## F2 Asymmetric treatment of retained earnings

## Source details

**Canonical URL:** [F2 Asymmetric treatment of retained earnings](https://www.imf.org/-/media/files/data/statistics/bpm6/approved-guidance-notes/f2-asymmetric-treatment-of-retained-earnings-final.pdf)

## Other formats

- [Markdown version](/-/media/files/data/statistics/bpm6/approved-guidance-notes/f2-asymmetric-treatment-of-retained-earnings-final.pdf.md)
- [Structured JSON version](/-/media/files/data/statistics/bpm6/approved-guidance-notes/f2-asymmetric-treatment-of-retained-earnings-final.pdf.json)

---

### SECTION I: THE ISSUE — Background and conceptual framing
- Retained earnings correspond to the net distributable income that has not been distributed to shareholders in the form of dividends (2008 SNA, paragraph 7.139; BPM6, paragraph 11.34).
- Retained earnings of foreign direct investment (FDI) enterprises are treated as remitted to direct investors and reinvested by them and are called reinvested earnings (RIE).
- BPM6 records RIE as distributed to direct investors in proportion to equity ownership; a direct investor is entitled to income irrespective of actual distribution (BPM6, paragraph 3.74).
- RIE is recorded as direct investment income in the current account and as a transaction in equity in the financial account (BPM6, paragraphs 8.15–8.16 and 11.33–11.47).
- 2008 SNA treats retained earnings of foreign corporations included in FDI as distributed and then reinvested by the shareholder; RIE is recorded only for equity in FDI and investment funds, not for other types of equity.
- Similar RIE-like treatment applies to collective investment funds where income earned by funds is deemed all distributed (D.443) and thus received by equity instrument holders (F.52) (ESA 2010 sub-codification; 2008 SNA, paragraph 7.152).
- Share buybacks have grown substantially and in some leading markets exceed dividends, acting as substitutes for dividends and potentially underestimating shareholder income (overestimating holding gains) and distorting savings distribution across sectors.

### Inconsistencies arising from current treatment
- Two main asymmetries:
  - Inconsistency 1: FDI relationships are treated differently from foreign portfolio investment relationships (nonresident investor owns less than 10 percent of equity).
  - Inconsistency 2: Domestic direct investment relationships are treated differently from investment relationships between residents and nonresidents.
- Allocation summary for retained earnings on equity other than Investment Fund shares:
  - Domestic Equity Relationship:
    - Direct Investment: Retained earnings allocated to saving of the corporation
    - Portfolio Investment: Retained earnings allocated to saving of the corporation
  - International Equity Relationship:
    - Direct Investment (FDI, RIE): Retained earnings allocated to saving of the shareholder
    - Portfolio Investment: Retained earnings allocated to saving of the corporation
- Consequences of the asymmetry:
  - Conditioning recording on investors’ degree of control shifts returns between income and revaluation depending on distribution decisions.
  - Net saving of an enterprise 100 percent owned by foreign direct investors equals zero (2008 SNA, paragraph 26.65); by contrast, an enterprise 100 percent owned by portfolio or resident investors records retained earnings as enterprise saving and investor wealth growth as revaluation.
  - Corporate inversions can shift net primary income receipts and related national accounts indicators across economies.
  - Asymmetry particularly relevant for public corporations and government finance, with implications for deficit measurement and incentives for dividend manipulation.

### Rationale and core conceptual debate
- Core issue: whether enterprises should have saving.
  - Argument for enterprise saving: aligns with corporations as distinct institutional units with autonomy over retention versus distribution decisions (2008 SNA, paragraphs 4.1–4.6; paragraph 4.38b).
  - Argument for imputing RIE to investors: direct investors may have control or effective access; retained earnings contribute to investors’ income and market value, suggesting recording as transaction vs revaluation.
- Practical disadvantages of wider imputation include uncertain assumptions required (service lives for consumption of fixed capital, definition/location of software and intellectual property, inventory pricing) — these assumptions are, however, already routinely used for FDI RIE and collective investment schemes.

### Interactions and potential asymmetry effects
- Resolving one asymmetry in isolation can aggravate others:
  - Applying RIE to foreign portfolio investment only would magnify cross-border vs domestic treatment differences.
  - Extending FDI treatment to domestic investment by controlling shareholders would expand direct vs portfolio differences.
- RIE thresholds:
  - FDI RIE based on "10 percent" threshold.
  - Public corporation control would be set at "50 percent" threshold for control-based considerations.
- Two rationales for extension:
  - To better measure national income (suggesting extension to cross-border portfolio investment).
  - By shareholder control/influence criterion (suggesting extension to domestic direct investment relationships).

### Options considered for addressing asymmetry
- Option 1: Keep the status quo (current BPM6 and 2008 SNA treatment).
- Option 2: Leave core accounts unchanged but add supplementary information on portfolio investment RIE and on public corporations/domestic enterprise RIE to national accounts and balance of payments.
- Option 3: Extend RIE applied to foreign direct investors to public controlled corporations and/or to cross-border portfolio investors.
- Option 4: Extend RIE applied to foreign direct investors to all equity holdings in national accounts and balance of payments.
- Option 5: Eliminate RIE for FDI in core accounts (with possibility to keep in supplemental tables or memorandum items).

