OECD Benchmark Definition of Foreign Direct Investment (Fifth Edition)

Annex C. Examples of foreign direct investment data collecting and reporting methods

834. This annex discusses some examples of the foundations needed to compile foreign direct investment (FDI) data via a survey system. These include data sources, survey design, need for a business register, mail-out procedures, data capture, compilation and dissemination practices, legal aspects, timeliness and revision practices. In addition to FDI surveys, compilers rely on many other data sources to compile FDI statistics, including international transactions reporting systems (ITRS), administrative data, company financial statements, the financial press, and commercial databases, that are not covered in this annex. IMF (2014[1]) provides general guidance on conducting enterprise surveys as well as model surveys for FDI statistics, and IMF (2015[2]) provides guidance specific to conducting surveys to collect FDI position data.

Data sources

Questionnaires

835. Statistical questionnaires often constitute the core source of information and are used to obtain information on direct investment from businesses involved in direct investment activities. Depending on the frequency of release of FDI results, quarterly or annual questionnaires can be used. Questionnaires requesting information on the consolidation used by the reporting entities when reporting their operations are also a very useful tool to assess the scope of reported data. In collecting data using a survey, it is generally preferable to have the operation conducted on a statutory basis.

Administrative data

836. Data gathered from government bodies can be useful to complete the data collected by the questionnaires. Local investment review agencies can also provide a good source of financial data to complement mailing lists of companies. In addition to their use in developing a population frame for FDI surveys, administrative data can also be a valuable data source in their own right to compile FDI statistics. Relevant data from statistical offices and official business registers or information from banking and financial supervision authorities can complement surveys in, for example, the compilation of reinvested earnings. Centralised financial statement databases, and other sources containing mandatory annual financial statements (preferably those that comply with international accounting standards or are audited) can all be useful sources for the compilation of FDI statistics and may even provide enough information to lighten survey burdens.

Monitoring of the financial press and commercial databases

837. Monitoring financial information from the business community as published in the financial press or in press releases from the stock exchange supervisory authority is often very important both to highlight the occurrence of and to better assess important FDI transactions. On occasion, it can be necessary to estimate a FDI transaction based on published reports without information directly from the company. Commercial databases and catalogues showing corporate structures of companies (e.g., Who Owns Whom) can also constitute reliable information to populate a frame of enterprises likely to be involved in FDI. These sources may also provide information to validate data collection. For the case of listed equity in FDI, they may provide information on dividend distributions that could be useful in the compilation of reinvested earnings.

Bilateral sources

838. Reconciliation of data between partner countries can be conducted to validate and try to harmonise one another's statistics. This exercise can be done on a regular basis and is often very valuable in assessing the quality of direct investment data. Databases from international organisations, like the OECD Foreign Direct Investment Statistics database and the International Monetary Fund’s Coordinated Direct Investment Survey (CDIS), are important sources of bilateral data for reconciliation purposes. Data sharing initiatives, like Eurostat’s EuroGroups Register (EGR), can provide useful structural economic information on enterprises that are part of a FDI framework.

Survey design

839. Survey design expertise is essential to better assess the proper layout and wording of the questionnaires in relation to respondents. Linkages with accounting terms and financial statements of companies greatly facilitate the completion of questionnaires by respondents.

Respondents’ relations

840. Contact with respondents should be maintained on a regular basis, particularly for the large, complex enterprises. Given the complexity of FDI and the complexities of the structures and operations of many multinational enterprises, it is essential that such contact be maintained at as high a level as possible on both sides on an ongoing basis. In this context, company visits by the compiler are very beneficial to both sides, and a team devoted to maintaining effective working relationships with companies is important. Such contact can ensure respondents understand the information required for FDI statistics, which is especially important if there is a change in the financial or accounting management within the company. Some statistical offices have created large case units specifically to manage relationships with large reporters and to validate reported data covering a variety of statistical domains. In sourcing data and in efforts to ensure the quality of the FDI results, access to audited company accounts (including management accounts) is crucial and should be pursued vigorously.

Business register

841. A frame of enterprises is needed to incorporate and maintain the list of those included in the FDI survey(s). It has to be designed specifically to include the necessary information for FDI, in particular ownership information. Moreover, use of a comprehensive business register for other surveys or other statistical purposes that includes the necessary ownership information to conduct FDI surveys is beneficial because it supports data linking.

842. Here are some examples of procedures that can help to maintain a business register suited for FDI. When an enterprise is identified, via a newspaper article or any other source, the precise information about the name and the address of the enterprise should be loaded into the mailing list.

843. The characteristics and details of the enterprises should then be included in the business register and a determination should be made on the basis of the relevant information whether it should be surveyed for FDI purposes. The business register should also include historical information about the enterprise from the first time it was surveyed as well as any details of any changes (e.g., whether it changed name, address, activity or ownership or even ceased operations). Here are some of the basic components of a business register:

• Enterprise ID (unique identifier)

• Enterprise identification and contact information, such as enterprise name, contact person at the enterprise, address of the enterprise, phone number, and email address.

• Enterprise characteristics, such as type of enterprise (e.g., private corporation or public enterprise), industry and industrial code, and institutional sector.

• Variables useful for identifying the surveys to be sent, such as number of employees, turnover, and size of balance sheet.

