Data sources and limitations
Investment Map integrates and organises data from a number of databases. It sources from FDI databases developed by UNCTAD, COMTRADE, the UN trade database, tariff databases developed by ITC and company databases maintained by Dun & Bradstreet.
The definition of industries in Investment Map is based on the ISIC revision 3 nomenclature. Tariff and trade data are available for about 5,300 products (6-digit level of the Harmonised System). Information on foreign affiliates covers 1,000 types of businesses (based on the US SIC87 nomenclature) and data on trade and FDI are available for the last four years.
The users of Investment Map should be aware of the limitations that the presented data suffer from. These constrains should always be born in mind, especially when making cross-country comparisons.
A first source of data problem is due to the combination of data coming from various sources and reconciled in the ISIC rev.3 nomenclature. This can lead to inaccurate industry classification. Such problems concern primarily data on FDI and foreign affiliates.
FDI data cannot be always allocated accurately to a given industry and therefore it becomes necessary to assign these data to an additional industry (not defined in the ISIC nomenclature) such as "unspecified secondary" or "unspecified wholesale trade". This problem of reallocating data also affects the geographical classification of FDI data by countries or territories, and part of the data may fall down in a residual category such as "Unspecified European Union" or "Unspecified South America". The residual "unspecified" categories generally represent small investments, but for some countries they may include large investments. In this framework, cross-country FDI comparisons, classified by industry, should only be considered as indicative.
Information on foreign affiliates is based on an American nomenclature, US SIC87 (predecessor of NAICS), composed of a taxonomy of 1,000 business types. The conversion of such data to the ISIC nomenclature can only be tentative in several cases. As an example, the SIC code "7389", labelled "business services, nes" might correspond to up to three ISIC industries: "trade", "transport, storage and communications" and "other services". The data of companies having such activities are duplicated in Investment Map for the three industries. Consequently, when analysing the company data for any given country, the user should check the line on top of the table, displaying the total number of companies.
For merchandise trade data, while the correspondence from the 6-digit level of the HS to the ISIC industries is generally straightforward, there is still an hybrid group of products, having both a manufacturing and a service component, such as paintings, pictures or movies, which have been allocated to an additional industry, "mixed goods (trade data)".
Data on trade in services in Investment Map presently cover the three broad categories of commercial services, defined by the IMF: travel services, transportation services and other commercial services. These three categories do not match with the ISIC nomenclature and have been consequently added as new industries.
Other data limitations
In addition to the classification problems, there are other data problems, which are more specific to each type of data.
FDI data are collected from National Banks and enterprise surveys. FDI flows are difficult to capture accurately (like trade in services data) by their nature, i.e. due to the intangibility of the financial flows. However, unlike trade in goods, which capture many relatively small size transactions, FDI data are characterised by fewer large size transactions, making them more sensitive to the accurate recording of these large observations.
International reporting practices set by the IMF (5th edition of the IMF Balance of Payments Manual) and the OECD are not followed uniformly. For example, some countries deviate from the 10% threshold value of foreign ownership (share in the equity of the foreign affiliate). Moreover, the three components of FDI flows, equity capital, intra-company loans and reinvested earnings may not be all included for all countries. Some countries do not report short-term intra-company loans, some countries still do not report re-invested earnings and a minority of countries report reverse investment (investment of a foreign affiliate in its parent firm).
Bilateral discrepancies observed between FDI data reported by two countries might also be due to the differences in FDI transactions recording time periods.
For most countries covered in Investment Map, the definitions and sources of FDI data are available in the first pages of the World Investment Directory. This directory is very comprehensive and covers for example whether FDI data reported by the country is in accordance with the recommendations of the 5th edition of the IMF Balance of Payments Manual or whether short terms intra-company loans are covered.
