultimate controlling enterprise, the UCP, which is at the top of the ownership chain, i.e., not controlled by any other entity” (paragraph 32). BD4 and the IMF’s Balance of Payments and International Investment Position Manual, Sixth Edition (BPM6, (IMF, 2009[4])) both recommend identifying the UCP of the resident fellow enterprise to extend the directional principle to fellow enterprises.
The 2010 MSITS defines the ultimate investor as the ultimate controlling institutional unit (UCI).3 The UCI is defined as “the first person in the chain – beginning with and including the first foreign parent – that is not controlled (that is, no other unit holds more than 50 per cent of the voting power of the UCI) by another person” (2010 MSITS, Box IV.2), where the first foreign parent is the first foreign entity in the chain of ownership. While the definition of UCI follows the ownership chain from the immediate direct investor (i.e., first foreign parent), it would identify the same UCI if it began at the DIE because FATS only cover control relationships. Thus, the 2010 MSITS defines the economy of the ultimate investor as “the country that ultimately controls, and therefore derives most of the benefits from controlling, the direct investment enterprise” (2010 MSITS, Chapter 4, Section H, paragraph 4.31).
This Benchmark Definition, along with the BPM7 and the 2025 System of National Accounts (2025 SNA, (United Nations et al., Forthcoming[5]), defines the UCP as the entity that ultimately controls an enterprise, identified by proceeding up the ownership chain from the enterprise through the controlling links (ownership of more than 50 per cent of the voting power) until an individual, household, or company that is not controlled by another company is reached. Then, this Benchmark Definition identifies the ultimate investor as either the UCP of the DIE under the winner takes all method or the UCP of the immediate direct investors under the proportional ownership method. The ultimate investing economy is then defined as the economy of the entity identified as the ultimate investor.4
Notes:
1. For more information on AMNE/FATS statistics, see Section 10.4.
2. In addition, Eurostat provides guidance to its member states for compilation of FATS in its European Business Statistics Compilers’ Manual for Foreign Affiliate Statistics, 2024 edition ( (Eurostat, 2024[6]), https://doi.org/10.2785/817091. .
3. The UCI replaced the ultimate beneficial owner (UBO) used in the 2002 MSITS (United Nations et al., 2002[7]). “The UBO is the first person in the chain - beginning with and including the first foreign parent - that is not majority owned by another person” (2002 MSITS, Box 8).
4. The concept of UCP is also used in application of the directional principle where the residency of the UCP of the fellow enterprise is used to determine the direction of the direct investment transaction and position; in estimating an upper bound on the amount of pass-through funds in an economy; and in defining a multinational enterprise.
Winner takes all method
506. The winner takes all (WTA) method is based on majority ownership of the DIE. The inward position is attributed to the economy of the ultimate controlling parent (UCP) of the DIE. The UCP is defined as the entity that ultimately controls an enterprise, identified by proceeding up the ownership chain from the enterprise through the controlling links (ownership of more than 50 percent of the voting power) until an individual, household, or company that is not controlled by another entity is reached. If there is no company, individual, or household that controls the resident company, then the resident company may be considered to be its own ultimate controlling parent. Under the WTA, there would only be one UIE identified for each DIE. While more complex ownership structures could be constructed with multiple chains of ownership, the principle that the entire FDI position in the DIE is reallocated to the UCP of the DIE should be followed.
507. FDI statistics cover both influence and control relationships, meaning that the WTA method will generally identify the compiling economy as the UIE when the DIE is not foreign-controlled.2 This could be the case if there is a domestic enterprise that holds a controlling interest in the DIE, or, if domestic investors as a whole own more than 50% of the voting power even if no single domestic investor controls the enterprise. This would lead to an overstatement of the amount of round-tripping into the economy (for more information on round-tripping, see Annex 6.A).3 Economies using the WTA method that are concerned that the amount of round-tripping is overstated may choose to publish a memorandum item with their inward position statistics by UIE that uses the PO method discussed below to more accurately reflect the amount of round-tripping into the economy.
Proportional ownership method
508. Under the proportional ownership (PO) method, the UIE is identified as the economy of residence of the UCP of the immediate direct investor into the DIE. That is, the UIE is the economy where the entity that controls the decision to make the direct investment into the DIE resides. If there is no enterprise that controls the immediate direct investor, then the immediate direct investor is effectively the ultimate investor in the DIE. Under the PO, there could be more than one UIE for a DIE. This is the method that was recommended in the fourth edition of this Benchmark Definition. While more complex ownership structures with multiple chains of ownership emanating from the DIE could be constructed, the main principle, that the ownership chain of each immediate direct investor is traced to identify the UIE, should be followed.
509. The PO method only identifies round-tripping when the UCP of the immediate direct investor is resident in the compiling economy. Therefore, it provides a more accurate picture of the amount of round-tripping in the economy than the WTA method.
Comparing and contrasting the WTA and PO methods
510. The example in Figure 8.1 can be used to illustrate the differences in the statistics that will result from using the WTA and the PO methods. Table 8.2 presents the results. As can be seen, the WTA method attributes the entire inward positions in economies 2, 4, and 5 to economy 1. While the PO method also attributes the inward positions in economies 2 and 4 to economy 1, it distributes the position in economy 5 between economies 1 and 3. While the identification of the UCP of an entity (either the DIE or the immediate direct investor) is often straightforward, Annex 8.A provides guidance for some special cases that have been encountered by statisticians compiling inward position statistics by UIE.
511. In practice, there is usually little difference between the two methods given that wholly-owned and majority-owned DIEs are often very common in FDI statistics. However, there are a couple of situations where the methods will yield different results. First, if there are restrictions on equity investment that prevent majority investment in particular industries or sectors, then the two methods will yield different results. In particular, the WTA method will not yield any information on the ultimate source of foreign influence investment in these industries or sectors. Second, the WTA method will not identify as sources of foreign investment economies that are more inclined to make minority investment. This could be an issue if these are economies that are of particular interest from a national security or development perspective.
Note: All values are in USD.
512. There are several advantages to each of the methods. First, the WTA method uses information on the UCP of the DIE that has already been collected to implement the directional principle for fellow enterprises. Second, it promotes alignment with the attribution by economy recommended in AMNE/FATS. For example, international guidelines for compiling FATS call for attributing resident foreign-controlled enterprises to the economy of the Ultimate Controlling Institutional Unit, which is determined by proceeding up a foreign affiliate’s chain of control until there is no further controlling interest. This could help the two sets of statistics to be used together although it should be noted that differences remain (see Section 10.4.7 for more information). FDI compilers could use information from AMNE/FATS as another source to identify the UIE, and it could be easier for users to understand the statistics if they are already familiar with the concept of ultimate investor from AMNE/FATS statistics. Finally, the WTA is more straightforward to implement than the PO as it is easier for the DIE to report their UCP rather than the UCP of their immediate direct investors and for the compilers to find this information using public information, commercial databases, or news reports.
513. There are also several advantages to the PO method. First, it identifies the UIE of all minority investment by identifying the UCP of the immediate direct investor. This means that it will better measure round-tripping as it will not incorrectly identify minority investment as controlled by resident investors as can happen under the WTA, but it will still correctly identify round-tripping taking place through minority investment if the UCP of the immediate direct investor is a resident of the compiling economy. Second, the PO method better reflects the financial structure of the multinational enterprise (MNE), in particular those investors that may be benefitting from investment in MNEs through minority ownership stakes. Third, the PO is more helpful for financial analysis as it accounts for all exposures to foreign risks as well as risk sharing through minority investment.