### Committee findings and preferences
- Majority of FITT members rejected Option 1 and favored applying RIE uniformly to controlling/influencing shareholders (foreign and domestic) conceptually.
- Practical challenges to uniform application include:
  - Need for databases to identify enterprises with a local controlling shareholder.
  - Attribution of income to ultimate shareholders requires ownership-chain knowledge.
  - Many statistical institutes lack complete resident ownership breakdowns except where security-by-security databases exist.
  - Extension to domestic links could use macro-adjustments via equity cross-sector whom-to-whom positions.
- Feasibility view:
  - Extending RIE only to public corporations (Option 3) is more practical and could be significant for economies where general government deficit is prominent.
  - Extending RIE to all cross-border equity links (part of Option 3/4) would improve national income measurement with compilation challenges that "may be less severe than for FDI or at least not more severe."
  - Removing RIE in FDI (Option 5) would enhance methodological consistency but hamper analytical usefulness of GNI under globalization.
  - Option 4 (full extension) offers coherence but poses significant practical implementation challenges and risk of delays, reduced reliability, and breaks in time series.

### Preferred pragmatic approach (Option 2)
- Option 2 recommended as a pragmatic compromise:
  - Permit publication of supplementary information based on alternative definitions without affecting core accounts’ internal consistency, international comparability, and time series continuity.
  - Conceptual acceptance that RIE should apply to all equity investments, while universal application is restricted to supplementary information.
  - Under supplementary extended RIE:
    - Core accounts: RIE continues to apply to FDI.
    - Supplementary tables: alternative balance of payments accounts where RIE is applied to portfolio investment; alternative measures of income, saving, and investment at national and sector levels where RIE is applied to all equity investment.
    - An alternative measure of government deficit could be compiled and labeled as such.
- FITT consultation numeric outcome:
  - Of 13 FITT members expressing views, "eight" supported Option 2 either as first choice or as preferred for feasibility; three supported Option 4 (one as second choice), one supported Option 3, and one supported Option 5.
- DITT members later split between Option 1 and Option 5; Option 2 received substantial support given feasibility difficulties with Options 3 and 4.
- Committee and AEG endorsed Option 2 to record allocation of retained earnings in a supplementary table and requested additional guidance for ESS and national account compilers.

### Testing, endorsement, and implementation guidance requests
- Feasibility testing approach:
  - Stage 1: survey about availability of source data for compiling RIE for portfolio investment equity (ESS) and domestic transactions (national accounts).
  - Stage 2: selected compilers provided experimental calculations.
- Findings presented at March 2022 Joint Committee and AEG Meeting and March 2023 AEG Meeting showed:
  - Data availability remains challenging but existing methods exist for calculating RIE for PI equity and domestic RIE transactions.
  - Some members requested further compilation guidance (to be provided in the new BPM Compilation Guide).
- Recommendation: feasibility of Option 2 be tested, especially if Options 3 or 4 are considered for core accounts.

### Share buybacks — policy and measurement considerations and proposed intermediate approach
- Observed facts and motivations:
  - Share buybacks have become a major distribution method due to tax efficiency and corporate flexibility; some companies never pay regular dividends and rely on buybacks.
  - In some leading markets buybacks have exceeded dividends; cited magnitudes include exceeding "one trillion dollars a year" and close to "five percent of GDP" in the US example.
- Current statistical treatment: buybacks treated as financial transactions (F.5) because sellers dispose of shares and receive cash.
- Guidance Note intermediate approach (expensing buybacks while keeping current core framework):
  - Rationale: increases distribution of income across the economy similarly to Option 4 while retaining the core concept that corporations retain saving (Options 2/1 boundary).
  - Conceptual question: whether handing out cash to shareholders taps corporations' savings; many believe they do and companies often borrow to finance buybacks.
- Proposed imputation method to expense buybacks:
  - Impute a dividend (D.42) for the amount and at the time of a share buyback:
    - Record imputed D.42 as a use/expenditure/debit in company accounts.
    - Record imputed D.42 as resource/revenue/credit across all shareholders, allocated using the REI allocation method.
    - Counterpart to the imputed D.42 is F.5 (same as REI D.43).
    - Company outflow currently coded (-)F.5L becomes D.42.
    - Shareholders' aggregate account records imputed D.42 against (+)F.5A.
    - Share seller records (-)F5.A against cash received (+)F2A, plus apportioned D.42/(+)F.5A.
  - Safeguard: new total D.42 should be superdividend tested to distinguish distributions of period earnings from liquidating transactions.
  - Treatment independence: recording should not vary by subsequent use of repurchased shares; subsequent uses treated separately.
- FITT and AEG support:
  - Broad support for treating buybacks as income distribution in core accounts conceptually, but preference to discuss details in a separate/subsequent guidance note; issue placed on post-review research agenda.