There are also specific issues that affect FDI data. These include:
- The identification of the ultimate owner and target company for cross-border mergers and acquisitions
- Valuation problems of FDI
- FDI under the fully consolidated systems
- Round-tripped investments and transhipped investments
- The role of special-purpose entities (companies) located in tax heavens
These issues are described in details in the following document (www.unctad.org/sections/wcmu/docs/C2em18p23_en.pdf), which was prepared in an expert meeting on capacity building in the area of FDI data compilation and policy formulation in developing countries, organised by UNCTAD in Geneva in December 2005. All the presentations of this expert meeting are available on-line at: (www.unctad.org/Templates/Page.asp?intItemID=3640). See for example the presentation of the representative of Luxembourg, who describes the strong share of flows in transit (estimated at 95% of the country's flows) in his country. This phenomenon also overestimates the "genuine FDI" of other countries such as the Cayman Islands, Bermuda, Saint Kitts and Nevis, Antigua and Barbuda, Aruba or Northern Marianas. This double counting also results in a non-negligible overestimation of world FDI flows.
Data on foreign affiliates, maintained by D&B, are collected by national counterparts, and present a great deal of variation in the depth of coverage and accuracy. Some countries, like Brazil and the Czech Republic are extremely well covered; while others, e.g. Mongolia or Jordan, are barely analysed. In addition, some of the fields such as the sales or the number of employees are not systematically reported. Cross-country comparisons are therefore very delicate to interpret. These data should rather be used as a complement to FDI flows and stocks, in order to assess the weight of foreign-owned companies in the different industries of a country or to analyse the global strategy of a multinational corporation.
Merchandise trade statistics are comprehensive in terms of product (more than 5,300 products under the Harmonized System), geographical (around 220 countries and territories covering 95% of world trade) and time coverage (data under the Harmonized System are available for the last decade). Moreover, they are readily available at moderate costs, and therefore it is an attractive source for market research and the assessment of trade performance. Notwithstanding the attractiveness of merchandise trade statistics, users should remark their following weak points:
Trade data are never complete. Smuggling and non-reporting represent a serious problem in a number of countries. In addition, trade statistics, like any source of information, are not free of mistakes and omissions.
Most countries include imports for re-exports and re-exports in their trade statistics. A low-income country may show up as an exporter of airplanes simply because its national airline has sold second-hand planes.
The export value refers to the total or contract value. According to international conventions for reporting trade statistics, the export value refers to the total or contract value, which may of course, be very different from local value-added. For many processing activities the local value added remains below 20% of the export value.
Only merchandise trade is covered. Detailed trade statistics are available only for merchandise trade and not for services, although the latter may account for a sizeable share of national exports.
Different products are categorized differently. Even at the lowest level of disaggregation, product groups in the trade nomenclatures do not necessarily reflect trade names and often contain a wide range of different products. Moreover, the product nomenclature is sometimes misleading. The labels of aggregated product groups are often very general and provide at times only limited guidance on the leading items within the group of products concerned.
Exchange rate fluctuations are not always recorded. Exchange rate fluctuations are not always properly recorded in international trade statistics. Values are normally aggregated over the period of one year in local currency and converted into US dollars.
Mirror statistics are sometimes used. For countries that do not report trade data to the United Nations, ITC uses partner country data, an approach referred to as mirror statistics. Mirror statistics are a second-best solution being better than having no data at all and allow the coverage of the over 50 countries that do not report consistently national trade statistics to COMTRADE. At the same time, they have a number of shortcomings when compared to the first-best solution of nationally reported data. First and foremost, they do not cover trade with other non-reporting countries. As a result, mirror statistics hardly cover South-South trade and would not be a suitable source for an assessment of intra-African trade. Second, there is the problem of transshipments, which may hide the actual source of supply. Third, mirror statistics invert the reporting standards by valuing exports in c.i.f. terms (i.e. including transport cost and insurance) and imports in f.o.b. terms (excluding these items).
In an effort to make some of these discrepancies more transparent, the option of viewing mirror statistics is available in Trade Map: Bilateral trade statistics, custom tariff by country and product at tariff line level, HS.