514. In summary, it could be argued that the WTA method better reflects the entity that is making the decisions regarding production in the compiling economy by identifying the entity that controls the DIE, but the PO method better reflects who is bearing the risks and reaping the rewards of the direct investment.
Inward FDI income by UIE
515. This Benchmark Definition recommends that economies compile statistics on inward FDI income by UIE on a supplementary basis. Just as capital can flow down an ownership structure, income can flow up it. This results in the similar issues as that of FDI positions. Specifically, presentations by immediate partner economy can obscure information on where the income is ultimately accrued within the MNE. Allocating income to the UIE reduces the influence of corporate structures on geographic disaggregation if the investment is controlled through intermediate economies. As with FDI positions, this usually reduces FDI values associated with offshore financial centres.
516. It is recommended that compilers use whichever method is used for compiling inward position statistics by UIE to also compile income statistics by UIE on a supplementary basis. Thus, if the WTA is used, it should be used for both, and likewise for the PO. While it could be argued that the PO method is more appropriate for FDI income because it reflects where the income ultimately accrues, consistency between the inward income and inward positions statistics by UIE was deemed to be more important.
8.2.3. Identifying the ultimate host economy
517. This Benchmark Definition recommends that economies compile outward FDI positions by UHE on a supplementary basis. As will be discussed further below, there is still more research that needs to be done on the concept of UHE, but the experiences of economies in developing these statistics will be useful in improving concepts, definitions, and methods. It is recommended that countries reallocate outward equity positions and, if possible, debt positions to the UHE. It is possible that compilers may find it feasible to produce UHE statistics for a subset of resident direct investors, such as operating entities or domestically-controlled direct investors. Given the number of issues that are still to be decided, economies that do publish outward position statistics by UHE are encouraged to include metadata that describes the definitions used and coverage of the statistics. In addition, the topic of UHE is on the Research Agenda (Annex D) for this Benchmark Definition.
518. Complex ownership structures can lead to distortions to the statistics of outward positions by immediate investing economy, most prominently through holding companies or SPEs in a third economy. This can make it difficult to use outward FDI statistics by immediate host economy to understand the geographic patterns of domestic MNE investment abroad.4 Statistics by UHE would also better illustrate the financial interdependencies between economies. For example, knowing the amount of investment broken down by UHE provides important information on how the risks of domestic direct investors are spread across economies and regions. MNEs do not always channel their investment through a single holding company or SPE but rather through a large number of companies (holding/SPE chains as well as operating units) until it reaches the entity with actual productive activities. Since the holding/SPE chains do not make an actual economic contribution to their host economy, it would be valuable to identify the economy in which operating (productive) DIEs are located.
519. To compile the outward position by UHE, the positions should be reallocated from the immediately held subsidiary to the economy(ies) of the subsidiary(ies) below it in the chain. Approaches to estimate FDI by UHE include:
Balance sheet information and ownership shares at each link in the chain can be used to show how the influence of the ultimate direct investor flows down the investment chain. Such a method could require a considerable amount of information on the ownership structure of the entire MNE. It could also require adopting certain conventions in the recording given the fungible nature of financing. This approach would provide the most detailed information on the financial structure of the MNE.
The value the direct investor would receive if that subsidiary were acquired by another enterprise; that is, measuring the ultimate investor’s position in the DIE in the specific economy where it is located. This presentation would provide a comprehensive view of the foreign operations of an economy’s direct investors for both direct and indirect ownership. This would result in statistics by UHE that are symmetric to the UIE. It would need to be determined if the value of the subsidiary should reflect only its value or the value of any entities below it in the chain. While the latter would be easier to collect, it would result in double-counting the value of entities lower in the chain as they would be reflected not only in their host economies but in the value of the host economies of subsidiaries above them in the chain.
Define the UHE as the last economy in the chain. This method alone would not be able to reallocate a proportion of the position in the immediate investing economy when the equity of the entity at the end of the chain is not enough to support the whole outward position. It can also attribute the position to smaller entities, such as retail operations, at the bottom of the chain, passing over larger entities, such as manufacturing facilities, further up the chain.
Define the UHE as the first operating unit even if it is not the end of an investment chain. This is the method recommended in BPM7 (BPM7, Annex 6). This would place less burden on respondents and compilers than the first two methods discussed above while still moving outward FDI from pass-through economies. A definition of operating unit would not simply be the first non-SPE but would rather identify the first unit whose purpose was not (only) administrative, management or holding assets to avoid reallocating positions to countries where these holding or administrative functions are prevalent. It would also be necessary to determine how to allocate positions across multiple affiliates when the SPE holds entities in multiple economies.
520. Annex 8.B provides more information on these possible definitions of UHE and methods to reallocate outward FDI positions using them. While it is not possible to use the example in Figure 8.1 to illustrate all four possible definitions of the UHE discussed above because more information on the balance sheet of each entity is needed for some of those methods, it is possible to illustrate the last economy in the chain and the first operating unit methods. If compilers in economy 1 used the last operating unit method, they would reallocate USD 80 of the position in economy 2 to economy 5 because its equity investment in enterprise E would support that amount; the remaining USD 40 would remain in economy 2. If economy 1 used the first operating unit method, the position would remain in economy 2 because enterprise B is an operating unit. Compilers in economy 2 may also reallocate the outward investment of enterprise B to the UHE. In this case, under both the last economy and first operating unit methods, they would reallocate the entire outward investment position to economy 5. Note that for economy 3, economy 5 is both the immediate and UHE for the outward by enterprise C. Countries that publish outward position statistics by UHE should describe the method used for the reallocation in accompanying metadata.
8.2.4. Pass-through funds
521. Pass-through funds in FDI are at the core of many of the problems that complex ownership structures pose for the use and interpretation of FDI statistics. As discussed in Chapter 6, separately identifying the financial transactions, positions, and associated income transactions to and from resident SPEs can be considered a lower bound on the amount of pass-through in an economy. This identifies funds passing through a subset of DIEs that are often established for this purpose. In Figure 8.1, the separate identification of resident SPEs would show the investment funds of the parent company A that pass through economy 4 but not those passing through economy 2 via enterprise B.
522. Presenting positions and transactions according to the residency of the UCP of direct investors can be used to estimate the extent of pass-through in an economy. Pass-through funds in an economy are carried out by direct investors that are foreign-owned and, so, are themselves DIEs. A simple breakdown of outward FDI positions and transactions based on the residency of the UCP would give an indication of the amount of outward FDI originating from that economy (i.e., the amount of outward FDI carried out by direct investors with resident UCPs) and the amount of funds and income passing through that economy (i.e., the amount of outward FDI carried out by direct investors with non-resident UCPs).
523. For inward FDI, the residency of the UCP shows the amount of round-tripping, which is a form of pass-through investment that takes place through non-resident entities but can inflate the outward and inward FDI figures of the compiling economy in a way that is similar to pass-through investment via resident entities. Combining the information of outward investment by direct investors with non-resident UCPs with the amount of inward investment by resident UCPs gives an upper bound estimate of the amount of funds that have passed through the economy.
524. Table 8.3 shows a breakdown of inward and outward FDI positions by the residency of the UCP. Result A identifies the amount of round-tripping in the economy (i.e., inward investment that has a resident UCP). B represents the amount of outward investment by domestic MNEs. D represents the amount of outward investment by foreign-owned direct investors – an indication of the amount of the inward investment into these direct investors that has passed through them to DIEs in other economies.5 C represents the inward investment by foreign investors, while C less D would give an indication of how much of this inward investment remains in the host economy.