### Empirical testing and data-availability evidence — survey and experimental results (key statistics)
- ESS survey participation:
  - "Eighty-one respondents from 75 countries completed the survey (Table 3)."
  - "Twenty-nine countries (39 percent) were interested in participating in the testing exercise for the reinvested earnings (REI) of portfolio investment (PI) assets and 31 countries (41 percent) for the REI of PI liabilities."
- Source data availability for RIE of portfolio investment:
  - For portfolio investment assets: "Seventy-seven percent of the respondent countries do not have the source data to compile the RIE of portfolio investment assets."
  - For portfolio investment liabilities: "63 percent of countries do not have it for RIE of portfolio investment liabilities."
- Interest in experimental exercise:
  - "Thirty-nine percent of countries expressed interest to participate in the testing exercise for the REI of PI assets and 41 percent for the REI of PI liabilities."
  - Non-participation reasons: lack of resources and absence of source data.
- Plans to develop source data:
  - "Around 20 percent of the respondent countries would develop the source data to compile the supplementary tables."
- Table 4 aggregate survey results (PI and data development):
  - Portfolio investment assets: Yes 16 (21%), No 59 (79%), Total 75 (100%).
  - Portfolio investment liabilities: Yes 26 (35%), No 47 (64%), No response 1 (1%), Total 75 (100%).
  - Interest in participation (PI assets): Yes 28 (37%), No 46 (61%), No response 1 (1%), Total 75 (100%).
  - Interest in participation (PI liabilities): Yes 30 (40%), No 43 (57%), No response 2 (3%), Total 75 (100%).
  - Plans to develop source data for PI assets: Yes 16 (21%), No 40 (53%), No response 19 (25%), Total 75 (100%).
  - Plans to develop source data for PI liabilities: Yes 13 (17%), No 32 (43%), No response 30 (40%), Total 75 (100%).
- Experimental calculation exercise for REI (portfolio investment):
  - Six respondents out of 31 volunteer countries responded to the second-phase questionnaire; only four provided data; one compiler could not undertake compilation due to other commitments.
  - Methods observed:
    - United States and Switzerland used aggregated stocks and aggregated performance indicators (indirect indicators, PER and payout ratios).
    - US formula example: REI percentage = (1 ÷ PER) – DY; RE in USD = (RIE% ÷ 4) x ((S_t0 + S_t1) ÷ 2).
    - Armenia and Portugal used more granular enterprise- or security-level data and financial statements.
  - Conclusion: aggregated performance indices introduce noise and composition differences; security-by-security data are preferable but often unavailable.
  - Overall conclusion: current data systems are not designed to collect RIE data for portfolio investments; economies need additional databases on (i) foreign companies, (ii) domestic companies with PI, and (iii) security-by-security holdings.

### National accounts consultation and experimental calculations for domestic RIE
- Participation and relevance:
  - "A total of 44 respondents contributed to this consultation."
  - Topic relevance rated high or medium for most respondents.
- RIE for public corporations (domestically controlled):
  - "For 56 percent of respondents, institutions have source data available to compile RIE for public corporations as supplementary items."
  - Among those with source data: 80 percent have GFS assets and liabilities data, 72 percent have GFS financial transaction data, 24 percent have other source data (enterprise databases, financial statements, BOP data).
  - Of institutions with source data, most (83 percent) have access to detailed audited financial statements including notes for public corporations.
  - Sixteen institutions expressed interest in participating in an experimental estimate exercise for public corporations.
- RIE for private domestic equity stakes:
  - "Approximately 50 percent of all respondents’ institutions have relevant source data available to compile RIE for private domestic equity stakes."
  - Dominant sources: financial statements, administrative tax data, enterprise income surveys.
  - Fourteen institutions are interested in participating in experimental estimates for private domestic equity stakes.
- Experimental calculations among national accountants (13 economies responded) used approaches including:
  - RE1 = net operating surplus plus net property income plus net current transfers.
  - RE2 = net operating surplus plus net property income.
  - RE3 (Residual) = Profits after taxes less dividends or Profits before taxes less (taxes and dividends).
  - For financial corporations, RE4 = interest revenue minus interest payable, plus fees/commissions, plus net dividend income, minus costs; central bank similar with no fees/commissions typically.
- Feasibility concerns:
  - Allocation of RIE to equity holders often hampered by lack of detailed shareholding information in administrative sources; SMEs and unlisted companies often excluded.
- Recommendation: economies should provide sectored RIE supplementary items, not change core accounts; BPM7 agreed to a supplementary table. AEG to consider guidance drawing on country experience (Portugal, Ireland).

### Policy and compilation recommendations (summary)
- Adopt Option 2 as pragmatic path forward: retain current core treatment (RIE for FDI) and publish supplementary tables applying RIE to portfolio investment and to domestic equity where feasible.
- Provide additional compilation guidance in updated statistical manuals and BPM Compilation Guide to support experiments and country testing.
- Encourage feasibility testing and pilot compilation exercises, prioritizing:
  - Supplementary RIE for portfolio investment (PI) assets and liabilities.
  - Supplementary RIE for public corporations and private domestic equity stakes.
- Develop and promote data infrastructure improvements where feasible:
  - Security-by-security holdings databases.
  - Databases enabling allocation to ultimate owners via ownership chains.
  - Enhanced enterprise surveys and incorporation of financial statement data.
- Consider expensing share buybacks via an imputed dividend (D.42) in conjunction with superdividend testing as an intermediate measure to better reflect distributions without having to change the core RIE paradigm immediately.
- Preserve safeguards to avoid perverse incentives (e.g., governments reselling shares to generate revenue) and to ensure superdividend tests identify liquidation-like transactions.

