## _tnm1401

## Source details

**Canonical URL:** [_tnm1401](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/tnm/2014/_tnm1401.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/tnm/2014/_tnm1401.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/tnm/2014/_tnm1401.pdf.json)

---

### Project purpose, scope, and timeline
- Project: IMF/DFID EDDI (External Data for Development Initiative) to upgrade BOP/IIP statistics in participating African countries.
- Six participating countries: The Gambia, Ghana, Kenya, Mauritius, Mozambique, and Nigeria.
- Core GDDS dissemination target for external sector aggregates: publication of an annual BOP statement within 6–9 months of the reference date; encouraged extension: an opening and closing annual IIP statement with the same timeliness requirement.
- Project start and methodological alignment:
  - EDDI began in April 2010.
  - Able to integrate BPM6 methodologies from the outset (BPM6 published in 2009).
  - CDIS emphasis following Coordinated Direct Investment Survey Guide (March 2010).

### Design and methodology highlights
- Primary project tasks:
  - Successful establishment of an annual enterprise survey.
  - Development of data sources for compiling portfolio investment.
- Enterprise survey design principles:
  - Smaller survey frames from the outset given budget constraints and skewed distribution of enterprises with foreign assets/liabilities; resulting frames for annual enterprise surveys have been less than 400.
  - Key phases in conducting an enterprise survey:
    - Establishment of a survey frame;
    - Questionnaire design;
    - Data collection;
    - Data review and validation;
    - Design of the database;
    - Grossing up techniques.
- Emphasis on CDIS participation:
  - With the exception of The Gambia, all EDDI countries agreed to participate in the CDIS.
  - Priority on inward direct investment reporting (immediate direct investor, fellow enterprises, reverse investment) and BPM6 changes (e.g., debt between selected affiliated financial corporations not classified as direct investment).

### Major methodological changes under BPM6 relevant to the project
- Instrument presentation preferred: equity securities, other equity, debt securities, currency and deposits, loans, trade credit and advances, other accounts receivable/payable, insurance, pension and standardized guarantees, financial derivatives and employee stock options, and Special Drawing Rights.
- More detailed classification of the financial sector: pension funds, insurance corporations, money market funds, investment funds, other financial intermediaries (implications for specialized questionnaires).
- Inclusion of standardized guarantees and stock options (not widely relevant in the region to date).
- Residence concept expanded to include corporate structures with little or no physical presence (major relevance for Mauritius and GBCs).

### Implementation experience — common findings and operational lessons
- Survey frames and business registers:
  - Common frame construction methods: listed companies, consultation with industry associations, regulatory bodies, investment agencies, social security funds.
  - Exploratory survey modules sometimes used to populate business registers.
  - Typical outcome: a census of enterprises meeting a foreign assets/liabilities threshold; skewed distribution where commonly 200–300 enterprises account for roughly three-quarters of BOP financial account transactions and positions.
  - Treatment of enterprise groups: consolidated reporting impractical; surveys conducted on an unconsolidated basis including resident subsidiaries/affiliates.
- Questionnaire design:
  - Built from draft BPM6CG forms (expanded Form 17/18 for other investment where needed).
  - Strategy: identify direct investment enterprises to complete the CDIS-based sections; other enterprises complete BPM6 "other investment" sections.
  - Some countries included services credits/debits in questionnaires (Kenya and Ghana).
- Data collection:
  - Poor electronic communications and unreliable mail led most countries to use enumerators to deliver and collect questionnaires.
  - Enumerators effective but costly; many countries saw response improve over successive surveys to over 70 percent for most countries.
  - Major enterprises generally required to provide questionnaire or at least a financial statement; response rates higher in this group.
- Data review and validation:
  - Best practices: internal questionnaire checks, cross-checks with previous survey responses, and verification against enterprise financial statements/annual reports.
  - Inward direct investment in equity (stocks and flows) generally highest accuracy due to correspondence with enterprise balance sheets.
  - Instrument breakdowns of external debt and resident/nonresident splits less reliable in many enterprises; notes to financial statements often used to detect misreporting.
  - Outward direct investment review is more demanding (need subsidiary/affiliate accounts abroad).
  - Databases designed to enable enterprise-by-enterprise scrolling, outlier detection, and cross-year consistency checks greatly enhance review.
- Database and IT:
  - Database requirements:
    - Facilitate compilation of BOP/IIP aggregates including CDIS needs;
    - Generate data review tables for enterprise-by-enterprise scanning;
    - Support grossing up techniques for consistency across periods;
    - Produce output tables and ad-hoc analyses.
  - Implementation status among participants:
    - Bank of Uganda served as a regional model.
    - Ghana and Mauritius have developed in-house data management software.
    - Nigeria and Mozambique using spreadsheet software; The Gambia taking steps.
    - Kenya used the MEFMI capital flows database and planned its own Access-based database.
  - Including IT staff in training workshops produced benefits.
- Grossing up and harmonization:
  - Grossing up needed when survey frames differ across surveys or for differential non-response.
  - Overlapping data between successive surveys or between surveys and non-survey sources is required for simplest grossing up.
  - Ghana executed a systematic grossing up to support a major revision of published BOP and IIP data in 2011.
- Quarterly surveys:
  - Envisaged once reliable annual surveys established.
  - Plan to survey approximately 20 large companies reporting quarterly accounts or management accounts to reduce burden.
  - Modified questionnaires developed for Mauritius and Ghana for quarterly use; preference for electronic reporting.
  - Mozambique already had an established quarterly enterprise survey with a frame of some 40 enterprises (used to supplement ITRS).
- Other direct reporting:
  - Scope to introduce reporting on direct investment in real estate and to add services and income (employee compensation) items to enterprise questionnaires.
  - For inward portfolio investment in listed companies, local stock exchanges/capital markets authorities often provide better transaction and position data than enterprise surveys.
  - Custodian bank reporting explored but not widely introduced due to bank reporting burdens.