525. As a result, this Benchmark Definition recommends that compilers produce outward FDI statistics according to the residency of the UCP of the direct investor on a supplementary basis. The topic of identifying pass-through funds is also added to the Research Agenda (Annex D).
8.3.1. General principles to identify industry classification
526. There is significant interest in the industry classification of both DIEs and their direct investors. DIEs and direct investors engage in a variety of economic activities. For a comprehensive economic analysis, enterprises should be grouped by type of economic or industrial activity. Under ideal circumstances, data should be available for compilers to classify both inward and outward direct investment on a dual basis to the industry of the DIE and the industry of its direct investor. Compilers are requested to present direct investment statistics on a directional basis and classified according to the economic activity of the counterparts involved.
527. If data for both inward and outward direct investment cannot be compiled on both of these bases, the Benchmark Definition recommends that data be compiled at least according to the activity of the DIE, for both inward and outward direct investment. In other words, inward direct investment should reflect the industry of the resident DIE and outward investment should reflect the industry of the non-resident DIE. Compilers are encouraged to provide supplementary statistics for inward and outward investment data according to the economic activity of the direct investor.
528. Each direct investor and each DIE must be classified into a single industry, even though many direct investors and DIEs are involved in a wide range of activities. The industry classification of the enterprise should be based on its principal economic activity. Ideally, the principal activity can be determined on the basis of which activity contributes most to the value added of the enterprise. In some cases, it is recognised that data on value added may not be available and that data on sales, revenues, or payroll or other bases may serve as a proxy. The detailed requirements of the standard and supplementary FDI series are given in Annex B, including the industry codes to be used.
529. A direct investor involved in a wide range of activities may make its overseas investment in each activity through numerous separate domestic subsidiaries specialising in a given activity, or it may make all of its overseas investment through a single domestic subsidiary established to handle overseas investment projects or a mixture of these approaches. For data that are presented based on the industry of the direct investor, the industry corresponds to the main activity of the direct investor, including all of its activities in its own economy of residence. This approach avoids distortions due to different organisational arrangements.
530. Results presented based on the industry of the DIE should be based on the reporting enterprise (if the reporting unit is a DIE and, at the same time, a direct investor, its industry classification should be based on the activities that it conducts and should exclude those conducted by its own FDI enterprises). It is recommended that the industry of the enterprise represents the principal activity of that enterprise, including all of its subsidiaries, associates and branches in its economy of residence.
531. When a change in the industry classification of a direct investor or DIE occurs, the reclassification of positions from one industry to the next should be recorded as arising from volume adjustments as discussed in Chapter 5.
8.3.2. Standard industrial activity presentations
532. The two standard presentations required for the geographical analysis (see Section 7.2.2) are also required for the industrial or economic activity analysis:
1. The standard industrial activity recommended is that all FDI financial transactions, positions and associated income transactions be compiled for the reporting economy on a directional basis, according to the industrial activity of the resident DIE involved for inward investment and of the DIE abroad for outward investment.
2. As for the partner economy allocation, the standard presentation requires the reporting economy presents on the same basis all FDI financial transactions, positions and associated income flows of all its resident SPEs.
8.3.3. International Standard Industrial Classification (ISIC)
533. Countries should compile data by industries that correspond to the major tabulation categories in the United Nations’ ISIC. The major categories in ISIC Revision 5 are:6
A. Agriculture, forestry and fishing
B. Mining and quarrying
C. Manufacturing
D. Electricity, gas, steam and air-conditioning supply
E. Water supply, sewerage, waste management and remediation activities
F. Construction
G. Wholesale and retail trade
H. Transportation and storage
I. Accommodation and food service activities
J. Publishing, broadcasting, and content production and distribution activities
K. Telecommunications, computer programming, consultancy, computing infrastructure, and other information service activities
L. Financial and insurance activities
M. Real estate activities
N. Professional, scientific and technical activities
O. Administrative and support service activities
P. Public administration and defence, compulsory social security
Q. Education
R. Human health and social work activities
S. Arts, sports and recreation
T. Other services activities
U. Activities of households as employers; undifferentiated goods- and services-producing activities of private households for own use
V. Activities of extra-territorial organisations and bodies
8.3.4. Holding companies
534. The classification of holding companies merits additional clarification. ISIC Rev. 5 continues to provide the ability to distinguish between two types of holding activities: financial and management activities:7
Holding companies engaged in financial services (activities of holding companies 6421): units that hold the assets (owning controlling-levels of equity) of a group of subsidiary corporations and whose principal activity is ownership of the group. The holding companies in this category do not provide any other service to the business in which the equity is held.
Holding companies engaged in management (activities of head offices 7010): this category includes the overseeing and managing of other units of the enterprise; undertaking the strategic or organisational planning and decision-making role of the enterprise. These units exercise operational control and manage the day-to-day operations of their related units. More precisely, they include:
head offices
centralised administrative offices
corporate offices
district and regional offices
subsidiary management offices.
535. If a holding company has subsidiary operations in the local economy as well as abroad, the approach to classifying the holding company should be based on the predominant activity of the local enterprise group.
536. To demonstrate this approach, consider that direct investor (A – the parent company) is located in economy 1 and establishes a holding company H1 in economy 2.
Case 1: Holding Company H1 has two manufacturing subsidiaries, company M1 and company M2, both in economy 2. In turn, company M1 has a holding subsidiary H2 in economy 3. The industry classification of H1 should be manufacturing (Figure 8.2).
Source: Based on OECD (2009[3]), OECD Benchmark Definition of Foreign Direct Investment, Fourth Edition (BD4), https://doi.org/10.1787/9789264045743-en.
Case 2: Holding Company H1 has two subsidiaries of which one is a resident holding company H2 in Economy 2 and a non-resident manufacturing subsidiary M1 in Economy 3. Holding subsidiary H2 in turn has a manufacturing subsidiary in Economy 4. The industry classification of H1 should be holding activities (Figure 8.3).
Source: Based on OECD (2009[3]), OECD Benchmark Definition of Foreign Direct Investment, Fourth Edition (BD4), https://doi.org/10.1787/9789264045743-en.
References
[9] Banco Central do Brasil (2018), Direct Investment Report – 2018, Banco Central do Brasil, Brasilia, https://www.bcb.gov.br/content/publications/directinvestmentreport/2017/dir_2017.pdf.
[11] Borga, M. and C. Caliandro (2018), Eliminating the Pass‐through: Towards FDI Statistics that better Capture the Financial and Economic Linkages between Countries, NBER Working Paper 25029, Cambridge, Massachusetts.
[8] Deutsche Bundesbank (2023), German foreign direct investment in 2021/2022, Deutsche Bundesbank, Frankfurt, https://www.bundesbank.de/en/press/press-releases/german-foreign-direct-investment-in-2021-2022-903736.
[6] Eurostat (2024), European Business Statistics Compilers’ Manual for Foreign Affiliate Statistics - 2024 Edition, Publications Office of the European Union, https://doi.org/10.2785/817091.
[4] IMF (2009), Balance of Payments and International Investment Position Manual, Sixth Edition (BPM6), International Monetary Fund, Washington D.C, https://doi.org/10.5089/9781589068124.069.
[1] IMF (Forthcoming), Integrated Balance of Payments and International Investment Position Manual, Seventh Edition (BPM7), International Monetary Fund, Washington D.C.