*Guidance Note — SECTION I: THE ISSUE (IMF).*

### SECTION I: THE ISSUE

### SECTION I: THE ISSUE

### Background
- Retained earnings correspond to the net distributable income that has not been distributed to shareholders in the form of dividends (System of National Accounts 2008 (2008 SNA), paragraph 7.139 and others; sixth edition of the Balance of Payments and International Investment Position Manual (BPM6), paragraph 11.34 and others).
- Retained earnings of foreign direct investment (FDI) enterprises are considered in the international statistical standards as being remitted to the direct investors and reinvested by them and are called reinvested earnings (RIE).
- The BPM6 records RIE as being distributed to direct investors in proportion to their equity ownership in the enterprise. A direct investor is seen as entitled to all the income generated by its subsidiaries, associates, and branches, irrespective of whether the income is distributed in the form of dividends (or branch profits) or retained as RIE (BPM6, paragraph 3.74).
- RIE are recorded as direct investment income in the current account and as a transaction in equity in the financial account (BPM6, paragraphs 8.15–8.16 and 11.33–11.47).
- The 2008 SNA treats retained earnings of foreign corporations included in FDI as though they had been distributed and then reinvested by the shareholder; the item is called RIE and the equal entry in the financial account is called reinvestment of earnings (2008 SNA, paragraph 26.88).
- RIE is recorded only for equity in FDI and investment funds, but not for other types of equity.
- The 2008 SNA does not recommend classification of resident-to-resident investment relationships as domestic direct investment; therefore, RIE treatment does not arise in domestic equity links.
- BPM6 and 2008 SNA rationale: a direct investor has significant influence on management and effective access to earnings; decision to retain earnings represents a conscious investment decision by direct investors; imputed distribution of retained earnings is included in direct investors’ income and then reinvested (BPM6, paragraph 11.41; 2008 SNA, paragraph 7.138; BPM6, paragraph 3.17).
- Dividends reduce RIE (which can be negative—see 2008 SNA, paragraph 26.64 and BPM6, paragraph 11.46); dividends thus affect the split between dividends and RIE and are financial account cash payments counterbalanced by financial transactions in equity.
- Treatment contrast: dividends from non-FDI corporations (portfolio or domestic direct investment) affect overall income paid by the corporation and corresponding decreases in corporate value are reflected as holding losses, not as transactions.
- Similar RIE-like treatment applies to collective investment funds (and some pension funds) where income earned by funds is deemed all distributed (D.443) and thus received by equity instrument (F.52) holders, irrespective of actual distribution (ESA 2010 sub-codification; 2008 SNA, paragraph 7.152). As a result, the savings of these investment funds are emptied.
- Recording dividends from non-FDI corporations is sometimes described as a deviation from strict accrual: investor income recorded at ex-dividend date rather than when investee income is earned (2008 SNA, paragraph 7.130). This creates an income/revaluation boundary issue and becomes more visible with quarterly accounts and growing share buybacks.
- Share buybacks have developed considerably and in some leading markets exceed dividends; buybacks act as substitutes for dividends, enabling tax optimization and flexibility, leading to underestimation of shareholder income (overestimating holding gains) and distorting savings distribution across sectors.

### Inconsistencies highlighted
- Two main inconsistencies caused by use of RIE only for certain transactions:
  - Inconsistency 1: FDI relationships are treated differently from foreign portfolio investment relationships (other than investment fund shares).
  - Inconsistency 2: Domestic direct investment relationships are treated differently from investment relationships between residents and nonresidents.
- The term direct investment and portfolio investment will be used both for cross-border and domestic equity relationships in the remainder of the Guidance Note.

### Inconsistency 1 — Direct Investment vs Portfolio Investment
- RIE transactions are not recorded for foreign portfolio investment (nonresident investor owns less than 10 percent of equity).
- For FDI, increase in value due to retained earnings is regarded as a transaction (RIE) rather than a revaluation; for portfolio investment, retained earnings are recorded as the saving of the enterprise and the associated increase in investor equity value is recorded as a revaluation.
- Justification for differentiation: portfolio investors are considered to have insignificant influence on management and limited input into enterprise saving decisions.
- Consequence: conditioning recording on investors’ degree of control shifts statistical recording of financial returns arbitrarily between the income balance and revaluation, depending on distribution decisions.
- The treatment does not affect net foreign assets’ value but affects the share of growth of net foreign assets attributed to transactions versus revaluations.
- The RIE rule was mainly designed to measure national income; asymmetric treatment may hinder this objective where net portfolio cross-border positions are large and where share buybacks are prevalent.
- Counterargument: minority shareholders may indeed lack influence over distribution vs reinvestment decisions.

### Inconsistency 2 — Foreign vs Domestic Investment Relationship
- Except for FDI, 2008 SNA treats retained corporate earnings as saving of the corporation; domestic direct investment is not an SNA concept, so RIE transactions are never imputed for resident-to-resident relationships.
- The income of resident shareholders always depends on how much the corporation chooses to distribute as dividends.
- The 2008 SNA research agenda considers extending RIE to domestic equity relationships, particularly for public corporations (2008 SNA, paragraph 7.140), but prior review rejected broad extension in favor of rules on superdividends and capital injections.
- The 2008 SNA warns that extending RIE to domestic links “would have serious implications for interpretation of the accounts since it would be built on a different paradigm from the current treatment of dividends and corporate saving” (2008 SNA, paragraph A4.29).
- SNA paradigm: corporations as independent institutional units that own assets, are responsible for liabilities, and engage in transactions (2008 SNA, paragraphs 4.1–4.6), with limited liability for owners (2008 SNA, paragraph 4.38b).
- The SNA paradigm is not applied for FDI and investment funds where RIE-like treatment exists.

