Asymmetries in the Coordinated Direct Investment Survey: What Lies Behind?
Author: Emma Angulo1, Alicia Hierro2, and Carlos Sánchez-Muñoz
  • 1 https://isni.org/isni/0000000404811396, International Monetary Fund
  • | 2 https://isni.org/isni/0000000404811396, International Monetary Fund

Contributor Notes

Author’s E-Mail Address: eangulo@imf.org; ahierro@imf.org
Publication Date:
21 Nov 2017
ISBN:
9781484330036
Language:
English
Keywords:
WP; investment position; enterprise; economy; CDIS
Download PDF (1.1 MB)

This paper analyzes asymmetries in direct investment positions reported in the Coordinated Direct Investment Survey (CDIS) following a top down approach. First, it examines asymmetries at global level; second, it examines asymmetries between CDIS reported and derived data for individual economies; and third, the paper analyzes data at bilateral economy level. Then, the paper explores seven main reasons for asymmetries, including those arising even when economies follow international standards. Finally, the paper includes a section on addressing bilateral asymmetries and concludes with specific planned actions to reduce asymmetries, including initiatives led by international organizations.

Abstract

This paper analyzes asymmetries in direct investment positions reported in the Coordinated Direct Investment Survey (CDIS) following a top down approach. First, it examines asymmetries at global level; second, it examines asymmetries between CDIS reported and derived data for individual economies; and third, the paper analyzes data at bilateral economy level. Then, the paper explores seven main reasons for asymmetries, including those arising even when economies follow international standards. Finally, the paper includes a section on addressing bilateral asymmetries and concludes with specific planned actions to reduce asymmetries, including initiatives led by international organizations.

I. Introduction

The CDIS is a worldwide statistical data collection effort led by the IMF designed to improve the availability and quality of data on direct investment, both at overall level in the international investment position (IIP) and by immediate counterpart economy. The CDIS database presents detailed annual data on inward direct investment positions starting from data since end-2009 cross-classified by economy of immediate investor (direct investment into the reporting economy), and data on outward direct investment positions (direct investment abroad by the reporting economy) cross-classified by economy of immediate investment.

The structure of the CDIS database allows easy comparison of data reported by one economy and mirror data (obtained from all other counterpart reporting economies). Mirror data may be useful in highlighting data gaps or errors, and therefore where follow up efforts may prove beneficial.

The IMF is making continuous efforts to broaden participation in the CDIS as well as to improve the quality of the reported data. As part of these efforts, in May 2016, the IMF’s Statistics Department (STA) conducted a project to raise awareness on (and to the extent feasible try to address) large bilateral asymmetries based on the 2015 CDIS release for end-2014 CDIS data, contacting a total of 47 CDIS participants.

STA conducted a similar project in 2013 based on the 2012 CDIS release with a more reduced scope (28 CDIS participants).2 Both exercises provided insights into the reasons for asymmetries. They also helped some economies detect errors and consider taking actions to further improve the quality of their direct investment data. Several economies have conducted recent studies triggered by these projects under the auspices of the IMF Committee on Balance of Payments Statistics (Committee).3 STA continues its efforts in the field of direct investment asymmetries in coordination with other international organizations.

This paper analyzes asymmetries in direct investment positions reported in CDIS following a top down approach. In the following three sections, it (i) examines asymmetries at global level; (ii) examines asymmetries between CDIS reported and derived data for individual economies; and (iii) analyzes data at bilateral economy level, by comparing data as reported by individual economies vis-a-vis their individual counterparts and considering main symmetries and main asymmetries. Lessons learned on the reasons for asymmetries and proposed ways to reduce bilateral asymmetries are presented in the last two sections of the paper.

II. CDIS Results for the Total World

This section examines asymmetries at global level; comparing total outward and total inward direct investment reported by all CDIS reporting economies and analyzing global asymmetries by instrument. The results are then compared with the global asymmetries obtained from direct investment assets vs. liabilities reported in the international investment position (IIP).

