## Reconciling BPM-based Direct Investment (DI) and Activities of Multinational Enterprise (AMNE) Statistics

## Source details

**Canonical URL:** [Reconciling BPM-based Direct Investment (DI) and Activities of Multinational Enterprise (AMNE) Statistics](https://www.imf.org/-/media/files/data/statistics/bpm6/approved-guidance-notes/d9-reconciling-bpmbased-direct-investment-and-amne-statistics.pdf)

## Other formats

- [Markdown version](/-/media/files/data/statistics/bpm6/approved-guidance-notes/d9-reconciling-bpmbased-direct-investment-and-amne-statistics.pdf.md)
- [Structured JSON version](/-/media/files/data/statistics/bpm6/approved-guidance-notes/d9-reconciling-bpmbased-direct-investment-and-amne-statistics.pdf.json)

---

### Background and conceptual differences
- Two complementary statistical sets for MNEs:
  - AMNE statistics: economic measures (number of affiliates, turnover, employment) and nationality-based classification by ultimate partner country using the Ultimate Controlling Investor (UCI) concept and majority-control population (typically > 50 percent equity ownership).
  - DI statistics: financial measures of transactions between resident companies and nonresident affiliates or parent companies, produced under the residency approach consistent with the 2008 SNA and BPM6, using influence relationships of 10 percent or more and typically reported by Immediate Controlling Parent (ICP).
- Key conceptual distinctions:
  - Residency approach (DI) versus nationality approach (AMNE).
  - AMNE assigns full economic values to the UCI for any control relationship (example: an 80 percent control relationship results in 100 percent assignment to the UCI); DI assigns values proportionally (example: 80 percent assignment).
  - AMNE populations limited to majority-owned affiliates; DI includes influence relationships at the 10 percent threshold.
  - AMNE data may be fiscal or accounting year–based; DI statistics are typically calendar-year based.
- Special measurement issues:
  - Investment passing through special purpose entities (SPEs) can mask ultimate source/destination in DI statistics; GN D.6 addresses Ultimate Investing Economy (UIE)/Ultimate Host Economy (UHE) and pass-through funds.
  - Reconciling requires detailed intra-group transactions and local enterprise information for variables like employment, while consolidating sales/turnover and assets/liabilities to avoid double-counting.

### Issues for discussion and possible reconciliation stages
- No agreed framework currently exists for reconciling AMNE with DI statistics; complexity arises from different populations, residency vs nationality approaches, and differing concepts of control and consolidation.
- Possible stages and actions for a reconciliation framework:
  - Explore existing sources on economic activities of MNEs that could be linked to DI.
  - Use ownership information from business registers to link DI and AMNE datasets.
  - Collect consolidated assets and liabilities of MNEs as part of data collection.
  - Collect AMNE-like activity variables: sales/turnover/output; employment; value added; imports and exports of goods and services; number of enterprises.
  - Expand data collection to include more financial and economic variables, particularly on intra-company activities.
  - Obtain detailed local enterprise–level information to facilitate full reconciliation with DI statistics.
- Analytical and policy uses of reconciled datasets:
  - Alternative national accounts presentations by foreign ownership and company size to identify primary income related to foreign affiliates, including re-invested earnings and dividends by type and size of ownership.
  - Complement Country by Country (CbC) reporting within BEPS: a linked DI–AMNE dataset could provide context to CbC returns and complement Mode 3 Trade in Services and extended Supply & Use tables.

### Outcomes from consultations (DIIT, BPTT, GZTT) and global survey findings
- DITT and BPTT member views:
  - General agreement that linked DI and AMNE statistics would be well received; emphasis on ensuring consistency in both statistics.
  - Suggested developing statistical definitions and primary breakdowns closer to accounting standards; candidate variables include income statement, employment data, and balance sheet data.
  - Institutional barriers: AMNE statistics often produced by agencies outside the BPM framework; limited data agreements impede sharing. Use of a common identifier (e.g., Legal Entity Identifier, LEI) could help but is not globally available or mandatory.
  - A reconciliation framework focused on control relationships was considered more feasible for compilers.
  - One BPTT member disagreed with including a reconciliation framework in BPM; most supported surveying compilers.
- GZTT agreement:
  - MNE definition in GN G.2: a legal entity that (1) has at least one non-resident affiliate or branch, and (2) exercises control over its affiliate(s) or branch(es) either directly—by owning over 50 percent of the voting power—or by indirect transmission of control. The MNE is the ultimate controlling parent (UCP). This definition applies consistently in national accounts and external sector statistics and aligns with identifying DI by the UIE using the “winner takes all” approach.
- OECD WGIIS proposed MNE framework highlights:
  - Measure total assets/liabilities and selected details as well as total economic activities of MNEs according to a common data reporting framework.
  - Cover both cross-border and domestic (local) financing.
  - Include direct and indirect control relationships (more than 50 percent ownership of voting equity).
  - Use stock data from balance sheets of direct investment enterprises as a matter of principle; consider flow data and acquisition type as supplements.
  - Allocate to ultimate controlling investor (UCP) and to ultimate host.
  - Consolidate statistically for the group to eliminate double counting due to funds in transit or round-tripping.
  - Provide a set of meaningful indicators to analyze MNEs.