### Source data changes and implications for ITRS
- Risks with ITRS:
  - Increased unreliability in many countries in an environment of relaxed exchange controls.
  - Misclassification risks when many small over-the-counter transactions exist, complex instruments are used, or customers conduct BOP transactions through accounts with banks abroad.
  - Misclassifications revealed by comparisons with enterprise survey data (e.g., Ghana and Kenya).
- ITRS redesign:
  - ITRS must function as a predictor of more reliable direct reporting sources and remain useful for transactions where direct reporting is infeasible (e.g., household portfolio investment abroad).
  - Redesign of ITRS classifications and residence definitions is a key late-stage project task and will likely lead to substantial revisions when reconciled with direct reporting and IIP data.

### Cooperation, capacity building, and donor interaction
- Regional and south–south cooperation was central: Bank of Uganda and Bank of Ghana provided mentoring and database assistance.
- Donor support examples:
  - DFID financed early enterprise surveys in The Gambia; World Bank financed Kenya’s initial enterprise survey under STATCAP.
  - MEFMI provided database and questionnaire support for Kenya’s first EDDI survey.
- Project encouraged internal funding and sustainability: with the exception of Kenya, compiling agencies expected to fund ongoing surveys internally.

### Country progress snapshots (status as of July 2013; objectives to March 2015)
- The Gambia:
  - Past DFID-funded surveys existed but loss of staff/institutional memory hindered progress.
  - Frame ~100 enterprises; recent missions reviewed ~40 larger enterprises for 2010–2011.
  - Planned completion date: April 2015 to re-establish annual enterprise survey and use resulting data through 2012.
- Ghana:
  - Established annual Foreign Private Capital Flows Survey (FPCF); frame expanded to some 320 enterprises.
  - Built database capability; integrated survey results into published BOP and IIP but publication timing delayed beyond GDDS timeliness.
  - Pilot quarterly FPCF design in progress with a 20-company frame for quarterly reporting.
  - Focus ongoing on inward portfolio investment data sources and reporting from Securities Exchange Commission.
- Kenya:
  - 2010 Foreign Investment Survey funded under World Bank STATCAP; preliminary IIP and revised BOP compiled.
  - 2013 FIS planned with a 500-enterprise frame drawn from 2010 responses; in-house Access-based database planned.
  - Aim to replace initial ITRS-based estimates with FIS-driven BOP/IIP statistics.
- Mauritius:
  - Three enterprise surveys completed; 2013 FALS underway with frame ~590 enterprises covering 2011 and 2012.
  - Implemented annual GBC (global business companies) surveys for GBC1s; GBC1 data included in annual IIP from 2009 and in quarterly BOP estimates from 2010.
  - Non-response by ~30 major enterprises remains an issue; plan to use ITRS and financial statements to support grossing up.
  - Preparing enhancements for SDDS Plus adherence and quarterly IIP compilation timeliness considerations for 2014.
- Mozambique:
  - Quarterly enterprise survey already established (frame ~40) and annual FALS being developed (target frame ~200).
  - Work focused on integrating annual FALS, quarterly FALS, and monthly ITRS and on portfolio investment and oil & gas direct investment data sources.
  - Cash versus accruals differences noted; publication of fully integrated IIP delayed but progress ongoing.
- Nigeria:
  - 2010 CDIS inward survey covered 250 enterprises; 2011 SOFAL expanded to 270 with improved 85 percent response (excluded free trade zones).
  - 2012 SOFAL frame 320 enterprises; high response maintained, improved JVC responses, but incomplete free trade zone coverage.
  - In-house database development planned, with field learning visit to Bank of Ghana arranged for June 2013.
  - Expectation that SOFAL will become preferred source for BOP financial account transactions; redesign of ITRS will be required.

### Key operational metrics and figures preserved from the source
- Participating countries: 6 (The Gambia, Ghana, Kenya, Mauritius, Mozambique, Nigeria).
- GDDS timeliness: 6–9 months for annual BOP; same timeliness proposed for annual IIP as an encouraged extension.
- Project start: April 2010.
- BPM6 publication year: 2009.
- CDIS Guide release: March 2010.
- Typical annual enterprise survey frames used in EDDI: less than 400.
- Common large-frame contributor group: commonly 200–300 enterprises accounting for roughly three-quarters of BOP financial account transactions and positions.
- Reported survey response: over 70 percent for most countries (after successive surveys).
- Example country frames and counts cited:
  - Ghana expanded frame to some 320 enterprises.
  - Mauritius 2013 FALS frame around 590 enterprises.
  - Mozambique initial quarterly frame some 40 enterprises; planned annual frame some 200 enterprises.
  - Kenya planned 2013 FIS frame of 500 enterprises (from 2010 results).
  - Nigeria SOFAL frames: 2010/2011/2012 reported frames of 250, 270, and 320 enterprises respectively; 2011 response 85 percent.
- Selected completion targets mentioned:
  - Ghana: use of grossing up techniques to support 2011 BOP/IIP revision (completed).
  - The Gambia: completion date set for re-establishing annual enterprise survey and using results through 2012 — April 2015.