Direct investment positions are concentrated in equity, representing around 80 percent of the total direct investment, which raises awareness of the importance of applying consistent valuation methods for unlisted equity to reduce the bilateral asymmetries between counterpart economies. The global asymmetries in net equity are always positive, that is, total outward equity positions are always larger than total inward equity positions, and on the other hand, the asymmetries in net debt instruments are always negative, that is, total inward debt positions are always larger than total outward debt positions for the seven years covered in the CDIS. Assuming that inward direct investments positions are more accurate than outward positions, for both equity and debt instruments, this section concludes that equity investment abroad tends to be overestimated while debt investment positions abroad tends to be underestimated.

A. Total Outward Direct Investment vs. Total Inward Direct Investment

In principle, at a global level, total outward direct investment positions should be equal to total inward direct investment positions reported by all economies. However, in the CDIS database, there are differences between outward and inward reported totals because (i) not all economies participate in the CDIS, (ii) not all CDIS reporting economies provide outward data, and (iii) there are some bilateral asymmetries between reporting economies.

Figure 1 shows the differences between total outward and total inward direct investment positions reported by the CDIS participating economies. In most cases, the number of reporting economies, indicated in Figure 2, increases over the years and is always larger for inward than for outward (outward data are more difficult to compile and not relevant for some small economies). The global difference is particularly small in 2013, when the maximum number of participants is reached (76 for outward and 110 for inward) and the data could be considered “final” after two years of collecting revised data.

Figure 1.
Figure 1.

Global Asymmetries in CDIS

(US Dollars, billions)

Citation: IMF Working Papers 2017, 261; 10.5089/9781484330036.001.A001

Figure 2.
Figure 2.

Number of CDIS Reporting Economies

Citation: IMF Working Papers 2017, 261; 10.5089/9781484330036.001.A001

The sign of the differences is not homogenous; therefore, it cannot be deduced that there are missing values in one or the other component (inward or outward). These global asymmetries can be analyzed in more detail, assessing the asymmetries produced in the main components of the direct investment positions; equity and debt instruments.

Figure 3 presents the global asymmetries by instrument. The figure shows that, for all periods, the weight of the net equity component in the total direct investment is much higher than the weight of the net debt instruments (for outward, the weight of equity ranges from 84 to 89 percent, and for outward it ranges from 77 to 81 percent). This results highlight the importance of applying a consistent valuation of unlisted equity. Compilers should apply the same valuation methods to reduce the bilateral asymmetries between counterpart economies.

Figure 3.
Figure 3.

Outward vs Inward by Instrument in the CDIS

(US Dollars, billions)

Citation: IMF Working Papers 2017, 261; 10.5089/9781484330036.001.A001

Source IMF’s CDIS, 2016 release

Also, it shows that the asymmetries in net equity are always positive, therefore, the total amounts reported under outward-equity are always larger than the total amounts reported under inward-equity for the seven years covered by the CDIS. As the number of inward reporters is larger than the number of outward reporters, these asymmetries cannot be explained by missing reporters. The differences are likely due to the different valuation methods applied by compilers. The investments abroad are higher valued than the investment in the reporting economy where the information is easier to measure and tends to be more accurate. Therefore, in principle, investment in equity abroad (outward equity) tends to be overestimated.

On the other hand, Figure 3 also shows that the global asymmetries in net debt instrument positions are always negative, that is, the total amounts reported under outward-debt instruments are smaller than the total amounts reported under inward-debt instruments for the seven years covered by the CDIS. As the inward-debt instruments positions are more accurate, mainly due to the higher number of inward reporters and the better coverage of inward debt instruments. It can be assumed that investment in debt instruments abroad (outward debt instruments) tends to be underestimated.

Finally, Figure 3 also shows that the differences at instrument level are larger than the total asymmetries because positive differences in equity are compensated with negative differences in debt.

B. Total Direct Investment Assets vs. Total Direct Investment Liabilities in the IIP

The values reported in the CDIS are different than the values reported in the direct investment component of the IIP due to three main reasons: (i) the number of economies reporting IIP data to STA is larger than the number of economies reporting CDIS data; (ii) direct investment data are presented on a net basis in the CDIS, following the directional principle, and on a gross basis in the IIP, following the assets/liabilities presentation; and (iii) the possibility of applying six different valuation methods for unlisted equity in the IIP, as a proxy for market value (para. 7.16 of BPM6), while, in the CDIS, unlisted equity should be valued at own funds at book value.