### DITT–OECD WGIIS survey outcomes (December 2, 2021 to January 17, 2022)
- Respondent capabilities and coverage:
  - Respondents able to differentiate between influence and control relationships in DI transactions and positions: 74 percent.
  - Specific differentiation counts:
    - 36 respondents can differentiate in DI asset positions.
    - 41 respondents can differentiate in DI liability positions.
    - 26 respondents can differentiate in net acquisitions of DI assets.
    - 29 respondents can differentiate in net incurrences of DI liabilities.
    - 24 respondents can differentiate in DI income receipts.
    - 29 respondents can differentiate in DI income payments.
  - 27 respondents cannot differentiate these relationships; nine respondents did not provide an answer.
- Classification by ultimate partner economy:
  - More than half of respondents who answered indicated ability to classify DI liabilities and/or income payments according to the UIE using the “winner takes all” approach.
  - Only 20 percent of respondents who answered indicated ability to classify DI assets and/or income receipts according to the UHE.
- Ability to collect economic-activity variables from DI sources:
  - Less than half of respondents indicated ability to collect AMNE-like economic activity information from DI data sources.
  - Economic activity variables with higher possibilities of being produced included:
    - “sales, turnover, or gross output”
    - “imports and exports of goods and services”
    - “employment”
    - “value added”
    - “number of enterprises”
  - 38 respondents answered positively to collecting economic-activity variables; 12 respondents did not answer.
- Linkages between DI surveys and FATS/AMNE at micro-level:
  - 80 percent or 62 respondents do not link DI surveys with FATS or AMNE surveys at the micro-level.
  - Five respondents did not answer this question.

### Conclusions from surveys and consultations
- The survey results suggest progress can be made in bringing DI and AMNE/FATS statistics closer together in:
  - classification by partner country, and
  - population coverage.
- Practical opportunities:
  - Many respondents can differentiate between influence and control relationships in DI statistics.
  - A number of respondents can classify inward DI positions by the ultimate as well as the immediate direct investor, which could help reconcile classification differences.
- Practical feasibility:
  - A reconciliation framework focusing on control relationships is considered more feasible for statistical compilers and aligns with GZTT and other ongoing international statistical work.
- Rejected alternative:
  - Requiring full reconciliation of DI and AMNE statistics using detailed affiliate-level results was rejected due to feasibility concerns (extensive data collection, responder burden, and cost), though countries are not discouraged from pursuing full reconciliation where feasible.

### Recommendations and practical options for compilers
- Specific recommendations:
  - Update Appendix 4 in BPM6 to reflect the agreed MNE definition in GZTT GN G.2.
  - Develop a framework for reconciling DI and AMNE statistics including:
    - Identifying additional breakdowns (e.g., differentiating influence from control relationships in DI statistics).
    - New supplemental presentations (e.g., classifying inward DI by ultimate investing economy).
    - Considering additional variables (e.g., adding consolidated assets and liabilities to AMNE statistics).
  - Encourage DI compilers to explore producing reconciled statistics either by directly collecting AMNE information or linking to existing information; where not collected, compilers are invited to explore options to begin collecting these data and learn from countries that record them.
- Options for linkage and data development:
  - Explore linking DI microdata with pre-existing datasets to investigate MNE economic activities; linkage would be aided by a common identifier such as the LEI (if available) and expanded public and private MNE information sources.
  - At minimum, consider the extent to which AMNE-type information exists and identify gaps; include as many aspects as feasible in ongoing statistical development toward producing reconciled DI–AMNE statistics.