### Principal policy and operational recommendations distilled from project experience
- Prioritize establishment of annual enterprise surveys focused on BOP/IIP needs, with smaller initial frames (less than 400) targeting enterprises with significant foreign assets/liabilities.
- Use exploratory modules and stakeholder consultations (industry associations, stock exchanges, regulators) to build and validate business registers and survey frames.
- Design questionnaires aligned with BPM6 and CDIS model forms; minimize additional non-BOP/IIP questions to preserve response rates.
- Require enterprises to enclose latest financial statements/annual reports to support data review and validation.
- Use enumerators where necessary to secure response, recognizing cost trade-offs; aim to transition to electronic reporting where feasible.
- Develop in-house data management software (Access or equivalent) to enable enterprise-level scrolling, cross-year comparisons, grossing up, and automated output tables; include IT staff in training.
- Implement grossing up techniques where survey frames and responses vary across years; ensure overlapping data sources to support grossing up and historical revisions.
- Redesign ITRS classifications and residence definitions to be mutually supportive with direct reporting and to act as reliable first estimates pending direct reporting.
- Sequence reforms: establish reliable annual surveys first, pilot a small-frame quarterly survey (≈20 firms), then expand quarterly reporting and specialized industry questionnaires (e.g., oil & gas, GBCs, real estate) as capacity permits.
- Promote regional peer learning and south–south cooperation (e.g., Bank of Uganda, Bank of Ghana, MEFMI) and coordinate donor support to catalyze sustainable, internally financed survey cycles.

### Questionnaire design, scope, data collection, review, and grossing up (section includes borrowing by direct investment enterprises from unrelated parties abroad)
- Questionnaire design and scope:
  - Required categories follow the format for reporting the IIP in Part F of the survey form 17 in the draft BPM6CG.
  - Distinctions and inclusions:
    - Direct investment borrowing from unrelated enterprises abroad and the outward equivalent.
    - Portfolio investment split between listed companies (transactions and positions valued at market prices, often from local stock exchange reporting) and unlisted companies (valued in relation to the book value of shareholders capital, typically from enterprise surveys).
    - Financial derivatives may be included in instrument classification but are often not reported or reported inaccurately. Two countries (Mozambique and Ghana) designed separate financial derivatives questionnaires for banks following a Bank of Uganda example.
    - All questionnaires follow BPM6 classification of external debt instruments between unrelated enterprises (loans, trade credit and advances, accounts payable/receivable) with breakdown by long and short-term contractual maturity.
    - Separate reporting usually included for insurance technical reserves, pension entitlements, employee stock options, investment fund shares, and financial derivatives, mainly to determine existence of cross-border liabilities and claims.
- Data collection practices:
  - Electronic collection (mail-outs, email) has been disappointing in the region; effective electronic reporting requires an established working relationship and is best for larger enterprises.
  - Most participating countries found enumerators to be the most effective method to deliver and collect questionnaires despite higher cost:
    - Common practice: enumerator presents questionnaire and returns after two week to collect completed questionnaire.
    - Enumerators contracted for the weeks in which data collection is scheduled.
    - Compensation often has two components: travel/per diem and payment for submission of mandated sections of completed questionnaires (commonly comprising all data needed for BOP and IIP compilation).
  - For first-time large-frame surveys, to avoid incentives for incomplete returns, paying only travel and per diem for enumerators may be preferable.
  - Establishing the survey frame before conducting the survey is essential to avoid frictions and poor coverage.
- Data review procedures and checks:
  - Multi-stage review process evolved as best practice:
    - (i) Initial consistency checks by enumerators during enterprise visit.
    - (ii) Supplementary consistency checks by BOP compilers before data entry, including checks against unconsolidated and consolidated financial statements.
    - (iii) Follow-up contact by assigned BOP compiler for unresolved queries using contact information on completed questionnaires.
    - (iv) Final review by a senior BOP compiler scrolling through the survey database enterprise-by-enterprise to identify gaps and outliers.
  - Key field checks required of enumerators:
    - Confirm enterprises reporting no BOP transactions/IIP positions have no nonresident transactions or foreign liabilities/assets, evidenced by financial statements; still collect contact information.
    - If enterprise financials use a foreign currency as unit of account, review conversion procedures into domestic currency.
    - Verify shareholding structure against enterprise financial statements.
    - For equity liabilities to nonresidents: ensure direct investment (10 percent or more) questions are completed; portfolio investment (less than 10 percent) question completed; internal consistency of data entered.
    - For external debt liabilities to nonresidents: ensure enterprises in direct investment relationships report financing from direct investor/fellow enterprises and from unrelated nonresidents; all enterprises report external debt by instrument categories; currency of denomination and country of creditor questions completed.
    - For income: retained earnings should equal net income after tax less dividends paid or due (not the stock of retained earnings reported under shareholders capital).
    - For investment abroad: where direct investment relationship (10 percent or more), report shareholders capital breakdown, income, and dividends paid for each subsidiary/affiliate abroad identified.
  - Priority in second-round compiler review: focus on larger enterprises (suggested example: largest 50 enterprises by size of foreign liabilities/assets).
  - Common data review issues:
    - Distinguishing resident vs nonresident banking/borrowing in enterprise balance sheets.
    - Use notes to financial statements on related party transactions and positions to identify intercompany borrowing from abroad.
- Data management software versus spreadsheets:
  - Excel spreadsheets create constraints (memory limits, maintenance issues, decentralization, corruption risk) as survey frames and outputs expand.
  - Data review tables in databases facilitate enterprise-by-enterprise scrolling to detect gaps/outliers; Bank of Uganda’s capital flows database used as a model.
  - Countries developing in-house databases with data review tables: Ghana, Kenya, Mauritius, Nigeria.
- Grossing up techniques and non-response estimation:
  - Grossing up follows draft BPM6CG recommendations and addresses:
    - (i) estimates for non-responding enterprises;
    - (ii) revisions for comparability across surveys with differing frames/response rates;
    - (iii) carry back/forward estimates for years without survey data.
  - Principle: find non-survey data series with observable relationships to survey data and apply grossing up across industries or size-thresholded enterprises.
  - Hansen and Hurwitz methodology applied:
    - Equation provided in source with defined symbols including: N = total number of enterprises in the frame, n = number of surveyed enterprises, x̄1' = sample average for enterprises responding during main survey round, m = number responding during main round, x̄2" = average for enterprises responding during follow-up round, δ = number of enterprises that did not respond during main round, k < δ and n = m + δ and m + k < n.
    - Key assumption: reporting enterprises during follow-up (k) have similar characteristics to those that do not respond at all (δ-k); best suited where population can be stratified into homogeneous groups.
  - Application examples:
    - Ghana applied stratification into four groups by stock of equity and used the equation to gross up across large-frame and small-frame surveys.
    - For BOP current account items not derived as changes in end-of-period positions, ratios from a base year applied to updated book values to estimate dividends and interest (examples provided in source using end–2007 to end–2008 ratios).
  - Recommended practice once annual surveys stabilize:
    - Estimate for non-responding enterprises on an enterprise-by-enterprise basis using previous survey responses and sector-level indicators; selective application advised where staffing constraints exist.
  - Rule of thumb from EDDI experience: an adequate survey response is one in which nearly all major enterprises and at least 70 percent of all other enterprises in the survey frame have reported.