[13] IMF BOPCOM (2018), Final Report of the Task Force on Special Purpose Entities, International Monetary Fund, Washington D.C, https://www.imf.org/external/pubs/ft/bop/2018/pdf/18-03.pdf.
[10] Leino, T. and J. Ali-Yrkko (2014), How Well does Foreign Direct Investment Measure Real Investment by Foreign-Owned Companies?, Bank of Finland Research Discussion Paper No. 12/2014, May 2014.
[3] OECD (2009), OECD Benchmark Definition of Foreign Direct Investment, Fourth Edition (BD4), OECD Publishing, Paris, https://doi.org/10.1787/9789264045743-en.
[12] U.S Bureau of Economic Analysis (2023), Experimental Ultimate Host Economy Statistics for U.S. Direct Investment Abroad, U.S. Bureau of Economic Analysis, Washington, D.C., https://www.bea.gov/system/files/papers/BEA-WP2023-9.pdf.
[2] United Nations et al. (2012), Manual on Statistics of International Trade in Services 2010, United Nations publication, ESA/STAT/Ser.M/86, Sales No. E.02.XVII.11, New York, https://unstats.un.org/unsd/publication/seriesm/seriesm_86rev1e.pdf.
[7] United Nations et al. (2002), Manual on Statistics of International Trade in Services, United Nations publication, ESA/STAT/Ser.M/86, Sales No. E.02.XVII.11, New York.
[5] United Nations et al. (Forthcoming), 2025 System of National Accounts, United Nations et al., New York.
537. This Benchmark Definition recommends that economies compile inward FDI positions by ultimate investment economy (UIE) on a standard basis and inward FDI income statistics by UIE on a supplementary basis. It presents two methods to designate the UIE. Both methods rely on identifying the ultimate controlling parent (UCP) of an entity. For the winner takes all (WTA) method, the UCP of the direct investment enterprise (DIE) needs to be identified while under the proportional ownership (PO) method, the UCP of the immediate direct investor(s) needs to be found. The UCP is identified by proceeding up the relevant entity’s ownership chain through the controlling links (ownership of more than 50% of the voting power) until an enterprise is reached that is not controlled by another enterprise. While the identification of the UCP is often straightforward, there are some special cases that pose difficulties.
538. This annex will first illustrate the reallocation process for a more complicated ownership structure than discussed in Chapter 8. Then, it will provide guidance for these special cases, including special types of UCP’s and multiple minority ownerships. This guidance often draws on the experience of compilers of activities of multinational enterprises and foreign affiliate statistics (AMNE/FATS).
539. In the case of inward FDI, the following presents a general example of how positions can be attributed to the economy of the ultimate investor(s) in the resident DIE.
540. Under the PO method, the ultimate investor is the enterprise that has control over the investment decision to have a FDI position in the DIE. As such, the ultimate investor controls the immediate direct investor. It is identified by proceeding up the immediate direct investor’s ownership chain through the controlling links (ownership of more than 50% of the voting power) until an enterprise is reached that is not controlled by another enterprise. If there is no enterprise that controls the immediate direct investor, then the direct investor is effectively the ultimate investor in the DIE.
541. The economy in which the ultimate investor is resident is the ultimate investing economy (UIE) for the investment in the DIE. It is possible that the ultimate investor is a resident of the same economy as the DIE. This is round-tripping.
542. The entire FDI position that is attributed to the economy of residence of the immediate direct investor is reallocated to the UIE.8 Where there is more than one immediate direct investor in a DIE, the entire inward FDI position of each immediate direct investor is reallocated to the respective UIEs based on the UCP of each immediate direct investor.
543. Annex Figure 8.A.1 illustrates ownership links for DIEs in economy 4.
Source: Based on OECD (2009[3]), OECD Benchmark Definition of Foreign Direct Investment, Fourth Edition, https://doi.org/10.1787/9789264045743-en.
544. According to the standard presentation by economy (on an immediate investor basis), Economy 4 would show USD 40 million of inward FDI positions from Economy 3. Under the PO method, the presentation of inward positions by UIE would be as follows:
the USD 10 million investment from C into F would be reallocated to Economy 2 as enterprise B is the ultimate investor in enterprise F (as the UCP of enterprise C)
the USD 4 million investment from C into G would be reallocated to Economy 2 as enterprise B is the ultimate investor in enterprise G (as the UCP of enterprise C)
the USD 6 million investment from D into G would remain allocated to Economy 3 as enterprise D is the ultimate investor in enterprise G (no enterprise exerts control over enterprise D, so enterprise D is its own UCP)
the USD 20 million investment from E into H would be reallocated to Economy 4 as enterprise I is the ultimate investor in enterprise H (as the UCP of enterprise E).
545. The standard presentation for Economy 4 would record USD 14 million inward FDI from Economy 2 as the UIE, USD 6 million inward FDI from Economy 3 as the UIE, and USD 20 million inward FDI from Economy 4 as the UIE (round-tripping). No investment in Economy 4 is reallocated to Economy 1 as enterprise A is not the UCP of any of the immediate direct investors in Economy 4.
546. From this example, it can be seen that the entire immediate direct investment position (i.e., between the immediate direct investor and the DIE) is reallocated to the UIE – the percentage ownerships in the chain of control above the immediate direct investor do not affect the value recorded. It can also be seen that where a DIE has multiple direct investors (for example, enterprise G), the investment from each immediate direct investor is reallocated to the respective UIEs based on the UCP of each of the immediate direct investors.
547. Under the WTA, the UCP of each DIE in economy 4 would be identified and the entire inward position in that DIE would be reallocated to the economy of that UCP. A key difference between the PO and WTA is that, under the WTA, there can only by one UIE for a DIE while, under the PO, there can be more than one UIE. Under the WTA method, the presentation of inward positions by UIE would be as follows:
the USD 10 million investment from C into F would be reallocated to Economy 4 as enterprise F is domestically controlled (this is a case where the estimated value of round-tripping is inflated)3
the USD 4 million investment from C into G and the USD 6 million investment from D into G would be allocated to Economy 3 as D is the UCP of enterprise G
the USD 20 million investment from E into H would be reallocated to Economy 4 as enterprise I is the UCP of enterprise H (this is a case of genuine round-tripping).
548. Sometimes the UCP can be identified, but the specific form that it takes or where it resides can pose difficulties for the statistics by UIE.
Tax havens, offshore financial centres, and special purpose entities
549. When the UCP is in a tax haven or offshore financial centre9, it is recommended that such an entity is identified as the ultimate investor only if it fulfils the definition of an institutional unit and has a substantial presence in the jurisdiction. If the UCP does not have a substantial presence in the jurisdiction, it is unlikely to be the top of the chain, and more research is needed to find the actual top, if possible.
550. If a special purpose entity (SPE) is identified as the UCP, it is unlikely to be the actual top of the ownership chain and more research is needed to find the actual top of the chain, if possible.
Ultimate investor is a natural person
551. When the UCP is a natural person, it is best to identify the UIE as the economy of residence of this person, because that is in keeping with the concept of residency in macroeconomic statistics. However, it may not always be clear where the natural person resides. In this case, the nationality of the natural person could be used as a proxy, followed, as a last resort, by the residency of the company directly owned by the natural person. Some compilers might alternatively prefer to leave the position under 'unallocated' or allocate it to the immediate investor when the residency of the ultimate owner as a natural person is not easily identifiable.