*Source: IMF Guidance Note "Reconciling BPM-based Direct Investment and AMNE Statistics" — SECTION I: THE ISSUE and related DITT survey outcomes and IMF Committee on Balance of Payments positions.*

### SECTION I: THE ISSUE

### d9-reconciling-bpmbased-direct-investment-and-amne-statistics - SECTION I: THE ISSUE

### Background
- Internationalization and globalization have produced increasingly complex corporate structures crossing multiple economic borders, driven by geographic diversification, trade-barrier responses, proximity to markets, and reducing the cost of labor, transportation, and taxes.
- Two complementary statistical sets provide insights on MNEs:
  - AMNE statistics: economic measures of activities (number of affiliates, turnover, employment) and nationality-based classification by ultimate partner country using the Ultimate Controlling Investor (UCI) concept and majority-control population (typically > 50 percent equity ownership).
  - DI statistics: financial measures of transactions between resident companies and their nonresident affiliates or parent company, produced under the residency approach consistent with the 2008 SNA and BPM6, using influence relationships of 10 percent or more and typically reported by Immediate Controlling Parent (ICP).
- Key conceptual differences:
  - Residency approach (DI) versus nationality approach (AMNE); nationality approach focuses on the location of the entity that ultimately controls the units.
  - AMNE assigns full economic values to the UCI for any control relationship (e.g., under AMNE, an 80 percent control relationship results in 100 percent assignment to the UCI), whereas DI assigns values proportionally (e.g., 80 percent).
  - AMNE populations are limited to majority-owned affiliates; DI includes influence relationships at 10 percent threshold.
  - AMNE statistics can be collected on a fiscal or accounting year basis; DI statistics are typically calendar-year based.
- Special issues:
  - Investment passing through special purpose entities (SPEs) can mask ultimate source/destination in DI statistics; GN D.6 addresses Ultimate Investing Economy (UIE)/Ultimate Host Economy (UHE) and pass-through funds.
  - Reconciling requires detailed intra-group transactions and local enterprise information for variables like employment, while consolidating variables such as sales/turnover and assets/liabilities to avoid double-counting.
- Existing guidance and related work cited:
  - Appendix 4 on AMNE in BPM6; Chapter 8 on foreign DI and Globalisation in OECD BD4; Eurostat FATS Recommendation Manual (2012); UN MSITS 2010.
  - OECD WGIIS Final Report (2013) recommended an analytical framework (not BD4 modifications) for harmonising financial and economic AMNE measures, including cross-border and domestic financing, and total assets and liabilities.
  - Globalization Task Team (GZTT) work on breaking down institutional sector accounts and GVA by foreign-controlled corporations, domestic MNEs, and other corporations.
- Survey and data-collection realities:
  - A survey by IMF’s DITT and OECD’s WGIIS found: more than half of respondents could identify cases of ultimate control for inward DI statistics; only 20 percent could identify the UHE; less than half could collect AMNE information from DI sources; most would need to link DI with FATS or AMNE microdata. Some countries have partial information; many outside the OECD may collect no AMNE information.

### Issues for Discussion
- No agreed framework currently exists for reconciling AMNE with DI statistics; reconciliation complexity arises from different populations, residency vs nationality approaches, and differing concepts of control and consolidation.
- Possible stages and actions for a reconciliation framework include:
  - Exploring existing sources on economic activities of MNEs that could be linked to DI.
  - Using ownership information from business registers to link DI and AMNE datasets.
  - Collecting consolidated assets and liabilities of MNEs as part of data collection.
  - Collecting AMNE-like activity variables: sales/turnover/output; employment; value added; imports and exports of goods and services; number of enterprises.
  - Expanding data collection to include more financial and economic variables, particularly on intra-company activities.
  - Having detailed local enterprise level information to facilitate full reconciliation with DI statistics.
- Potential analytical and policy uses:
  - Alternative presentations of national accounts by foreign ownership and company size to identify primary income related to foreign affiliates, including re-invested earnings and dividends by type and size of ownership.
  - Complementing Country by Country (CbC) reporting within BEPS: a linked DI-AMNE dataset could provide context to CbC returns and complement Mode 3 Trade in Services and extended Supply & Use tables.