The number of economies compiling IIP data (see Figure 5) is larger than the number of economies participating in the CDIS because the IIP data are compiled on an aggregated basis and CDIS data must be provided broken down by counterpart economy, which is more difficult to compile.

Figure 4.
Figure 4.

Global Asymmetries in the IIP

(US Dollars, billions)

Citation: IMF Working Papers 2017, 261; 10.5089/9781484330036.001.A001

Source IMF’s Balance of Payments and IIP database
Figure 5.
Figure 5.

Number of IIP Reporting Economies

Citation: IMF Working Papers 2017, 261; 10.5089/9781484330036.001.A001

The results shown in Figures 4 and 6 for the IIP are similar to those described in Figures 1 and 3 for the CDIS. The weight of the equity component in the total direct investment is also higher than the weight of the debt instruments, for both assets and liabilities, representing approximately 72 percent of the assets and 65 percent of the liabilities. Also, the asymmetries for equity are always positive and the asymmetries for debt instruments are always negative. Therefore, the previous findings are verified with these results.

Figure 6.
Figure 6.

DI Assets vs DI Liabilities by Instrument in the IIP

(US Dollars, billions)

Citation: IMF Working Papers 2017, 261; 10.5089/9781484330036.001.A001

Source IMF’s Balance of Payments and IIP database

III. CDIS Reported vs CDIS Derived Data for Individual Economies

While the previous section examined asymmetries for the total world, this section focuses on asymmetries between reported and derived data for individual economies and discusses the usefulness and limitations of derived measures of direct investment (see more details in Annex I).

To this end, this section identifies the ten economies with the largest asymmetries between reported and derived data for inward and outward direct investment, respectively. The section further analyzes selected economies and identifies the reasons for the asymmetries.

Asymmetries between reported and derived data can be spread among several counterparts or concentrated in a few counterparts. Asymmetries can be explained due to missing data by the counterparts or to bilateral asymmetries. Some counterpart economies do not provide data either because they do not participate in the CDIS, or because the data provided are considered confidential, or fall below a reporting threshold. Mirror data are useful to cross check and verify countries’ own estimates, to highlight data gaps or errors, and to identify areas for improvement. If reported data by a given economy are lower than derived data, it can suggest under coverage; therefore, asymmetries at bilateral level should be analyzed.

A. Top Ten Differences Between Reported and Derived Data

Tables 1 and 2 below show the top ten differences between CDIS reported and derived data for inward and outward direct investment, respectively. Inward data reported by one given economy are compared with outward data reported by its reporting counterparts (derived inward); and outward data reported by the given economy are compared with inward data reported by its reporting counterparts (derived outward).

Table 1.

Top Ten Differences Inward/Inward Derived

(US Dollars, millions)

article image
Source IMF’s CDIS, 2016 release
Table 2.

Top Ten Differences Outward/Outward Derived

(US Dollars, millions)

article image
Source IMF’s CDIS, 2016 release

As shown in Table 1, while China, P.R. Mainland reported inward direct investment amounting to US$2,580 billion at end-2015, all its CDIS counterparts reported outward investment in China P.R.: Mainland for only US$1,149 billion (representing only 44 percent of the inward reported). These mirror data reported by all its counterparts can then be used as a measure of the derived inward.

B. Top Differences in Inward Reported and Derived: Analysis for Selected Economies

In the case of China, P.R.: Mainland, 44 percent of the difference in inward is explained by its counterpart China, P.R. Hong Kong, which reports a much lower amount of investment in China, P.R. Mainland (outward investment reported by Hong Kong in Mainland is US$608 billion, versus US$1,239 billion reported by Mainland as inward investment received from Hong Kong). In addition, 41 percent of the difference is explained because some of the main counterparts do not report outward data (for example, British Virgin Islands accounts for US$328 billion followed by Singapore with US$111 billion, representing respectively 23 and 8 percent of the total difference). There are also some smaller differences acting in the opposite direction (meaning that inward reported by Mainland is lower than mirror data reported by counterpart); South Africa is the most significant case, but represents only 4 percent of the total difference in absolute terms.