Ultimate investor is a collective investment institution
552. Sometimes, it may be difficult to determine the residence of certain direct investors even if their identity is known, as in the cases of collective investment institutions (CIIs). CIIs pool funds from individual investors that may reside in different economies and are often managed by a professional investor. Their structure is driven by tax and regulatory considerations, so they may be resident in one economy but be managed from another jurisdiction while the investment could pass through entities in another economy. In macroeconomic statistics, CIIs are institutional units, and the economy where it resides should be the UIE.
553. In some cases, the DIE has several investors but none of them has a majority ownership. Those structures are characterised by multiple minority ownership of shareholders from abroad and in the home economy itself. A special case of multiple minority ownership is equally-owned DIEs, often in the form of joint ventures. All of these cases pose challenges for the application of the WTA as there is no immediate single non-resident or resident majority owner of the DIE. In these special cases, under the WTA, compilers might consider the DIE as being foreign-owned when looking at the combined position of its multiple (immediate) foreign direct investors. For compilers applying the PO, the immediate direct investor is considered to be the UCP if it is not controlled; therefore, these cases do not pose any specific challenge.
Case 1: Multiple minority ownership
554. In Case 1, shown in Annex Figure 8.A.2, none of the non-resident or resident investors own more than 50% of the voting rights. Hence, this case poses a challenge for the application of the WTA. While the 40% of the voting power held by investors in the same economy as the DIE represents the largest ownership share by one specific economy, it is recommended that the DIE be designated “foreign-controlled” because the majority (60%) of the voting power is held abroad. If there is evidence that both foreign economies are acting in concert and speaking in one voice, then the largest non-resident investor could be deemed the UCP (UCP of DI2 in economy 2).10 Given the difficulties that compilers may face to apply this criterion, they might still choose to allocate the position to the largest foreign investor or to leave the position unallocated (specific metadata should explain the procedure chosen).
Note: ‘DI’ refers to direct investor.
Case 2: Equally-owned enterprises (i.e., joint ventures)
555. When the voting power in a DIE is held equally between two or more entities, it is not possible to identify the UCP of the DIE (Annex Figure 8.A.3). In some shareholding models, the capital shares differ from voting power. If in the case of equal voting power in which one of the direct investors has larger capital shares, then the UIE should be the economy of this investor. In addition, an investor is also to be classified as a de facto dominant controller if any available information (e.g., annual report, administrative sources) implies that this investor has the greatest influence on the DIE.11 If any of this information is difficult to obtain for the compiler, then another alternative could be to allocate the total FDI value to the economy with the higher aggregated FDI position by UIE because the relative “misallocation” error is smaller than for the other economy. Finally, the compiler could choose to leave the position unallocated.
Note: ‘DI’ refers to direct investor.
556. The presented approaches could also be used for similar enterprise structures, e.g., three times 33% or four times 25%. However, if in this case, two (three) owners of the DIE are resident in the same economy and, thereby, generating a majority for that economy overall, it is recommended that this economy should be considered as the UIE. In Annex Figure 8.A.4, UCP of DI1 and UCP of DI2 are resident in the same economy and therefore, the majority of shares (in total 66%) is in that economy (economy 1), which would be the UIE.12
Note: ‘DI’ refers to direct investor.
Case 3: The majority of shares in the DIE is actively traded
557. In Annex Figure 8.A.5 below, the majority of shares in the DIE is actively traded (55% of the shares are owned by resident and non-resident portfolio investors). The DIE has no foreign controlling entity as there is no single immediate or indirect foreign investor holding more than 50% of the voting power. At the same time, less than 50% is held by foreign investors, meaning there is no combined foreign control. There is also no resident investor who individually reaches 10% of the voting power, hence no resident investor can qualify as a direct investor. In this case, compilers following the WTA approach might consider leaving these FDI positions by UIE unallocated as it cannot be argued that there is a domestic entity that can exercise any control or even influence on the DIE.
558. However, some compilers following the WTA might still prefer to allocate the 45% of the foreign-owned shares in the DIE to the country where the entity controlling the largest minority foreign direct investor resides (i.e., UCP of DI3 in Economy 3), or to the country of the largest minority foreign direct investor if the UCP cannot be identified (i.e., DI3 in Economy 3). The latter situation could arise if the UCP was resident in a different economy than DI3 as compilers of the WTA cannot generally climb the control chain for minority direct investors.
Note: ‘DI’ refers to direct investor. ‘ROW’ refers to the rest of the world. *In this specific case, 55% of the shares in the DIE are owned by resident and non-resident portfolio investors (free float).
559. This annex further elaborates on the definitions of ultimate host economy (UHE) in Chapter 8 and describes methods that could be used to produce outward direct investment statistics by UHE, which have focused on equity positions so far. When considering compiling these supplementary UHE statistics, each statistical office should balance the needs of users with the resources required and information available to determine if they should produce them and which method to use.
560. The methods to produce UHE statistics can be grouped under three major themes that are covered in more detail in the rest of this annex. The list of methods included are not an exhaustive list; as research on this topic continues, it is possible that other methods or combination of methods will be developed. In addition to the possible advantages and disadvantages discussed, confidentiality concerns could also make a compiler favour one method over the other. The extent of round-tripping could also influence which method to use.13 Some methods by construction might not allow for the existence of round-tripping, while others might do a better job of capturing it for economies where it is significant. For example, for some compilers the last unit, or last-operating unit, approach could lower the amount of the UHE position that gets classified in the compiling economy if the compiling economy sees significant pass-through for entities that are ultimately controlled by resident entities.
561. Push-down methods, as the name implies, move (push) the position down the FDI ownership chain to a single entity, without considering intermediate or subsequent linkages, if any. These methods should allow for a reallocation, to varying extents, of the outward position from financial centres to economies where the productive activities of multinational enterprises (MNEs) are carried out. These methods should be less data-intensive and relatively easier to implement and maintain than methods that consider all the units in the ownership structure. However, complications can arise with complex ownership chains seeing multiple owners of some direct investment enterprises (DIEs) within the enterprise structure.
562. On the other hand, a drawback of these methods is that precisely because they do not consider all the units in the chain, they could lose detail on the global financial structure of MNEs, which some users and policymakers might be interested in. Another drawback of these methods is that there could be cases where the equity of the DIE further down the chain is not large enough to support the whole outward position of the directly held DIE. Push-down methods are further outlined below.
First operating unit method
563. The first-operating unit method is recommended largely for practical reasons. The first-operating unit can be the first non-holding company and/or non-special purpose entity (SPE) going down the ownership of the directly held DIE.14 It could also be one whose purpose was not (only) administrative, management and/or holding assets. While there is likely considerable overlap between all these distinctions, subtle differences between them can yield different results, empirically but also conceptually.15 Information on the purpose of a DIE can be hard for compilers to collect, especially as the reporting MNEs themselves may not know it. From a conceptual angle, looking through both holding companies and SPEs should approximate what many users think of as “true” FDI, i.e., the enterprise in the chain that is the main reason the ownership chain exists in the first place. An example of this method is available in Deutsche Bundesbank (2023[8]).