### Allocation of retained earnings under current standards (as summarized)
- Table 1 representation (allocation of retained earnings on equity other than Investment Fund shares):
  - Domestic Equity Relationship:
    - Direct Investment: Retained earnings allocated to saving of the corporation
    - Portfolio Investment: Retained earnings allocated to saving of the corporation
  - International Equity Relationship:
    - Direct Investment (FDI, RIE): Retained earnings allocated to saving of the shareholder
    - Portfolio Investment: Retained earnings allocated to saving of the corporation

### Issues for discussion
- Core issue: whether enterprises should have saving.
  - Letting corporations have saving aligns with treating them as distinct institutional units with autonomy over decisions like retaining earnings versus paying dividends.
  - Assuming an imputed transaction in income and equity irrespective of the corporation’s distribution decision may contradict the standards’ definition of transactions that emphasize mutual agreement (e.g., BPM6, paragraph 3.4).
  - The argument against imputation is weaker for 100 percent owned subsidiaries (particularly special purpose entities (SPEs)) or controlled entities.
- Legal existence and informational value:
  - Allowing enterprises to have saving is consistent with limited liability legal status and avoids over-simplified allocation of retained earnings’ benefits/costs to common shareholders.
  - The level of enterprise saving is a useful indicator of intention to fund investment from internal resources; some analysts exclude RIE in FDI when measuring foreign capital inflows because RIE is generated in the host economy.
- Terminology and presentation considerations:
  - Users often better understand “retained earnings” or “undistributed incomes” than the term “saving” for corporations (see 2008 SNA, paragraph 9.11).
  - Extending RIE to all corporations’ equity links would set their net saving (B.8n) to zero, but retained earnings could still be observed via net RIE; recipients’ saving would also remain observable by disaggregating the RIE component.
  - For households, increases in share prices from retained earnings contribute to consumption incentives and current non-RIE treatment may understate the saving ratio; creation of an alternative balance could address communication issues.
- Arguments for earnings accruing to investors as earned:
  - Owners’ limited liability may have limited practical impact; direct investors may have obligations for debts (reputational) or in cases of FDI in a branch.
  - Retained earnings contribute to change in market value of the corporation; the key question is whether that contribution is better recorded as a transaction (RIE) or as a revaluation.
  - Any equity holder could, in principle, sell shares to realize a situation equivalent to distributed earnings; companies may distribute dividends in kind (extra shares), economically similar to retaining earnings and recorded as income (2008 SNA, paragraph 7.129).
- Asymmetry consequences:
  - Mixed treatment causes asymmetries: enterprise saving depends on the investor mix (direct investors, portfolio investors, resident investors).
  - Net saving of an enterprise 100 percent owned by foreign direct investors equals zero (2008 SNA, paragraph 26.65); by contrast, an enterprise 100 percent owned by portfolio or resident investors records all retained earnings as the enterprise’s saving, and investor wealth growth is recorded as a revaluation.
  - Corporate inversions can make the asymmetry’s impact evident by shifting net primary income receipts and related national accounts indicators (e.g., GNI) between economies.
- Public corporations and government finance:
  - 2008 SNA research agenda identifies asymmetry as particularly relevant for public corporations (2008 SNA, paragraphs A4.29 and 7.140).
  - Extending RIE to public corporations would reduce incentives for governments to use ownership of controlled corporations to artificially optimize deficit figures (e.g., by distributing large dividends to meet deficit targets and later recapitalizing via capital injections).
  - New superdividend and capital injection rules (2008 SNA, paragraphs 7.131 and 22.138) aimed to contain such manipulation but are ad-hoc and subject to contestation.
  - Extending RIE to public corporations would shift government deficit recognition to when investees’ earnings are earned (rather than when distributed) and would increase deficit when investees’ losses occur (rather than when actually covered), improving measurement of government net lending/net borrowing at any point in time.

*Guidance Note — SECTION I: THE ISSUE (IMF).*

### 25.      The treatment of FDI is based on the presumption that the foreign investor has control of

### f2-asymmetric-treatment-of-retained-earnings-final - 25.      The treatment of FDI is based on the presumption that the foreign investor has control of

### Rationale and conceptual issues regarding retained earnings imputation
- The treatment of FDI presumes the foreign investor has control of or influence on the company and therefore direct access to its net income; routing retained earnings of public corporations to the government would improve logical consistency where there is a controlling shareholder.
- Routing retained earnings in cases of controlling shareholder would eliminate potential large swings in government saving that can occur when classification of a public corporation changes between market producer and non-market producer.
- Imputing distributions of retained earnings to all kinds of shareholders would:
  - Make the role of saving in the growth of shareholders’ assets more visible.
  - Make measures of institutional sector saving, national saving, and the current account balance more meaningful.
  - Reveal a more significant role of saving in building household wealth (complements holding gains).
  - Better reflect resources available to residents where the economy has a large net negative position in foreign portfolio investment by taking into account claims of foreign portfolio investors on retained earnings of resident corporations.
- Practical disadvantages of giving retained earnings a more prominent role include uncertain assumptions required for estimation, such as:
  - Service lives used to model consumption of fixed capital.
  - Definition and location of software and other intellectual property assets.
  - Assignment of prices to inventory additions and withdrawals.
- These assumptions are routinely made for compilation of RIE on FDI and for allocation of retained earnings on collective investment schemes.