### Portfolio investment data collection and valuation
- Best practice guidance from draft BPM6CG and Coordinated Portfolio Investment Survey Guide used as starting point.
- Inward portfolio investment liabilities:
  - Enterprise surveys provide reliable data on number of shares owned by nonresidents (share registers).
  - Enterprises usually report portfolio investment at book values (percentage of shares held applied to shareholders capital in balance sheet); transactions reported as change in book values adjusted for revaluations attributable to portfolio investors.
  - Local stock exchange/capital markets authority often best source for listed company holdings and may report transactions and positions at market prices (inclusive of accrued dividends in principle, but in practice many report equity prices excluding accrued dividends; no current adjustments made to include accrued but not paid dividends/interest).
  - Custodian data and securities databases are alternative sources but have limitations (distinguishing direct vs portfolio investment; reluctance to mandate custodian reporting).
  - Government bond holdings by nonresidents obtained from government bond registers or central depository systems (usually managed by the central bank).
  - Securities issued on foreign capital markets: partner country sources (e.g., BIS database on international bond issues) assumed holders are nonresident; enterprise questionnaires did not generally request information on foreign market issues due to rarity.
- Issues and special cases:
  - Nominee accounts and round-tripping obscure investor residence; legal disclosure requirements vary.
  - Portfolio investment assets: enterprises report holdings as holders of securities; banks may distinguish banking book (book values) and investment book (market prices).
  - Households’ portfolio investment abroad: household surveys are unreliable; broker surveys may be used when households commission brokers to buy/sell securities issued abroad.
  - Cross-listed securities: local stock exchange data and broker information used to estimate resident purchases of nonresident cross-listed securities.

*Source: IMF Technical Notes and Manuals, Part II of the EDDI project report (content as provided).*

### Part II reports on the project’s progress and results thus far, highlighting the experiences of

### _tnm1401 - Part II reports on the project’s progress and results thus far, highlighting the experiences of 

### Project purpose, scope, and timeline
- Project: IMF/DFID EDDI (External Data for Development Initiative) to upgrade BOP/IIP statistics in participating African countries.
- Six participating countries: The Gambia, Ghana, Kenya, Mauritius, Mozambique, and Nigeria.
- Core GDDS dissemination target for external sector aggregates: publication of an annual BOP statement within 6–9 months of the reference date; encouraged extension: an opening and closing annual IIP statement with the same timeliness requirement.
- Project start and methodological alignment:
  - EDDI began in April 2010.
  - Able to integrate BPM6 methodologies from the outset (BPM6 published in 2009).
  - CDIS emphasis following Coordinated Direct Investment Survey Guide (March 2010).

### Design and methodology highlights
- Primary project tasks:
  - Successful establishment of an annual enterprise survey.
  - Development of data sources for compiling portfolio investment.
- Enterprise survey design principles:
  - Smaller survey frames from the outset given budget constraints and skewed distribution of enterprises with foreign assets/liabilities; resulting frames for annual enterprise surveys have been less than 400.
  - Key phases in conducting an enterprise survey:
    - Establishment of a survey frame;
    - Questionnaire design;
    - Data collection;
    - Data review and validation;
    - Design of the database;
    - Grossing up techniques.
- Emphasis on CDIS participation:
  - With the exception of The Gambia, all EDDI countries agreed to participate in the CDIS.
  - Priority on inward direct investment reporting (immediate direct investor, fellow enterprises, reverse investment) and BPM6 changes (e.g., debt between selected affiliated financial corporations not classified as direct investment).