564. From a practical angle it might be more feasible to identify non-resident holding companies than non-resident SPEs (or vice-versa). A compiler can choose to look through just holding companies if it is more feasible than non-resident SPEs, especially if there is a lot of overlap between the two. With modern complex ownership structures, there is not always going to be a factory (or hotel, mining operation, etc.) at the end of, or somewhere along, each ownership chain. There could even be an extreme case where all of the DIEs in the chain meet the SPE definition. MNEs can set up DIEs to hold portfolio-type investment or lease intellectual property, for example, and these DIEs can account for a significant amount of the outward position that should not “disappear” when assigning positions to UHE. The existence of these entities does entail that even with UHE statistics, some of the position will remain in financial centres.
Last unit method
565. The last-unit method assigns the position of the directly held DIE, no matter whether it is operating or not, to the unit at the bottom of the chain. Similar to the first-operating method, it should re-assign part of the position in financial centres to other economies where productive activities are carried out, but some is likely to remain in the former for the same reasons as the first-operating unit method. An example of this method is found in Banco Central do Brasil (2018[9]).
566. The last unit method is more advantageous than the first-operating method in cases where the first operating unit mixes operating and holding activities (that is, a "near-SPE") and is followed by a chain of SPEs or holding companies before there is another non-SPE DIE at the bottom of the chain.16 That advantage also applies to cases where there is a chain of “near-SPEs” before reaching a DIE at the bottom of the ownership chain that carries out most of the operating activities. On the other hand, one potential disadvantage when compared to the first-operating method includes the possibility of attributing the position to entities at the bottom of the chain engaged in activities auxiliary to the main activity of the MNE that is carried out by entities further up the chain; for example, retail or service-providing entities at the bottom of the chain, passing over manufacturing entities further up in the chain. Another drawback of these methods is that there could be cases where the equity of the DIE at the bottom of the chain is not large enough to support the whole outward position of the directly held DIE.
567. Methods that look at the financial structure of MNEs could provide the best picture in terms of how direct investors organise their global financial structure. Under these methods, the position of the directly held DIE is distributed along the DIEs in the chain, reflecting how much of the position “stays” at each tier.
568. While these methods could arguably provide a better picture of MNEs’ global financial structure, they are the most information intensive because compilers would need accurate and detailed information of all the DIEs of the MNE. These methods could also create difficulties from a statistical compilation point of view due to the complex organisational structures MNEs use. For example, distributing the ownership of the directly held DIE can be tricky if a unit along the chain has negative equity or a DIE in a lower tier is larger than the one above, so any method needs to account for these possibilities.17
Pass-through method
569. The pass-through method makes use of the direct investors’ equity position in each DIE that “passes through” to be invested in another DIE further down the chain. This method accounts for the possibility of negative positions and external financing (Leino and Ali-Yrkko, 2014[10]; Borga and Caliandro, 2018[11]). However, it is complex. The value of the direct investor’s pass-through equity in each DIE is a function of the equity that the direct investor has in it (“inward equity”) and the equity the DIE has in units further down the chain (“outward equity”). Pass-through equity can then be calculated for each DIE using the conditions below for pass-through equity depending on inward equity, and outward equity. Once inward equity and pass-through equity have been calculated, the adjusted equity position of each DIE is calculated as inward equity minus pass-through equity.
where IE = Inward equity; OE = Outward equity; and PTE = Pass-through equity.
570. In the example shown in Annex Figure 8.B.1, the direct investor owns DIEs A, B, and C. DIE A is 100% directly owned by the direct investor, has an equity position of USD 100 million, and has USD 50 million in equity in affiliate B. DIE A itself owns 80% of B, which has USD 40 million in the equity of C, of which it has complete control (100%).
Notes: IE=Inward equity; OE=Outward equity; PTE=Pass-through equity.
571. On an immediate basis, the compiler of the resident direct investor would show USD 100 million of the outward position in the economy of DIE A. However, after applying the pass-through method, the compiler would now show USD 50 million in the economy of DIE A (inward equity less pass-through equity gives USD 100 million less USD 50 million), USD 18 million in the economy of B (USD 50 million less USD 32 million), and USD 32 million in the economy of C (USD 32 million less USD 0 million) on an UHE basis.
Parent’s equity share of affiliates’ assets method
572. The parent’s equity share of affiliates’ assets method makes use of the direct investor’s (“parent”) equity share of the DIEs’ (“affiliate”) assets. The reallocation method involves transferring, at each ownership tier, that proportion of the equity position in higher-tier DIEs that represents investment in lower-tier DIEs to the DIEs at the next tier down in the ownership chain.
573. The direct investor’s equity position in first-tier DIEs is reallocated to lower-tier DIEs in proportion to the equity ownership shares connecting the DIEs to each other and to the direct investor. This is also done in proportion to the share of the total financing of the DIEs accounted for by equity funding supplied directly or indirectly by the direct investor. The equity position for first-tier DIEs is then adjusted by subtracting that portion of the unadjusted position that has been reallocated to the lower-tier DIEs. Where the ownership chain extends beyond the second tier, this process becomes an iterative one, in which the reallocation procedure is repeated at each tier.
574. The example depicted in Annex Figure 8.B.2 consists of a single first-tier DIE (affiliate A) and a single second-tier DIE below it (affiliate B). The parent company’s (direct investor's) unadjusted equity position in affiliate A is calculated as the product of the parent’s percentage equity stake (100%) and the total owners’ equity of affiliate A (USD 200), which is USD 200. The other USD 50 in the total assets of affiliate A is financed by borrowing by the affiliate.
1. Equals parent’s percentage equity stake in affiliate A times owners’ equity of affiliate A.
2. Equals unadjusted equity position for affiliate A minus adjusted equity positions for affiliate B.
3. Equals (1) parent’s percentage equity stake in affiliate A times owners’ equity of affiliate A divided by (2) total assets of affiliate A.
4. Equals parent's percentage equity stake of affiliate A times affiliate A’s percentage equity stake in affiliate B times owners’ equity of affiliate B.
575. The parent’s equity share of affiliate A’s assets calculated as the parent’s USD 200 equity investment divided by the USD 250 in total assets of affiliate A, which is 80%. This percentage and affiliate A’s equity investment in affiliate B (100%) are used to compute the proportion of the parent’s USD 200 direct equity investment in affiliate A that is reallocated to affiliate B.
576. Computationally, the parent’s adjusted equity position in affiliate B is the product of (1) affiliate A’s equity stake in affiliate B (100%), (2) affiliate A’s parent’s equity share of affiliates’ assets (80%), and (3) the total owners’ equity of affiliate B (USD 100), which is USD 80. The logic behind this formula is explained next. The product of (1) and (3) represents affiliate A’s equity investment in affiliate B. However, we are trying to calculate the direct investor’s indirect equity investment in affiliate B, which is less than this amount because the direct investor has a claim on (just) 80% of affiliate A’s assets. Therefore, (1) times (3) must be multiplied by (2) to derive an estimate of the amount of the direct investor’s equity position in affiliate A that should be reallocated to affiliate B. The parent’s adjusted equity position in affiliate A is thus calculated as the USD 200 unadjusted equity position less the USD 80 adjusted equity position in affiliate B, which is USD 120. The parent’s total direct investment position in affiliate A would be this amount plus any portion of the affiliate’s USD 50 in debt that represented borrowing from the parent.
577. The outward position by UHE can also be produced by apportioning the position using a third dataset, such as activities of multinational enterprises/foreign affiliates statistics (AMNE/FATS). For example, if X% of an outward FATS (OFATS) related variable is in economy A, then X% of the outward position gets assigned to economy A. Related variables that could be used to scale the position include employment, physical assets, sales/turnover, or some combination of these.