### Interactions and potential asymmetry effects
- Resolving a single asymmetric treatment in isolation may make other asymmetries more problematic:
  - Applying the RIE treatment to foreign portfolio investment only would increase the difference between cross-border and domestic investment treatment.
  - Extending current FDI treatment to domestic investment by controlling shareholders would expand differences between direct and portfolio relationships.
- RIE on FDI is based on a "10 percent" threshold, while the threshold for public corporations would be set at "50 percent"; an associate's control over distribution is less solid than a controlling parent.
- Extending RIE is justified on two rationales:
  - To better measure national income (GNI, etc.), warranting extension to cross-border portfolio investment.
  - By shareholder control/influence criterion, justifying extension to domestic direct investment relationships.
- Extensions to domestic investment links can complicate interpretation of institutional sector saving due to changes over time in enterprise classification.

### Options considered (five options)
- Option 1: Keep the status quo (current treatment in both BPM6 and 2008 SNA).
- Option 2: Leave core balance of payments accounts and national accounts unchanged but add supplementary information on portfolio investment RIE to balance of payments (and possibly memorandum items) and national accounts; add supplementary information on public corporations RIE and overall investment in resident enterprises RIE to national accounts.
- Option 3: Extend concept of RIE applied to foreign direct investors to public controlled corporations and/or to cross-border portfolio investors.
- Option 4: Extend concept of RIE applied to foreign direct investors to all equity holdings in the national accounts and balance of payments accounts.
- Option 5: Eliminate asymmetries by discontinuing current treatment of RIE for FDI (with possibility to keep treatment in supplemental tables or memorandum items).

### Committee findings, feasibility, and preferred approach
- Majority view of FITT members rejected Option 1 and favored enhancing coherence by applying RIE uniformly to controlling/influencing shareholders, both foreign and domestic.
- Practical challenges to uniform application include:
  - Need for adapted databases to distinguish enterprises with a local controlling shareholder.
  - Attribution of income to ultimate shareholders requires knowledge of ownership chains.
  - Many statistical institutes/central banks distinguish resident vs nonresident ownership but lack complete resident ownership breakdowns except where security-by-security databases exist.
  - Extension to domestic links could be done via macro-adjustments using equity cross-sector whom-to-whom positions.
- Extending RIE only to public corporations (Option 3) is more practical than extending to all domestic controlling-shareholder links and would improve coherence slightly for some economies; for others, it could be a significant improvement given prominence of general government deficit and could allow dropping superdividend and capital injection rules.
- Extending RIE to all cross-border equity links (Option 3) would improve measurement of national income with compilation challenges that "may be less severe than for FDI or at least not more severe."
- Removing RIE in FDI (Option 5) would enhance methodological consistency but would hamper analytical and policy usefulness of GNI in the context of globalization because RIE partially corrects distortions caused by MNE operations.
- Extending RIE to all enterprises regardless of owners’ residency and control (Option 4) would:
  - Allow a coherent, unified approach attributing saving to ultimate beneficiaries.
  - Present significant practical implementation challenges across sectors, resident/nonresident relationships, and likely produce significant effects on current account balance, national saving, and household saving.
  - Risk delays, reduced reliability, and important breaks in time series without simplifying assumptions.

### Role and rationale for Option 2 (pragmatic compromise)
- Option 2 allows publication of supplemental information based on alternative definitions without affecting internal consistency, international comparability, and time series continuity of core accounts; development timetable can be flexible.
- Pragmatic outcome: conceptual acceptance that RIE should apply to all equity investments, while recommending universal application only in supplementary information.
- Under extended RIE in supplementary tables:
  - RIE would continue to be applied to FDI in the core accounts.
  - Supplementary information would present alternative balance of payments accounts where RIE is applied to portfolio investment and alternative measures of income, saving, and investment at national and institutional sector levels where RIE is applied to all equity investment.
  - An alternative measure of government deficit could be compiled and labeled as such.
- FITT consultation: majority favored Option 4 or 3 conceptually but expressed preference for Option 2 as pragmatic; of 13 FITT members expressing views, "eight" supported Option 2 either as first choice or as preferred for feasibility; among others three supported Option 4 (one as second choice), one Option 3, and one Option 5.
- DITT members later expressed split views between Option 1 and Option 5, while Option 2 received substantial support given feasibility difficulties with Options 3 and 4.

### Testing, endorsement, and implementation guidance requests
- Committee and AEG endorsed Option 2 to record allocation of retained earnings in a supplementary table for international and domestic transactions and positions, noting:
  - Initial split in views but pragmatic reasons (data collection challenges, implications for macro aggregates) led to endorsement of Option 2.
  - Request for additional guidance to external sector statistics (ESS) and national account compilers in updated statistical manuals.
- Feasibility testing among ESS and national account compilers involved:
  - Stage 1: asking about availability of source data for compiling RIE for PI equity (ESS) and domestic transactions (national accounts).
  - Stage 2: selected compilers provided experimental calculations.
  - Results presented at March 2022 Joint Committee and AEG Meeting and March 2023 AEG Meeting showed data availability remains challenging but existing methods exist for calculating RIE for PI equity and domestic RIE transactions.
  - Some members requested further compilation guidance on acceptable/recommended methods (to be provided in the new BPM Compilation Guide).
- Given practical difficulties, it was recommended that feasibility of Option 2 be tested, especially if Options 3 or 4 affecting core accounts were adopted due to larger resource implications and impact on headline macro indicators.