### Major methodological changes under BPM6 relevant to the project
- Instrument presentation preferred: equity securities, other equity, debt securities, currency and deposits, loans, trade credit and advances, other accounts receivable/payable, insurance, pension and standardized guarantees, financial derivatives and employee stock options, and Special Drawing Rights.
- More detailed classification of the financial sector: pension funds, insurance corporations, money market funds, investment funds, other financial intermediaries (implications for specialized questionnaires).
- Inclusion of standardized guarantees and stock options (not widely relevant in the region to date).
- Residence concept expanded to include corporate structures with little or no physical presence (major relevance for Mauritius and GBCs).

### Implementation experience — common findings and operational lessons
- Survey frames and business registers:
  - Common frame construction methods: listed companies, consultation with industry associations, regulatory bodies, investment agencies, social security funds.
  - Exploratory survey modules sometimes used to populate business registers.
  - Typical outcome: a census of enterprises meeting a foreign assets/liabilities threshold; skewed distribution where commonly 200–300 enterprises account for roughly three-quarters of BOP financial account transactions and positions.
  - Treatment of enterprise groups: consolidated reporting impractical; surveys conducted on an unconsolidated basis including resident subsidiaries/affiliates.
- Questionnaire design:
  - Built from draft BPM6CG forms (expanded Form 17/18 for other investment where needed).
  - Strategy: identify direct investment enterprises to complete the CDIS-based sections; other enterprises complete BPM6 "other investment" sections.
  - Some countries included services credits/debits in questionnaires (Kenya and Ghana).
- Data collection:
  - Poor electronic communications and unreliable mail led most countries to use enumerators to deliver and collect questionnaires.
  - Enumerators effective but costly; many countries saw response improve over successive surveys to over 70 percent for most countries.
  - Major enterprises generally required to provide questionnaire or at least a financial statement; response rates higher in this group.
- Data review and validation:
  - Best practices: internal questionnaire checks, cross-checks with previous survey responses, and verification against enterprise financial statements/annual reports.
  - Inward direct investment in equity (stocks and flows) generally highest accuracy due to correspondence with enterprise balance sheets.
  - Instrument breakdowns of external debt and resident/nonresident splits less reliable in many enterprises; notes to financial statements often used to detect misreporting.
  - Outward direct investment review is more demanding (need subsidiary/affiliate accounts abroad).
  - Databases designed to enable enterprise-by-enterprise scrolling, outlier detection, and cross-year consistency checks greatly enhance review.
- Database and IT:
  - Database requirements:
    - Facilitate compilation of BOP/IIP aggregates including CDIS needs;
    - Generate data review tables for enterprise-by-enterprise scanning;
    - Support grossing up techniques for consistency across periods;
    - Produce output tables and ad-hoc analyses.
  - Implementation status among participants:
    - Bank of Uganda served as a regional model.
    - Ghana and Mauritius have developed in-house data management software.
    - Nigeria and Mozambique using spreadsheet software; The Gambia taking steps.
    - Kenya used the MEFMI capital flows database and planned its own Access-based database.
  - Including IT staff in training workshops produced benefits.
- Grossing up and harmonization:
  - Grossing up needed when survey frames differ across surveys or for differential non-response.
  - Overlapping data between successive surveys or between surveys and non-survey sources is required for simplest grossing up.
  - Ghana executed a systematic grossing up to support a major revision of published BOP and IIP data in 2011.
- Quarterly surveys:
  - Envisaged once reliable annual surveys established.
  - Plan to survey approximately 20 large companies reporting quarterly accounts or management accounts to reduce burden.
  - Modified questionnaires developed for Mauritius and Ghana for quarterly use; preference for electronic reporting.
  - Mozambique already had an established quarterly enterprise survey with a frame of some 40 enterprises (used to supplement ITRS).
- Other direct reporting:
  - Scope to introduce reporting on direct investment in real estate and to add services and income (employee compensation) items to enterprise questionnaires.
  - For inward portfolio investment in listed companies, local stock exchanges/capital markets authorities often provide better transaction and position data than enterprise surveys.
  - Custodian bank reporting explored but not widely introduced due to bank reporting burdens.

### Source data changes and implications for ITRS
- Risks with ITRS (International Transactions Reporting Systems):
  - Increased unreliability in many countries in an environment of relaxed exchange controls.
  - Misclassification risks when many small over-the-counter transactions exist, complex instruments are used, or customers conduct BOP transactions through accounts with banks abroad.
  - Misclassifications revealed by comparisons with enterprise survey data (e.g., Ghana and Kenya).
- ITRS redesign:
  - ITRS must function as a predictor of more reliable direct reporting sources and remain useful for transactions where direct reporting is infeasible (e.g., household portfolio investment abroad).
  - Redesign of ITRS classifications and residence definitions is a key late-stage project task and will likely lead to substantial revisions when reconciled with direct reporting and IIP data.

### Cooperation, capacity building, and donor interaction
- Regional and south–south cooperation was central: Bank of Uganda and Bank of Ghana provided mentoring and database assistance.
- Donor support examples:
  - DFID financed early enterprise surveys in The Gambia; World Bank financed Kenya’s initial enterprise survey under STATCAP.
  - MEFMI provided database and questionnaire support for Kenya’s first EDDI survey.
- Project encouraged internal funding and sustainability: with the exception of Kenya, compiling agencies expected to fund ongoing surveys internally.