### Outcomes from Consultations and Surveys
- DITT members’ views:
  - General agreement that linked DI and AMNE statistics would be well received, with emphasis on ensuring consistency in both statistics.
  - Suggested developing statistical definitions and primary breakdowns closer to accounting standards; include income statement, employment data, and balance sheet data among candidate economic and financial variables for AMNE statistics.
  - Institutional barriers noted: AMNE statistics often produced by agencies outside BPM framework; limited data agreements impede sharing. Use of a common identifier (e.g., Legal Entity Identifier, LEI) could help but is not globally available or mandatory.
  - A reconciliation framework focused on control relationships would be more feasible for compilers.
- BPTT members’ views:
  - Broad agreement with recommendations; support aligning DI and the definition of a MNE in national accounts (from GZTT).
  - Most agreed to a survey of compilers’ positions; one member disagreed with including a reconciliation framework in BPM.
  - General acknowledgement that partial DI-linked AMNE information would be useful; optional supplementary series separating majority control could be included.
- Global consultation and survey results:
  - Global consultation generally favored a framework to reconcile DI with AMNE statistics, focused on control relationships even though this extends beyond BPM scope.
  - The DITT–OECD WGIIS survey (conducted December 2, 2021 to January 17, 2022) showed respondents at different starting points: some with established AMNE/FATS data collections and time series, others facing major data-access challenges.
  - Only a minority reported linking DI microdata with AMNE/FATS collections or collecting economic activity on DI surveys; most indicated classifying DI by UIE and distinguishing influence vs control relationships were possible.
  - The GZTT agreed on a MNE definition in GN G.2: a legal entity that (1) has at least one non-resident affiliate or branch, and (2) exercises control over its affiliate(s) or branch(es) either directly—by owning over 50 percent of the voting power—or by indirect transmission of control. The MNE is the ultimate controlling parent (UCP). This definition applies consistently in national accounts and external sector statistics and aligns with identifying DI by the UIE using the “winner takes all” approach.

### Recommendations and Practical Options
- Specific recommendations:
  - Update Appendix 4 in BPM6 to reflect the agreed MNE definition in GZTT GN G.2.
  - Develop a framework for reconciling DI and AMNE statistics including:
    - Identifying additional breakdowns (e.g., differentiating influence from control relationships in DI statistics).
    - New supplemental presentations (e.g., classifying inward DI by ultimate investing economy).
    - Considering additional variables (e.g., adding consolidated assets and liabilities to AMNE statistics).
  - Encourage DI compilers to explore producing reconciled statistics either by directly collecting AMNE information or linking to existing information; where not collected, compilers are invited to explore options to begin collecting these data and learn from countries that record them.
- Options for compilers towards reconciliation:
  - Explore linking DI microdata with pre-existing datasets to investigate MNE economic activities; linkage would be aided by a common identifier such as the LEI (if available) and expanded public and private MNE information sources (e.g., OECD ADIMA, Euro pean Groups Register (EGR), Moody’s Orbis).
  - At minimum, consider the extent to which AMNE-type information exists and identify gaps; include as many aspects as feasible in ongoing statistical development toward producing reconciled DI-AMNE statistics.
- Practical note on feasibility:
  - A reconciliation framework focusing on control relationships is considered more feasible for statistical compilers and aligns with GZTT and other ongoing international statistical work.

*Source: IMF Guidance Note "Reconciling BPM-based Direct Investment and AMNE Statistics" — SECTION I: THE ISSUE.*

### 22.      The IMF Committee on Balance of Payments strongly supported developing a framework

### Reconciling BPM-based Direct Investment (DI) and Activities of Multinational Enterprise (AMNE) Statistics

### IMF Committee position and framework development
- The IMF Committee on Balance of Payments strongly supported developing a framework for reconciling DI and AMNE statistics for possible inclusion in the BPM7 and launching a survey to gather feedback from compilers on such framework.
- The Committee proposed including a definition of MNE.
- The Committee discussed that the framework should:
  - not only aim to reconcile DI with AMNE statistics but also be fully coordinated and ultimately aligned with the work on MNEs by the OECD WGIIS (referred to as GZTT in the text).
  - introduce additional breakdowns of DI statistics to better align with the concept of MNEs used in AMNE.
- This version of the Guidance Note (GN) incorporates the Committee’s discussions.