For the Netherlands, differences in inward are spread among many countries, the most relevant ones correspond to United Kingdom (US$189 billion), Luxembourg (US$187billion), and Switzerland (US$134 billion) representing, respectively 14, 14, and 10 percent of the total difference. In addition, 29 percent of the difference is explained because some of its main counterparts do not report outward data (Bermuda accounts for US$291 billion followed by Cayman Islands with US$49 billion). There are also some smaller differences acting in the opposite direction (meaning that inward reported by the Netherlands is lower than mirror data reported by its counterpart); United States is the most significant case, but represents only 5 percent of the total difference in absolute terms.

In the case of China, P.R. Hong Kong, however, around 95 percent of the difference in inward data is explained because its main counterpart economies do not report outward data (British Virgin Islands and China, P.R.: Mainland account for US$552 and US$349 billion respectively, followed by Cayman Islands and Bermuda with US$85 and US$74 billion).

For the United Kingdom, the difference has opposite sign that in the previous cases, meaning that reported inward data is much lower that the derived inward data. Main difference is with Luxembourg (-US$593 billion), representing 50 percent of the total asymmetry, followed by the Netherlands and the United States, with differences of -US$263 and -US$160 billion, representing respectively 22 and 13 percent of the total difference. There are also some differences acting in the opposite direction (meaning that inward reported is higher than inward derived); this is mostly due to some counterparts not reporting outward data (mainly Bermuda, Cayman Islands, and British Virgin Islands with differences representing respectively 6, 3, and 2 percent of the total difference in absolute terms).

C. Top Differences in Outward Reported and Derived: Analysis for Selected Economies

For the Netherlands, 74 percent of the difference in outward data is explained due to different figures reported by its counterparts and 23 percent is explained by missing reporters. Differences are spread among many countries and most relevant ones correspond to United Kingdom (US$263 billion), the United States (US$232 billion), and Luxembourg (US$188 billion) representing respectively 14, 13, and 10 percent of the total difference. In addition, 23 percent of the difference in outward data is explained because main counterparts do not report inward data (Bermuda accounts for US$165 billion followed by Puerto Rico with US$109 billion, representing respectively 9 and 6 percent of the difference).

In the case of Luxembourg, 49 percent of the difference in outward data is explained because its counterpart the United States reports a much lower figure for the corresponding mirror data (inward reported by the United States as received from Luxemburg is US$328 billion, versus US$1,011 billion reported by Luxembourg as outward investment in the United States). A similar situation applies for United Kingdom, accounting for 43 of the difference (inward reported by United Kingdom as received from Luxemburg is US$105 billion, versus US$698 billion reported by Luxembourg as invested in United Kingdom). Only 12 percent of the difference in outward data is explained because main counterparts do not report inward data, Gibraltar and Bermuda being most significant ones (representing each around 5 percent of the difference). In some other cases, there are differences that have opposite sign (this is, its counterparts are reporting higher figures in mirror data, for example the Netherlands and Germany, accounting for 13 and 6 percent of the total difference in absolute terms).

In the case of United States, a significant part of the difference in outward is explained because main counterparts do not report inward data; Bermuda and Cayman Islands represent 27 and 26 of the difference respectively. In addition, the difference is also explained due to lower figures reported by its counterparts. Differences are spread among many countries; the most relevant one corresponds to United Kingdom (US$160 billion), representing 16 percent of the difference, followed by Canada and the Netherlands with differences of 7 percent each. In some other cases, there are differences that have opposite sign (the counterparts are reporting higher figures in mirror data, than the outward reported by the United States); for example, with Luxembourg there is a difference of -US$233 billion, and with Mexico there is a difference of -US$130 billion, representing respectively 24 and 13 percent of the total difference in absolute terms.

For the United Kingdom, the difference has opposite sign that in the previous cases, meaning that reported outward data is much lower that the derived outward data. Main difference is with Luxembourg (-US$347 billion), representing 52 percent of the total asymmetry, followed by the Netherlands and the United States, with differences of -US$189 and -US$130 billion, representing respectively 28 and 20 percent of the total difference. These findings are like those for the inward. There are also some differences acting in the opposite direction (meaning that outward reported by United Kingdom is higher than outward derived); this is mostly due to counterparts that do not report inward data (mainly Jersey, Cayman Islands, and British Virgin Islands, with differences representing respectively 4, 3, and 2 percent of the total difference in absolute terms) but also to countries that report mirror data with lower amounts (for example Belgium, with a difference representing 3 percent of the total difference in absolute terms).