### Country progress snapshots (status as of July 2013; objectives to March 2015)
- The Gambia:
  - Past DFID-funded surveys existed but loss of staff/institutional memory hindered progress.
  - Frame ~100 enterprises; recent missions reviewed ~40 larger enterprises for 2010–2011.
  - Planned completion date: April 2015 to re-establish annual enterprise survey and use resulting data through 2012.
- Ghana:
  - Established annual Foreign Private Capital Flows Survey (FPCF); frame expanded to some 320 enterprises.
  - Built database capability; integrated survey results into published BOP and IIP but publication timing delayed beyond GDDS timeliness.
  - Pilot quarterly FPCF design in progress with a 20-company frame for quarterly reporting.
  - Focus ongoing on inward portfolio investment data sources and reporting from Securities Exchange Commission.
- Kenya:
  - 2010 Foreign Investment Survey funded under World Bank STATCAP; preliminary IIP and revised BOP compiled.
  - 2013 FIS planned with a 500-enterprise frame drawn from 2010 responses; in-house Access-based database planned.
  - Aim to replace initial ITRS-based estimates with FIS-driven BOP/IIP statistics.
- Mauritius:
  - Three enterprise surveys completed; 2013 FALS underway with frame ~590 enterprises covering 2011 and 2012.
  - Implemented annual GBC (global business companies) surveys for GBC1s; GBC1 data included in annual IIP from 2009 and in quarterly BOP estimates from 2010.
  - Non-response by ~30 major enterprises remains an issue; plan to use ITRS and financial statements to support grossing up.
  - Preparing enhancements for SDDS Plus adherence and quarterly IIP compilation timeliness considerations for 2014.
- Mozambique:
  - Quarterly enterprise survey already established (frame ~40) and annual FALS being developed (target frame ~200).
  - Work focused on integrating annual FALS, quarterly FALS, and monthly ITRS and on portfolio investment and oil & gas direct investment data sources.
  - Cash versus accruals differences noted; publication of fully integrated IIP delayed but progress ongoing.
- Nigeria:
  - 2010 CDIS inward survey covered 250 enterprises; 2011 SOFAL expanded to 270 with improved 85 percent response (excluded free trade zones).
  - 2012 SOFAL frame 320 enterprises; high response maintained, improved JVC responses, but incomplete free trade zone coverage.
  - In-house database development planned, with field learning visit to Bank of Ghana arranged for June 2013.
  - Expectation that SOFAL will become preferred source for BOP financial account transactions; redesign of ITRS will be required.

### Key operational metrics and figures preserved from the source
- Participating countries: 6 (The Gambia, Ghana, Kenya, Mauritius, Mozambique, Nigeria).
- GDDS timeliness: 6–9 months for annual BOP; same timeliness proposed for annual IIP as an encouraged extension.
- Project start: April 2010.
- BPM6 publication year: 2009.
- CDIS Guide release: March 2010.
- Typical annual enterprise survey frames used in EDDI: less than 400.
- Common large-frame contributor group: commonly 200–300 enterprises accounting for roughly three-quarters of BOP financial account transactions and positions.
- Reported survey response: over 70 percent for most countries (after successive surveys).
- Example country frames and counts cited:
  - Ghana expanded frame to some 320 enterprises.
  - Mauritius 2013 FALS frame around 590 enterprises.
  - Mozambique initial quarterly frame some 40 enterprises; planned annual frame some 200 enterprises.
  - Kenya planned 2013 FIS frame of 500 enterprises (from 2010 results).
  - Nigeria SOFAL frames: 2010/2011/2012 reported frames of 250, 270, and 320 enterprises respectively; 2011 response 85 percent.
- Selected completion targets mentioned:
  - Ghana: use of grossing up techniques to support 2011 BOP/IIP revision (completed).
  - The Gambia: completion date set for re-establishing annual enterprise survey and using results through 2012 — April 2015.

### Principal policy and operational recommendations distilled from project experience
- Prioritize establishment of annual enterprise surveys focused on BOP/IIP needs, with smaller initial frames (less than 400) targeting enterprises with significant foreign assets/liabilities.
- Use exploratory modules and stakeholder consultations (industry associations, stock exchanges, regulators) to build and validate business registers and survey frames.
- Design questionnaires aligned with BPM6 and CDIS model forms; minimize additional non-BOP/IIP questions to preserve response rates.
- Require enterprises to enclose latest financial statements/annual reports to support data review and validation.
- Use enumerators where necessary to secure response, recognizing cost trade-offs; aim to transition to electronic reporting where feasible.
- Develop in-house data management software (Access or equivalent) to enable enterprise-level scrolling, cross-year comparisons, grossing up, and automated output tables; include IT staff in training.
- Implement grossing up techniques where survey frames and responses vary across years; ensure overlapping data sources to support grossing up and historical revisions.
- Redesign ITRS classifications and residence definitions to be mutually supportive with direct reporting and to act as reliable first estimates pending direct reporting.
- Sequence reforms: establish reliable annual surveys first, pilot a small-frame quarterly survey (≈20 firms), then expand quarterly reporting and specialized industry questionnaires (e.g., oil & gas, GBCs, real estate) as capacity permits.
- Promote regional peer learning and south–south cooperation (e.g., Bank of Uganda, Bank of Ghana, MEFMI) and coordinate donor support to catalyze sustainable, internally financed survey cycles.