578. The apportioning method is likely the easiest method to implement if the compiler already produces a dataset that can be used to apportion since it does not require the use of microdata (company-specific ownership chains). However, this method, like the push-down methods, is unlikely to provide a full picture of the financial structure of MNEs. Additionally, the resulting UHE statistics could be significantly different depending on which related variable is used to apportion. If the position is apportioned using employment, the position places more weight on economies with labour-intensive DIEs, while capital-intensive DIEs have higher weights with variables like physical assets.
579. It is possible to combine several aspects of the methods mentioned above to develop hybrid methods for producing UHE statistics. As also stated above, compilers should consider data and resource availability along with the needs of users when determining which method to use, and a combination of characteristics of different methods could be the best approach.
Last “operating” unit
580. The last “operating” unit method combines aspects of both push-down methods (i.e., the first-operating unit method and last unit method). Under this approach, the position of the directly held DIE is assigned to the last operating unit down the ownership chain. Looking beyond the first-operating unit could help with issues of near-SPEs towards the top of the chain that might otherwise get positions assigned to them. It also avoids assigning the position to smaller DIEs at the very bottom of the chain that do not engage in productive activities, passing over larger entities further up in the chain, such as manufacturing facilities, although it could still assign the position to a smaller productive DIE at the bottom of the chain that is not large enough to support the entire position.
Micro-level apportioning
581. The outward position could also be apportioned at the micro-level, which can be particularly useful if the compiler has some information on DIEs along the chain but not enough for methods requiring detailed knowledge of MNE financial structures. For example, the position of the directly held DIE could be distributed along the chain by apportioning using a related variable available to the compiler, such as physical assets. The directly held position would then be distributed to the DIEs along the chain based on the percentage of the related variable that each DIE accounts for out of the total of that related variable for all the DIEs in the chain. Alternatively, the compiler could use the information on the ownership chain to determine the first-operating (or last) unit of the chain, and use the percentage of the related variable the DIE accounts for out of the total of that related variable for all the DIEs of the MNE (in this last case, aspects from both push-down and apportionment methods are combined).
582. The United States Bureau of Economic Analysis (2023[12]) provides a comparison of the different methods discussed above using data on United States outward FDI.
583. Currency unions (CUs) and economic unions (EcUn) play an increasingly important role in the world’s economy. This Annex summarises the specific features of the compilation and uses of CUs/EcUns’ foreign direct investment (FDI) statistics.
Definition of currency and economic unions and corresponding territories
584. For statistical purposes, a CU is defined as a group of economies that adopts a common currency and has a central decision-making body (usually a currency union central bank) with the authority to issue the legal tender of the area and conduct a single monetary policy.18 A CU is established by means of a formal intergovernmental legal agreement (e.g., a treaty).
585. The economic territory of a CU consists of the economic territory of the CU economies that comprise the CU, plus the currency union central bank. Any other regional organisations that comprise the same or a subset of the same economies are included in the CU.
586. For statistical purposes, an economic union (EcUn) is defined as a group of economies established by means of an intergovernmental legal agreement among sovereign economies or jurisdictions with the intention of fostering greater economic integration. In an economic union, some of the legal and economic characteristics associated with a national economic territory are shared among the different economies/jurisdictions. These elements include: (1) the free movement of goods and services within the EcUn and a common tax regime for imports from non-EcUn economies (free trade zone); (2) the free movement of capital within the EcUn; and (3) the free movement of (individual and legal) persons within the EcUn. Also in an EcUn, specific regional organisations are created to support the functioning of the EcUn under points (1) to (3). Some form of cooperation/coordination in fiscal and monetary policy usually exists within an EcUn.
587. An EcUn territory consists of the economic territory of the member economies/jurisdictions and the regional institutions that comprise the same or a subset of the same economies.
Specific issues arising for the FDI statistics of the currency/economic union
588. The concept of FDI statistics does not differ per se for currency or economic unions, taken as a single economic entity, from that of individual economies. As other subscriptions to the capital of international organisations, the subscription to the currency union central bank, or to regional organisations is not recorded under direct investment even though these contributions may exceed 10% of the voting shares of these institutions.19 Furthermore, specific legal status may exist for groups of companies in CUs or EcUns, as illustrated by the “European Company status”. Such arrangements may have consequences on national FDI statistics, e.g., by facilitating the transfer of assets and activities across economies, but the involved FDI cross-border20 transactions and positions would generally be recorded according to the standard methodology as described in this Benchmark Definition.
589. The compilation of FDI statistics could be similar to that of individual economies, if a specific data collection were performed on the basis of the residency within the union. However, such an approach may duplicate the data collection with the national statistics, still necessary in such situations (if only because various key national policies, e.g., tax policies, are under the national responsibility, within such unions). In this context, currency and economic union FDI statistics are usually based on the sources used for the compilation of national FDI statistics. This raises a number of specific compilation issues, which this Annex intends to briefly describe.
Compiling “extra-currency/economic union” transactions and positions
590. The compilation of FDI statistics for the currency/economic union by simple addition of gross national data would unduly inflate their gross stocks and flows because these would include transactions and positions between members. In principle, the addition of only the net national transactions and positions would solve this problem but would provide only net direct investment stocks and flows (i.e., assets minus liabilities) for the currency/economic union. However, this approach would not allow the compilation of assets and liabilities for the currency/economic union because transactions between economies belonging to the union would be included on both sides. Furthermore, it would be subject to distortions in case of asymmetric recording of positions/transactions within the currency/economic union. In this context, currency/economic union FDI has to be compiled as the sum of its residents’ positions/transactions towards non-residents of the union (thereafter: “extra” positions/transactions), plus those of the institutions of the union.
591. National contributions for compiling financial transactions data in currency and economic unions are allocated along the debtor-creditor approach, as opposed to the transactor approach, as a way to ensure bilateral symmetry. This convention means that cross-border transactions in financial claims are allocated to the economy of residence of the non-resident debtor, and cross-border transactions in liabilities are allocated to the economy of residence of the non-resident creditor.
Sector and industry attribution
592. A legal entity resident in one economy is always a separate institutional unit if one or more of its owners is resident in another economic territory. In case all its owners were resident in another economy, the sector of this entity would be defined only according to its own characteristics.
593. However, when compiling FDI statistics for the currency/economic union as a whole, such entities having all of their owners in other economies of the union would in principle be treated as resident of the same economy (i.e., the CU or EcUn) and, thus, become ancillary to their parent. As such, they would be allocated to the institutional sector of their parent company. It is recognised, though, that this reclassification may raise significant practical difficulties, and would need a close cooperation between compilers in the economies involved.
594. This Benchmark Definition recommends that when the enterprise is part of a local enterprise group, the industry classification be that of the local enterprise group's primary activity. The primary activity of the "local group" at the level of a CU/EcUn may however be different from the primary activity of the “local group” at national level. Although it might be, in principle, possible to construct the FDI statistics of a CU/EcUn according to the primary activity of the "local group" at the level of a CU/EcUn, the loss of valuable detailed information and the increase in the error due to reclassifications of potentially substantial amounts would render such statistics for the CU/EcUn significantly less useful.
Determination of the ultimate investing economy
595. The compilation of inward position statistics by ultimate investing economy (UIE), as recommended by this Benchmark Definition, is complicated for a currency/economic union because it is necessary to avoid duplication.