Mirror data are useful to for cross checking purposes; however, it should be used with caution, as some counterpart economies may not provide data. The analysis should be done beyond the comparison between reported and derived data to better understand and explain asymmetries; asymmetries at bilateral level should be analyzed.

IV. CDIS Results by Compiling Economy

The next step in the top down approach to analyze asymmetries is comparing data bilaterally for individual economies and their individual counterparts. This section shows bilateral comparisons for 20 smallest and 20 largest asymmetries.

A. Smallest Bilateral Asymmetries

Table 3 presents inward data reported by 23 economies with differences not exceeding three percent of the total when compared with outward data reported by counterpart economies. These small differences reflect that: (i) economies undertake bilateral reconciliations or comparisons with their main counterpart economies to improve the accuracy of their estimates; (ii) economies use the mirror data provided by their counterpart economies to adjust their own estimates; (iii) data sources used by the two economies involved to collect direct investment positions are accurate and the coverage is adequate; and/or (iv) compilers follow the same methodological standards.

Table 3.

Twenty Smallest Bilateral Asymmetries as of end-2015

(US Dollars, millions)

article image
Source IMF’s CDIS, 2016 release

B. Largest Bilateral Asymmetries

The CDIS exercise has proven to be of utmost importance in identifying and creating awareness of large asymmetries between data provided by some economies versus data provided by their counterpart economies.

Table 4 shows the top twenty bilateral asymmetries as of end-2015. The table comprises eleven economies of which seven (Canada, China, Hong Kong, Luxembourg, Netherlands, Switzerland, the United Kingdom, and the United States) are also included in Table 3 (smallest bilateral asymmetries). Section V discusses the main reasons for asymmetries, based on numerous communications with CDIS reporting economies, existing CDIS metadata, and a survey conducted by STA in May 2016, which had a large response rate.

Table 4.

Twenty Largest Bilateral Asymmetries as of end-2015

(US Dollars, millions)

article image
Source IMF’s CDIS, 2016 release

V. Main Reasons for Bilateral Asymmetries

The root for asymmetries is often at national level due to inadequate or partial data sources, lack of information, and noncompliance with recommended guidelines. Bilateral asymmetries are mainly due to limitations faced by economies in their compilation process or to the use of methodologies deviating from the standards. Nevertheless, in some cases (mainly related to the case of fellows) asymmetries may arise even when economies follow current international methodological standards.

This section focuses and provides example of seven main reasons for CDIS bilateral asymmetries: (a) way of recording positions between fellow enterprises when the ultimate controlling parent (UCP) is nonresident; (b) different coverage and lack or partial coverage; (c) different geographic allocation arising from the use of the immediate vs ultimate counterpart economy; (d) unallocated or confidential data; (e) methodological differences in the valuation for listed and unlisted equity and debt instrument or the identification and exclusion of debt between affiliated financial corporations; (f) use of different data sources and estimation techniques; and (g) criteria for identifying entities in a direct investment relationship.

A. Asymmetries Arising from the Recording of Positions Between Fellow Enterprises when the Ultimate Controlling Parent (UCP) is Nonresident

CDIS bilateral asymmetries may result from the treatment of direct investment positions between fellow enterprises (see Annex II for definitions). These asymmetries arise when comparing outward direct investment reported by the economy of one fellow enterprise with inward direct investment positions reported by the economy of the second fellow enterprise, when the UCP is nonresident for both economies. However, these asymmetries are not real because they can be eliminated when comparing the total direct investment positions (outward minus inward) between the two economies hosting the fellow enterprises.

This asymmetry only arises when the UCP of the two fellow enterprises is resident in a third economy, different than the economies of the two fellow enterprises. In this case, the application of the extended directional principle, to record positions between fellows, gives rise to inward positions in one economy and negative inward positions in its counterpart economy (rather than the corresponding outward positions). This circumstance causes asymmetries of the same size and opposite sign when comparing both economies’ inward and outward data.