### Share buybacks — policy and measurement considerations (Annex 1 summary)
- Share buybacks: companies repurchase their own shares on the market; motives include distributing shares for stock option programs, later resale, or releasing funds to shareholders as substitute or addition to regular dividends.
- In last three decades, buybacks have developed strongly as a distribution method because:
  - (i) Share buybacks are often highly tax-efficient for shareholders as a whole.
  - (ii) Share buybacks are very flexible for companies (can be stopped without reputational effects associated with cutting dividends).
  - Some companies have policies of never distributing regular dividends and instead carry out heavy share buybacks.
- Current statistical treatment: share buybacks are treated as financial transactions (F.5) because sellers dispose of shares and receive cash (unchanged net worth); consistency requires the buyer (company) also records a financial transaction.
- Argument for expensing buybacks as income distribution (D.42):
  - From whole-shareholders’ perspective, buybacks release cash made on earnings in a different legal form than dividends.
  - Limited cases where companies distribute all earnings via buybacks imply no income is recorded in the SNA at present, potentially biasing income measurement, saving rates, and GNI.
  - Example: share buybacks have exceeded dividends paid out by quoted companies in the US for a number of years—exceeding "one trillion dollars a year"—close to "five percent of GDP".
- FITT and AEG broadly supported treating share buybacks as income distribution in core accounts but preferred a more detailed discussion in a separate/subsequent guidance note; issue moved to the post-review research agenda.

*Source: Guidance Note (GN) on asymmetric treatment of retained earnings.*

### 5.      In this sense, expensing share-buybacks while keeping the current core framework could thus be

### F2 Asymmetric treatment of retained earnings

### Expensing share buybacks as an intermediate approach
- The Guidance Note (GN) presents expensing share-buybacks while keeping the current core framework as an intermediate approach between Option 4 and Options 2/1.
- Rationale and effects:
  - This approach "would significantly increase the distribution of income across the economy (as Option 4 would do) but still stay within the boundary of Options 2/1 (which insist that the saving of corporations should not be set to zero)."
  - The key conceptual question is whether handing out cash to shareholders taps corporations' savings; the GN notes that "Many think they do, and de facto many companies borrow to do so."
- Accounting conventions:
  - Under IFRS, share buybacks are reported similarly to dividends as financing transactions; dividends are treated as income in the SNA by convention to measure shareholder (notably household) income.
  - Alternative views noted: some argue true income on equity is company earnings (dividend plus REI, D.42+D.43); others propose excluding dividends from income altogether.

### How to expense share buybacks (proposed imputation method)
- Key concerns about direct reclassification:
  - A direct reclassification of the cash seller transaction as D.42 is unreasonable because the seller is conducting a financial transaction, and such direct classification could create perverse incentives (e.g., governments reselling shares to generate revenue) and make GNI vary with nonresident participation.
  - Source data may not reliably identify whether a share sale was to the issuing company (buyback) or to the market; intermediated buybacks may make identification infeasible.
- Proposed imputation procedure (needs further investigation):
  - Impute a dividend (D.42) for the amount and at the time of a share buyback:
    - Record the imputed D.42 in the company accounts as a use/expenditure/debit.
    - Record an imputed D.42 as resource/revenue/credit across all shareholders, allocated using the REI allocation method.
    - The counterpart to the imputed D.42 is F.5 (in the same way as REI D.43).
  - In the company accounts the outflow of cash currently coded (-)F.5L becomes D.42.
  - In the shareholders' aggregate account there is an imputed D.42 against (+)F.5A.
  - The share seller records (-)F5.A against cash received (+)F2A, in addition to its apportioned D.42/(+)F.5A.
  - This preserves that the buyback transaction itself remains a financial transaction for the seller, while using buyback amount/time as the basis to impute a dividend applying the REI method.
  - The GN notes precedent: the REI method is already applied in other contexts (e.g., collective investment schemes D.443) with different rationale.
- Safeguards:
  - Because the imputed D.42 would be added to regular dividends, "the new total D.42 should nonetheless be superdividend tested."
  - Purpose of superdividend test: to ensure buybacks distributing period earnings are treated as income, while buybacks that essentially liquidate a large part of the company are treated as financial transactions.
- Treatment independence of subsequent use:
  - The proposal rejects varying recording by subsequent use of repurchased shares because classification should not depend on ulterior motive and because such information is generally unavailable; subsequent uses should be treated separately.

### Outcomes of the survey on data availability and testing for REI (portfolio investment)
- Survey participation:
  - "Eighty-one respondents from 75 countries completed the survey (Table 3)."
  - "Twenty-nine countries (39 percent) were interested in participating in the testing exercise for the reinvested earnings (REI) of portfolio investment (PI) assets and 31 countries (41 percent) for the REI of PI liabilities."
  - Lack of relevant source data was the main reason preventing participation.
- Source data availability (survey summary):
  - For portfolio investment assets:
    - "Seventy-seven percent of the respondent countries do not have the source data to compile the RIE of portfolio investment assets."
  - For portfolio investment liabilities:
    - "63 percent of countries do not have it for RIE of portfolio investment liabilities."
- Interest in experimental exercise:
  - "Thirty-nine percent of countries expressed interest to participate in the testing exercise for the REI of PI assets and 41 percent for the REI of PI liabilities."
  - Non-participation reasons: lack of resources and absence of source data.
- Plans to develop source data:
  - "Around 20 percent of the respondent countries would develop the source data to compile the supplementary tables."
  - Countries planning development intended to include additional tables in enterprise surveys; countries with large PI holdings often considered it infeasible to require corporations to report.
- Table 4 aggregate results (survey questionnaire summary):
  - For portfolio investment assets: Yes 16 (21%), No 59 (79%), Total 75 (100%).
  - For portfolio investment liabilities: Yes 26 (35%), No 47 (64%), No response 1 (1%), Total 75 (100%).
  - Interest in participation (PI assets): Yes 28 (37%), No 46 (61%), No response 1 (1%), Total 75 (100%).
  - Interest in participation (PI liabilities): Yes 30 (40%), No 43 (57%), No response 2 (3%), Total 75 (100%).
  - Plans to develop source data for PI assets: Yes 16 (21%), No 40 (53%), No response 19 (25%), Total 75 (100%).
  - Plans to develop source data for PI liabilities: Yes 13 (17%), No 32 (43%), No response 30 (40%), Total 75 (100%).