596. As an example, an investment from economy 1 to economy nth may be performed, by means of a single investment chain, via entities resident in economies 2 and 3, both in the same CU/EcUn and to which neither A nor Z belongs (see Annex Figure 8.C.1). It may be expected that national statistics in economies 2 and 3 would both identify the UIE of their inward FDI positions as economy 1. However, for the CU/EcUn FDI statistics, the position towards A should be reported only once. One approach to ensure this outcome is that UIE positions would only be reported in CU/EcUn statistics if they refer to extra CU/EcUn positions from the point of view of the immediate counterpart; in the example, only B would report the UIE as economy 1 since it has an inward investment position with an entity outside of the EcUn. Another possibility is that economies remove the funds passing through their economy and only report the positions that remain in their economy by UIE.
Application of the extended directional principle
597. The application of the extended directional principle requires determining the resident or non-resident status of ultimate parents from the perspective of the currency/economic union.
Need to coordinate compilation practices
598. International statistical standards intentionally provide only a set of core definitions and concepts and leave some flexibility to national compilers to define their compilation methods and detailed compilation practices and adapt them to their specific environment. For instance, this Benchmark Definition allows for several approaches regarding the valuation of unlisted equity. While economies not belonging to a CU/EcUn may define their compilation practices independently from other economies, although still taking on board the interest of international comparability of these data, those belonging to a CU/EcUn have to coordinate closely in order to produce meaningful FDI data for the union as a whole.
Need to coordinate the recording of large transactions/positions
599. As explained above, aggregated FDI statistics are usually compiled from data collected by national compilers. Beyond the close coordination of compilation practices, the data quality of the union’s FDI statistics requires the coordination of the recording of some large transactions (and resulting positions) involving several economies. For example, a company in economy 1 may disinvest from economy 2, belonging to a CU/EcUn, to invest in another economy 3 of the same CU/EcUn (economy 1 is not a member). While such a development should have no impact on the FDI statistics of the CU/EcUn, it may be important in practice to coordinate the recording of the corresponding transactions/positions of economies 2 and 3 to ensure that they are compiled in exactly the same way (especially in terms of timing).
600. This coordination is not straightforward as national statistical confidentiality rules usually do not allow national compilers to share confidential information with statistical bodies outside their national borders. To facilitate the maintenance and improvement of data quality through data checking, the establishment of an appropriate statistical data confidentiality regime is desirable at least, and probably essential. Such a regime would ideally address the sharing and protection of confidential data both (i) among national compilers and (ii) between national compilers and the CU/EcUn compilers in order to enable the compilation of consistent and reliable aggregated union statistics.
601. While national FDI statistics still play a prominent role for policy makers in the economies participating in such unions, where tax and/or some other policies are still defined at national level, FDI statistics for CU/EcUns are important too. This allows for the overall assessment of the benefits of the union and supports the conduct of its policy, concerning trade, competition or labour rules, for example. The analysis of the developments of globalisation within and outside the union, or the determinants to GDP growth for the union as a whole may also strongly benefit from information on inward FDI.
602. Other economies may find it analytically useful to identify certain currency and/or economic unions among their partners. Corresponding data should be equal to the sum of positions/transactions towards the economies belonging to the corresponding CU or EcUn, plus possibly institutions of the CU or EcUn where appropriate.
603. In case the composition of the CU or EcUn would change over time, compilers may decide either to show back data only according to the latest composition of the CU/EcUn, or to show the CU/EcUn according to its composition at each point in time, and to take the changes in composition into account by means of a reclassification.
← 1. Whenever possible, compilers are encouraged to present their FDI statistics by partner economy and by industry excluding resident SPEs and for resident SPEs separately allowing the disaggregated series by partner economy and by industry for all entities to be derived. This would place the lowest level of confidentiality restrictions on series excluding resident SPEs and better support analyses of the impact of FDI.
← 2. In some cases, such as when the majority of shares are actively traded, it is not clear that there is domestic control over the DIE; compilers who follow the WTA method can use various approaches to identify the ultimate investor (see Annex 8.A).
← 3. While technically, cases of foreign investment with national control do not correspond to round-tripping of funds, they will be perceived as such by users of FDI statistics by UIE. Hence, treatments of those cases by compilers using the WTA are referred to in this chapter as ‘inflated round-tripping’.
← 4. It also causes issues with the economic activity patterns of outward FDI statistics. This issue is also on the Research Agenda (Annex D).
← 5. This ignores the possibility of negative investment positions as well as of external financing raised by the foreign-‑owned direct investor. These issues can be explored as part of the research into this topic recommended in the Research Agenda (Annex D).
← 6. Recognising the revisions of ISIC, the Benchmark Definition recommends that compilers should adjust industry classifications to all subsequent revisions of ISIC.
← 7. The ISIC definition of holding companies, applied within the industry classification of FDI statistics is different from the definition in the System of National Accounts (United Nations et al., Forthcoming[5]) used for institutional sector classification.
← 8. While the guidance calls for reallocating both the equity and debt positions, some compilation systems may use data sources for intra-company debt that do not enable the identification of the UIE. In these cases, only the equity position should be reallocated.
← 9. An offshore financial centre is defined as a jurisdiction in which the majority of the financial transactions are conducted by resident financial corporations on behalf of non-resident clients. Lists of offshore financial centres are widely available.
← 10. This approach is aligned with the treatment recommended in the 2010 Manual on Statistics for International Trade in Services (2010 MSITS, (United Nations et al., 2012[2])) and in the European Business Statistics Compilers’ Manual for Foreign Affiliate Statistics (Eurostat, 2024[6]).
← 11. This approach is aligned with with the treatment recommended in the 2010 MSITS and in the European business statistics compilers’ manual for foreign affiliate statistics (Eurostat, 2024[6]).
← 12. This approach is aligned with the treatment recommended in the 2010 MSITS and in the European business statistics compilers’ manual for foreign affiliate statistics (Eurostat, 2024[6]).
← 13. Ideally the amount of round-tripping identified in the outward position statistics by UHE should equal the amount identified in the inward position statistics by UIE. In practice, this will depend on the methods chosen to reallocate the positions to the UHE and UIE and on data quality.
← 14. It was also considered whether head offices should be ‘looked through’ even though they are operating units. The decision of whether to look through these entities under the ‘first operating unit’ method is left to the Research Agenda when more countries can share their experiences with developing experimental statistics by UHE.
← 15. Information about the main economic activity of the enterprise can be used as a proxy (e.g., the statistical classification of economic activities in the International Standard Industrial Classification).
← 16. Near-SPEs are “hybrid companies displaying both SPE-like (financial intermediation) and non-SPE-like activities”. For more details, see (IMF BOPCOM, 2018[13]).
← 17. MNEs can structure their worldwide operations with DIEs that consistently report losses or are net lenders to their parent companies and have negative positions. Additionally, a DIE in a lower tier might be larger than the DIE in a higher tier due to investment from unaffiliated parties (“external financing”). This is assuming such cases are accurately reported to, or otherwise known, by the compiler. Additional difficulties can arise from these methods requiring statistical compilers, and reporting companies, to correctly capture detailed information at every tier.
← 18. See Annex 3 on Regional Arrangements of BPM7 (IMF, Forthcoming[1]).
← 19. According to BPM7 Annex 3 on Regional Arrangements (IMF, Forthcoming[1]), subscriptions to currency union central banks should be recorded under other investment.
← 20. It is worth noting that cross-border refers to the CU/EcUn as a whole. Intra-CU/EcUn transactions and positions are not relevant for the FDI statistics of the CU/EcUn, although they still are for the FDI statistics of CU/EcUn member economies.