Example

Fellow enterprise A is resident in economy A and its fellow enterprise B is resident in economy B, and both economies apply the extended directional principle to record data on fellows:

If the UCP is resident in a third economy (e.g., economy C) different than the economies of the two fellow enterprises:

  • - Economy A will record all assets and liabilities between the fellows A and B in inward (because the UCP is nonresident) and

  • - Economy B will record all assets and liabilities between the fellows A and B in inward (because the UCP is nonresident), and with opposite sign.

In this case, there would be a bilateral asymmetry between inward data reported by one economy and outward data reported by the counterpart economy, because both economies will record all assets and liabilities between fellow enterprises under inward because the UCP is nonresident for both fellows.

To eliminate these asymmetries caused by the recording of positions between fellow enterprises when the UCP is nonresident for both fellow enterprises, total net direct investment positions (outward minus inward) reported by one economy should be compared with total net direct investment positions (inward minus outward) reported by its counterpart economy:

Some of the largest bilateral asymmetries included in Table 4 are reduced when total net direct investment positions are compared and therefore, the effect of this asymmetry is eliminated.

Table 5-A includes two examples on large bilateral asymmetries that are reduced when netting the differences applying the formula described above. The asymmetry between outward data reported by Luxemburg in Netherlands and the corresponding inward reported by the Netherlands is very large ($187 billion). A similar asymmetry is obtained when comparing outward reported by the Netherlands in Luxemburg with the corresponding inward reported by Luxemburg ($188 billion). However, when comparing total net direct investment reported by these two economies, the asymmetry is reduced to $1.2 billion. The same applies to the asymmetries between Brazil and the Netherlands which show a decrease from around $85 billion to $3.5 billion.

Table 5-A.

Bilateral Asymmetries from Total Direct Investment as of end-2015

(US Dollars, millions)

article image
Source IMF’s CDIS, 2016 release

The same information could be presented as follows:

Table 5-B.

Bilateral Asymmetries in Total Direct Investment as of end-2015

(US Dollars, millions)

article image
Source IMF’s CDIS, 2016 release

In order to compare bilateral data using the total direct investment positions, countries should report both inward and outward positions, but some CDIS reporting economies do not provide outward data.

Annex 3 includes additional examples and analyzes asymmetries that could arise in one of the components of direct investment (direct investment between fellow enterprises), when one economy collects consolidated data from local enterprise groups (LEGs).

B. Different Coverage, Lack or Partial Coverage

Different coverage, lack of coverage, or partial coverage for special purpose entities (SPEs), fellow enterprises, nonfinancial private sector, and real estate may create large asymmetries. The capacity to compile data from these institutional units greatly varies from one economy to another, including the systems used to collect them which could imply different coverage.

Special Purpose Entities (SPEs)

Direct investment positions involving SPEs should be included in the CDIS (see Annex IV). The different coverage levels for SPEs by CDIS participants have significant impact on CDIS asymmetries. The CDIS metadata questionnaire reveals that SPEs are not evenly covered by all participating economies. Some CDIS reporters fully cover SPEs (such as Netherlands), while others only collect partial data (such as Luxembourg and Mauritius), and for other economies SPEs are not covered (such as Seychelles and Curacao). For some countries, SPEs phenomenon is negligible or not relevant (e.g., Italy and South Africa).

In addition, some economies include SPEs’ data in their CDIS, but are not able to allocate SPEs investments to any specific country or consider these investments confidential. In these cases, data related to SPEs are reported under “Not Specified (including Confidential)”, with no counterpart economy detail. This could give rise to significant asymmetries at bilateral economy level, but would have no impact at aggregate level.

Resident SPEs in Direct Investment Data

Asymmetries resulting from different coverage levels for SPEs can vary by economy, depending on the significance SPEs have over their total direct investment. Currently, the CDIS does not separately identify data for SPEs; however, OECD requests direct investment data to be reported separately for resident SPEs.