### Summary of the experimental calculation exercise for REI (portfolio investment)
- Participation and data provision:
  - Six respondents out of 31 volunteer countries responded to the second-phase questionnaire; only four provided data; one compiler could not undertake compilation due to other commitments.
- Methods used by respondents:
  - United States and Switzerland used aggregated stocks and aggregated performance indicators; they employed indirect indicators to estimate total earnings and dividends and then computed the residual.
  - US method:
    - Used price-earnings ratios (PER) (MSCI excluding US index for PI assets, and S&P 500 for PI liabilities) as a performance indicator.
    - Calculated REI percentage as (1 ÷ PER) – DY.
    - Calculated RE in USD as (RIE% ÷ 4) x ((S_t0 + S_t1) ÷ 2).
  - Switzerland used a similar aggregated method: applied average payout ratio from Bloomberg to IIP stocks; difference between payout and dividend payout represents REI.
  - Armenia and Portugal used more granular data:
    - Armenia compiled only liabilities using balance sheets and income statements of individual enterprises (net profit and dividends); nonfinancial corporations do not report any data for PI.
    - Portugal:
      - For PI assets used Centralized Securities Database (CSDB) (accrued income factor) and outstanding equity amounts by country and issuance sector; distribution of earnings obtained from direct monthly company reports; REI = estimated total income – distributed earnings.
      - For PI liabilities, used direct investment reinvested earnings weighted by percent owned by non-resident portfolio investors.
- Limitations and needs:
  - Aggregated performance indices introduce "noise" and composition differences; more granular, security-by-security data are needed for accurate calculations.
  - Several countries (Nicaragua, Botswana, Curaçao) could not compile the data.
  - Overall conclusion: current data collection systems are not designed to collect RIE data for portfolio investments; economies need additional databases containing information on (i) foreign companies, (ii) domestic companies with PI (similar to DI compilation), and (iii) security-by-security databases.

### Summary of survey on data availability for national accounts and experimental calculations for domestic transactions
- National Accounts consultation:
  - "A total of 44 respondents contributed to this consultation."
  - Topic relevance: high or medium relevance for most respondents.
- RIE for public corporations (domestically controlled):
  - "For 56 percent of respondents, institutions have source data available to compile RIE for public corporations as supplementary items."
  - Among those: 80 percent have GFS assets and liabilities (stock) data, 72 percent have GFS financial transaction data, and 24 percent have other source data (e.g., enterprise databases, financial statements, and Balance of Payments data).
  - "Of the respondents, whose institutions have these source data available, most (83 percent) institutions have access to detailed (audited) financial statement informal including notes for public corporations."
  - Sixteen institutions are interested in participating in an experimental estimate exercise to prepare RIE data for public corporations.
- RIE for private domestic equity stakes:
  - "Approximately 50 percent of all respondents’ institutions have relevant source data available to compile RIE for private domestic equity stakes."
  - Dominant data sources: financial statements, administrative tax data, and enterprise income surveys.
  - Fourteen institutions are interested to participate in an experimental estimate exercise to prepare RIE data for private domestic equity stakes.
- Experimental calculations among national accountants (13 economies responded):
  - Methodologies for RIE for public and private nonfinancial corporations:
    - RE1 = net operating surplus plus net property income plus net current transfers.
    - RE2 = net operating surplus plus net property income.
    - RE3 (Residual) = Profits after taxes less dividends or Profits before taxes less (taxes and dividends).
  - For public and private financial corporations:
    - RE4 for deposit-taking and other financial corporations = interest revenue minus interest payable, plus income from fees and commissions, plus net dividend income, minus costs (fees, staff expenses, administrative costs).
    - For central bank same as above, except no fees/commissions in typical cases.
    - RE3 (Residual) also identified as an option.
  - Data sources used: surveys, administrative data (tax schedules, regulatory), company financial statements; one economy used commercial databases (Dun & Bradstreet, Bureau van Dijk).
- Feasibility and issues:
  - Some economies can allocate RIE to equity holders annually or quarterly; estimation more plausible for private corporations than public entities.
  - Major concern: data required to allocate RIE to equity holders is often unavailable (shareholding information not always in administrative sources); estimates may exclude companies without public financial statements and SMEs.
- Recommendation from the national accounts consultation:
  - Endorse recommendation for economies to provide information for sectored RIE as supplementary items, not part of core accounts.
  - A supplementary table allows economies to investigate data sources and methods while ensuring consistency between the System of National Accounts and the Balance of Payments community (BPM7 has agreed to a supplementary table).
  - The AEG is asked to consider guidance for economies on compiling estimates, drawing on experience from countries such as Portugal and Ireland.

---


_Source: https://www.imf.org/-/media/files/data/statistics/bpm6/approved-guidance-notes/f2-asymmetric-treatment-of-retained-earnings-final.pdf_