*Source: IMF Technical Notes and Manuals, Part II of the EDDI project report (content as provided).*

### section includes borrowing by direct investment enterprises from unrelated parties abroad

### _tnm1401 - section includes borrowing by direct investment enterprises from unrelated parties abroad

### Questionnaire design and scope
- Required categories follow the format for reporting the IIP in Part F of the survey form 17 in the draft BPM6CG.
- Distinctions and inclusions:
  - Direct investment borrowing from unrelated enterprises abroad and the outward equivalent.
  - Portfolio investment split between listed companies (transactions and positions valued at market prices, often from local stock exchange reporting) and unlisted companies (valued in relation to the book value of shareholders capital, typically from enterprise surveys).
  - Financial derivatives may be included in instrument classification but are often not reported or reported inaccurately. Two countries (Mozambique and Ghana) designed separate financial derivatives questionnaires for banks following a Bank of Uganda example.
  - All questionnaires follow BPM6 classification of external debt instruments between unrelated enterprises (loans, trade credit and advances, accounts payable/receivable) with breakdown by long and short-term contractual maturity.
  - Separate reporting usually included for insurance technical reserves, pension entitlements, employee stock options, investment fund shares, and financial derivatives, mainly to determine existence of cross-border liabilities and claims.

### Data collection practices
- Electronic collection (mail-outs, email) has been disappointing in the region; effective electronic reporting requires an established working relationship and is best for larger enterprises.
- Most participating countries found enumerators to be the most effective method to deliver and collect questionnaires despite higher cost:
  - Common practice: enumerator presents questionnaire and returns after two week to collect completed questionnaire.
  - Enumerators contracted for the weeks in which data collection is scheduled.
  - Compensation often has two components: travel/per diem and payment for submission of mandated sections of completed questionnaires (commonly comprising all data needed for BOP and IIP compilation).
- For first-time large-frame surveys, to avoid incentives for incomplete returns, paying only travel and per diem for enumerators may be preferable.
- Establishing the survey frame before conducting the survey is essential to avoid frictions and poor coverage.

### Data review procedures and checks
- Multi-stage review process evolved as best practice:
  - (i) Initial consistency checks by enumerators during enterprise visit.
  - (ii) Supplementary consistency checks by BOP compilers before data entry, including checks against unconsolidated and consolidated financial statements.
  - (iii) Follow-up contact by assigned BOP compiler for unresolved queries using contact information on completed questionnaires.
  - (iv) Final review by a senior BOP compiler scrolling through the survey database enterprise-by-enterprise to identify gaps and outliers.
- Key field checks required of enumerators:
  - Confirm enterprises reporting no BOP transactions/IIP positions have no nonresident transactions or foreign liabilities/assets, evidenced by financial statements; still collect contact information.
  - If enterprise financials use a foreign currency as unit of account, review conversion procedures into domestic currency.
  - Verify shareholding structure against enterprise financial statements.
  - For equity liabilities to nonresidents: ensure direct investment (10 percent or more) questions are completed; portfolio investment (less than 10 percent) question completed; internal consistency of data entered.
  - For external debt liabilities to nonresidents: ensure enterprises in direct investment relationships report financing from direct investor/fellow enterprises and from unrelated nonresidents; all enterprises report external debt by instrument categories; currency of denomination and country of creditor questions completed.
  - For income: retained earnings should equal net income after tax less dividends paid or due (not the stock of retained earnings reported under shareholders capital).
  - For investment abroad: where direct investment relationship (10 percent or more), report shareholders capital breakdown, income, and dividends paid for each subsidiary/affiliate abroad identified.
- Priority in second-round compiler review: focus on larger enterprises (suggested example: largest 50 enterprises by size of foreign liabilities/assets).
- Common data review issues:
  - Distinguishing resident vs nonresident banking/borrowing in enterprise balance sheets.
  - Use notes to financial statements on related party transactions and positions to identify intercompany borrowing from abroad.
- Data management software versus spreadsheets:
  - Excel spreadsheets create constraints (memory limits, maintenance issues, decentralization, corruption risk) as survey frames and outputs expand.
  - Data review tables in databases facilitate enterprise-by-enterprise scrolling to detect gaps/outliers; Bank of Uganda’s capital flows database used as a model.
  - Countries developing in-house databases with data review tables: Ghana, Kenya, Mauritius, Nigeria.

### Grossing up techniques and non-response estimation
- Grossing up follows draft BPM6CG recommendations and addresses:
  - (i) estimates for non-responding enterprises;
  - (ii) revisions for comparability across surveys with differing frames/response rates;
  - (iii) carry back/forward estimates for years without survey data.
- Principle: find non-survey data series with observable relationships to survey data and apply grossing up across industries or size-thresholded enterprises.
- Hansen and Hurwitz methodology applied:
  - Equation provided: x' = N * ( x̄1' * (m/n) + x̄2" * (δ/n) )  (presented in source as an equation with defined symbols)
  - Definitions preserved from source: N = total number of enterprises in the frame, n = number of surveyed enterprises, x̄1' = sample average for enterprises responding during main survey round, m = number responding during main round, x̄2" = average for enterprises responding during follow-up round, δ = number of enterprises that did not respond during main round, k < δ and n = m + δ and m + k < n.
  - Key assumption: reporting enterprises during follow-up (k) have similar characteristics to those that do not respond at all (δ-k); best suited where population can be stratified into homogeneous groups.
- Application examples:
  - Ghana applied stratification into four groups by stock of equity and used equation 1 to gross up across large-frame and small-frame surveys.
  - For BOP current account items not derived as changes in end-of-period positions, ratios from a base year applied to updated book values to estimate dividends and interest (examples: ratio of dividends declared to book value of equity at end–2007 applied to updated book value at end–2008; ratio of interest due to nominal value of debt at end–2007 applied to updated debt at end–2008).
- Recommended practice once annual surveys stabilize:
  - Estimate for non-responding enterprises on an enterprise-by-enterprise basis using previous survey responses and sector-level indicators; selective application advised where staffing constraints exist.
- Rule of thumb from EDDI experience: an adequate survey response is one in which nearly all major enterprises and at least 70 percent of all other enterprises in the survey frame have reported.