• For FDI surveys, it is also useful to identify the economy of immediate direct investors in resident direct investment enterprises (DIEs) and the economy of non-resident DIEs owned by resident direct investors.

844. The business register should be maintained on an ongoing basis and reflect any updates necessary due to births, deaths or changes applied to an enterprise. Therefore, each time a change is made to an enterprise, it should be reflected in the frame and kept for future reference. The information gathered from surveys, administrative sources or from the financial press is used to build and maintain an up-to-date business register.

Mail out and follow-up procedures

845. A legal authority is often needed to ensure that the statistical information is collected: it is compulsory in many countries for the enterprises to complete and return the questionnaires being sent to them by statistical authorities. Enterprises that refuse or neglect to do so may be guilty of an offence and may be subject to legal enforcement procedures and penalties. If an enterprise does not return its questionnaire(s) within a certain time period, it is often appropriate to send a number of reminders indicating that their questionnaires are still outstanding. If the respondent requires more time to complete the questionnaire, the information is to be noted so that the respondent is not sent reminders. In some cases, it may be necessary for compilers to pursue legal enforcement where a survey is mandatory.

846. It can be useful to maintain information on communications with the companies, including on follow-up communication with those that have not responded as well as to clarify responses received. Examples of this kind of information include response status (e.g., mailed, received, or exempt), dates of follow-up for non-response, and dates of queries about their response (including date, subject matter, and outcome). This information could be maintained in a separate related database – a survey management system (i.e., linked to the main register via the enterprise identification number). Summary counts of questionnaires mailed, received and the percentage outstanding should be easily retrievable from the response log file.

Data capture and procedures

847. Systems can be put in place to automate the data capture of the questionnaires received (spreadsheets or web systems). This function can improve considerably the quality of data if thoroughly carried out. FDI data captured from questionnaires can then be reviewed by financial analysts. Special efforts should be made with respondents involving large transactions or with missing critical information.

848. To help the analysts in the validation and manipulation of the massive amount of data being captured, a computer system has to be developed. This system should include all the information on the status of the enterprise, on the data being captured and the other sources of information described above. The description of the following procedures provides examples of the main functions to process the information.

Validation of data

849. This procedure validates the information after the data capture process. An essential part in the validation of data is the edit function built into the system. This edit function should follow pre-programmed automatic procedures that can identify erroneous data entries and build an error message file to be corrected.

850. Another function should allow the analyst to browse and correct data when further research using information from outside of the system justifies it. In this context, audited company accounts data can be very helpful as can other related information supplied to the same compiling authority or to another public authority, industry body, etc.

851. It is also possible to include some validation checks in the data capture process to assist companies in reporting data correctly.

Imputation and estimation

852. When conducting a survey, it is normal for some enterprises not to have responded at the time that estimates need to be compiled. Values need to be imputed for these respondents. The imputation process can use information from previous responses from the enterprise, responses to other surveys from the enterprise, responses from similar enterprises, and information from the financial press and other data sources.

853. It is also common not to send questionnaires to all enterprises known to be involved in FDI relationships. The information collected from selected enterprises may be used to represent the entire population of resident direct investors and DIEs. This can involve the application of weights (i.e., grossing or scaling factors) to reported data. Alternatively, an under-coverage adjustment may be based on data from other sources for the non-selected enterprises.

Assessment of the quality of the data

854. The initial estimates of direct investment need to undergo several updates and have to be validated to try to minimise under-coverage in the resulting estimates. The response rate needs to be assessed regularly, especially for respondents involved in many and/or large FDI transactions.

Dissemination

855. After estimates referring to a certain period (for transactions) or point in time (for positions) are published, it is possible that further information will become available that would cause revisions to the original estimates. This information may result from improved data from respondents, improved compilation processes or other investigations into these data. Any compilation system will need to assess the time lag between the survey process and the data dissemination as well as the revision practices.

856. As data need to be reported at the enterprise level, the analyst has to be able to generate reports according to very specific analytical needs. For publication purposes, data need to be available by economy, by industry, and by institutional sector (e.g., non-financial corporations or government), resulting in a multitude of reports that will be needed for comprehensive analysis of FDI data. Direct investment data can be released in a very aggregated form (e.g., as part of the balance of payments) but also on a more disaggregated basis as requested by this Benchmark Definition. The systems need to provide flexibility to be able to tabulate any combination of economy, industry and characteristics of direct investment that can be requested.

857. Statistical legislation is generally necessary to protect confidentiality of the data provided. Such legislation normally prohibits the disclosure of information relating to any individual person, business or organisation. Very strict rules need to be applied to protect the direct, indirect or inadvertent disclosure of confidential information. These legal protections are very important as they not only reassure the respondents concerning the security of the information they provide, but they can also engender a positive response culture. A good level of confidence established with the enterprises regarding the protection of their data will have a positive impact on the response rate.

References

[2] IMF (2015), Coordinated Direct Investment Survey Guide, International Monetary Fund, Washington, DC., https://www.imf.org/external/np/sta/pdf/cdisguide.pdf.

[1] IMF (2014), BPM6 Compilation Guide, International Monetary Fund, Washington, DC., https://www.imf.org/external/pubs/ft/bop/2014/pdf/guide.pdf.