### Rejected alternatives and rationale
- Requiring full reconciliation of DI and AMNE statistics using detailed affiliate-level results was rejected.
  - Rationale: in many cases, a fully reconciled MNE dataset with affiliate-level transactions would not be feasible due to extensive data collection, responder burden and cost.
  - Compilers would need to be sure of user demand for these statistics and that companies can provide the required level of detail.
  - The Committee noted that this rejection should not discourage countries that wish to pursue a full reconciliation of AMNE and DI statistics.

### Supplementary information — OECD (2013) highlights of the proposed MNE framework
- The OECD WGIIS proposed that the new MNE Framework should:
  1. measure total assets/liabilities and selected details as well as total economic activities of MNEs according to a common data reporting framework (i.e., going beyond, but also including FDI);
  2. cover both cross-border and domestic (local) financing;
  3. include direct and indirect control relationships (more than 50 percent ownership of voting equity);
  4. use stock data from balance sheets of direct investment enterprises as a matter of principle and to consider use of flow data, and data on the acquisition of equity by type (M&A, greenfield, etc.) as a supplement;
  5. allocate to ultimate controlling? investor (UCP) and to ultimate host;
  6. consolidate statistically for the group to eliminate double counting due to funds in transit or round-tripping;
  7. provide a set of meaningful indicators to analyze MNEs.
- Diagrammatic population distinctions included: Majority-owned affiliates and UCIs (FATS); Affiliates with no cross-border links; Minority-owned foreign affiliates.

### DITT survey on harmonizing DI with AMNE/FATS — survey context
- GN D.9 explored differences in populations, classifications, and variables collected between DI and AMNE statistics that hinder joint use.
- Key differences highlighted:
  - AMNE statistics typically cover control relationships only; DI statistics cover both influence (ownership of 10 percent or more of voting power) and control relationships (over 50 percent).
  - AMNE statistics are classified by the ultimate partner economy; DI statistics are typically classified by the immediate partner economy.
  - AMNE statistics provide economic measures of activities (e.g., turnover, employment); DI statistics measure financing.

### Survey outcomes — current compilation practices and capabilities
- Respondent coverage:
  - Most respondents (74 percent) provided information about their ability to differentiate between influence and control relationships in their DI transactions and positions.
  - Specific differentiation capabilities reported:
    - 36 respondents can differentiate in DI asset positions.
    - 41 respondents can differentiate in DI liability positions.
    - 26 respondents can differentiate in net acquisitions of DI assets.
    - 29 respondents can differentiate in net incurrences of DI liabilities.
    - 24 respondents can differentiate in DI income receipts.
    - 29 respondents can differentiate in DI income payments.
  - 27 respondents cannot differentiate these relationships; nine respondents did not provide an answer.
- Classification by ultimate partner economy:
  - More than half of respondents who answered indicated they would be able to classify DI liabilities and/or income payments according to the UIE using the same concept of ultimate investor used in AMNE or FATS statistics (the “winner takes all” approach discussed in GN D.6).
  - Only 20 percent of respondents who answered indicated they would be able to classify DI assets and/or income receipts according to the UHE.
- Ability to collect economic-activity variables from DI data sources:
  - Less than half of respondents indicated they would be able to collect information on economic activities of MNEs similar to FATS from their DI statistics data sources.
  - Economic activity variables with higher possibilities of being produced included:
    - “sales, turnover, or gross output”
    - “imports and exports of goods and services”
    - “employment”
    - “value added”
    - “number of enterprises”
  - Thirty-eight respondents answered positively to the question on collecting economic-activity variables; 12 respondents did not answer this question.
- Linkages between DI surveys and FATS/AMNE at micro-level:
  - The extensive majority, 80 percent or 62 respondents, do not link their DI surveys with FATS or AMNE surveys at the micro-level.
  - Five respondents did not answer this question.

### Conclusions drawn from the survey
- The survey results suggest progress can be made in bringing DI and AMNE/FATS statistics closer together in:
  - classification by partner country, and
  - population coverage.
- Practical opportunities identified:
  - Many respondents can differentiate between influence and control relationships in their DI statistics.
  - A number of respondents can classify inward DI positions by the ultimate as well as the immediate direct investor, which could help reconcile classification differences.

---


_Source: https://www.imf.org/-/media/files/data/statistics/bpm6/approved-guidance-notes/d9-reconciling-bpmbased-direct-investment-and-amne-statistics.pdf_