### Portfolio investment data collection and valuation
- Best practice guidance from draft BPM6CG and Coordinated Portfolio Investment Survey Guide used as starting point.
- Inward portfolio investment liabilities:
  - Enterprise surveys provide reliable data on number of shares owned by nonresidents (share registers).
  - Enterprises usually report portfolio investment at book values (percentage of shares held applied to shareholders capital in balance sheet); transactions reported as change in book values adjusted for revaluations attributable to portfolio investors.
  - Local stock exchange/capital markets authority often best source for listed company holdings and may report transactions and positions at market prices (inclusive of accrued dividends in principle, but in practice many report equity prices excluding accrued dividends; no current adjustments made to include accrued but not paid dividends/interest).
  - Custodian data and securities databases are alternative sources but have limitations (distinguishing direct vs portfolio investment; reluctance to mandate custodian reporting).
  - Government bond holdings by nonresidents obtained from government bond registers or central depository systems (usually managed by the central bank).
  - Securities issued on foreign capital markets: partner country sources (e.g., BIS database on international bond issues) assumed holders are nonresident; enterprise questionnaires did not generally request information on foreign market issues due to rarity.
- Issues and special cases:
  - Nominee accounts and round-tripping obscure investor residence; legal disclosure requirements vary.
  - Portfolio investment assets: enterprises report holdings as holders of securities; banks may distinguish banking book (book values) and investment book (market prices).
  - Households’ portfolio investment abroad: household surveys are unreliable; broker surveys may be used when households commission brokers to buy/sell securities issued abroad.
  - Cross-listed securities: local stock exchange data and broker information used to estimate resident purchases of nonresident cross-listed securities.

### Lessons from Ghana (experience and operational lessons)
- Ghana and Mauritius made most progress in establishing annual enterprise surveys, compiling annual IIP statements based on direct reporting, and integrating IIP with BOP financial account.
- Ghana practical lessons and milestones:
  - Initial FPCF surveys: 2008 (about 280 enterprises) and 2009 (shortened questionnaire, more enumerators, improved training); initial survey response 70 percent, improved through editing and follow-up.
  - Exploratory survey module added to an existing large-frame GSS survey to identify enterprises meeting foreign liabilities/assets threshold.
  - Survey frame expanded to 320 enterprises after exploratory survey; response increased to 85 percent with increased follow-up time.
  - Emphasis on identifying capital-intensive sectors (e.g., oil and gas) and major enterprises within them; oil and gas sector data remained a challenge.
  - Grossing up techniques applied to obtain comparable series across differing frames and response rates; preference for enterprise-by-enterprise estimation where feasible.
  - Published revised BOP and IIP data integrated changes in IIP with transactions, revaluations, and other volume changes based on enterprise accounts; Ghana published revised IIP and BOP financial account data incorporating survey results.
  - Database development: Bank of Ghana compiled a database using Access-based data management software modeled on Bank of Uganda; staffing and funding were increased (BOP unit staff expansion from eight to ten plus two IT staff) to support survey and database work.

### Summary of conclusions and achievements across participating countries
- Project setup and funding:
  - All six countries secured internal funding for annual enterprise surveys.
  - High-level buy-in sought through missions prior to the opening workshop and design of country work programs.
- Survey frames and questionnaire evolution:
  - Survey frames now well established and strengthened with successive surveys; exploratory surveys used in Ghana and Mozambique; ITRS used in some countries to identify enterprises (Mauritius, Mozambique).
  - Within frames, major enterprises accounting for the bulk of financing from abroad are identified for targeted high response.
  - Questionnaire approaches: three countries use BPM6 compliant Form 18 expanded for portfolio/other investment (The Gambia, Mauritius, Nigeria); three use Bank of Uganda–based questionnaires (Ghana, Kenya, Mozambique).
  - Many countries expanded questionnaires to meet other stakeholder needs due to survey costs and funding contributions.
- Data collection:
  - Most countries shifted from mail-outs to enumerator-based collection (The Gambia, Ghana, Kenya, Mozambique, Nigeria); Mauritius successfully used mail-outs though major enterprise follow-up required.
  - Electronic reporting is being explored selectively for industries with large enterprises or listed companies.
- Data review and databases:
  - Use of financial statements to validate returns and to estimate non-response is established.
  - Data review checks and the use of databases with data review tables are being implemented (Ghana, Kenya, Mauritius, Nigeria).
  - Ghana successfully established an Access-based database; several countries have IT working groups to develop in-house databases.
- Grossing up application status:
  - Grossing up techniques applied in Ghana and Mozambique; other countries expected to apply once survey frames and response adequacy are achieved.
- Publication and integration:
  - All countries produced preliminary IIP statements from first enterprise surveys; publication deferred until surveys produced reliable annual data.
  - Ghana is the only country to have published revised IIP and BOP financial account data fully integrating enterprise survey results; Mozambique nearly achieved this but faces reconciliation issues due to cash-based transactions and accrual-based positions.

*Source: TNM/13/02 Technical Notes and Manuals section on enterprise surveys, data collection, review, grossing up, portfolio investment, and Ghana experience.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/tnm/2014/_tnm1401.pdf_