Table 6 shows 2015 data extracted from OECD database with the identification of resident SPEs data for inward and outward direct investment positions. For Luxembourg, the share of direct investment positions through SPEs is nearly the full amount of direct investment positions (over 90 percent) for both inward and outward. For the Netherlands, the share of SPEs is 82 percent in inward and 76 percent in outward. For Hungary, the share of SPEs is respectively 57 and 77 percent. For other economies, the SPEs contribution is nil (for example for United States, Italy, and Japan).

Table 6.

Total Direct Investment Positions with Split by Resident SPEs – OECD Data

article image
Data extracted on 03 Oct 2017 16:17 UTC (GMT) from OECD.Stat(Dataset: FDI positions, main aggregates BMD4)Note zero values (resident SPEs not present or not significant) were excluded.

Nonresident SPEs in Direct Investment Data

While some countries are main destinations of SPEs (such as Netherlands, which deals mainly with pass-through funds), others are characterized by outward investment in SPEs. This is the case of Russia; it is not perceived as a transit economy for international cross-border financial flows, and there are no SPEs in Russia, however, Russian enterprises traditionally use SPEs registered abroad in the conduct of their economic activities for buyback operations, holding companies, and conduit companies. Most popular jurisdictions for establishing holding companies for investing in Russia include the Netherlands as well as Bahamas, Bermuda, and British Virgin Islands. Russian companies also choose off-shore countries for registration of such SPEs aimed at optimization of tax payments (IMF, BOPCOM 16/21).

Lack of Participation in CDIS of Economies Hosting SPEs

Many economies hosting SPEs do not participate in the CDIS. As explained in Annex I, derived data based on mirror information are a very useful source of information for economies that do not compile direct investment data, and can give insights of their total inward and outward direct investment positions with counterpart economy breakdown.

The table below shows the top 15 economies that did not report 2015 CDIS data, ranked based on their derived inward and outward data as reported by their counterpart CDIS reporting economies. For 2015, inward and outward derived data for non-reporters represent 10 and 12 percent of the total reported inward and outward positions respectively. There are many economies that do not report CDIS; however, investments are concentrated in few economies. The top 15 non-reporters represent 89 and 94 percent of the total non-reported data for inward and outward respectively. An interesting finding is that most significant non CDIS reporters are economies that host SPEs, with British Virgin Islands, Bermuda, and Cayman Islands heading the ranking.

There are challenges in data collection for SPEs, including lack of statistical capacity, unavailability of information due to the confidential nature of these entities, and unwillingness to invest in compilation due to their limited economic impact on the domestic economies (IMF, BOPCOM 16/18).

Table 7.

Top 15 Economies that Did Not Report 2015 Inward/Outward CDIS (ranking based on derived data)

(US Dollars, millions)

article image
Source IMF’s CDIS, 2016 release

Fellow Enterprises

Equity and debt instrument positions between fellow enterprises are to be classified as direct investment and therefore included in the CDIS data.4 However, some economies do not cover these positions under direct investment, mostly because they are not able to identify them; they would instead classify these positions under portfolio or other investment, causing asymmetries with counterpart economies. Several economies are updating their surveys to collect data on investment with fellow enterprises, in their efforts to transitioning to the sixth edition of the Balance of Payments and International Investment Position Manual (BPM6).

About half of the 107 economies reporting 2015 inward data separately identify investment between fellow enterprises. For the outward CDIS, 32 out of 73 economies separately identify investments for fellows. The fact that economies do not separately identify investments with fellows does not necessarily mean that they do not cover them in the CDIS. Countries that collect direct investment data using the local enterprise group as reporting unit cannot separately identify data between fellow enterprises (e.g., the United States). The importance of investments between fellows varies by economy, and therefore its potential impact on asymmetries. The importance could be measured based on separate fellow enterprise data reported in CDIS. Inward positions with fellow enterprises represent 40 percent of total inward direct investment in Belgium while it only represents 0.6 percent in Switzerland. Annex V shows the positions between fellow enterprises relative to the total inward and outward investment for CDIS participants that separately identify fellows.

Nonfinancial Private Corporations

Low participation in the direct investment surveys of the nonfinancial private corporations, which in many economies comprise the bulk of target respondents, is a key challenge for compilers. Gaps in the coverage of investments for the nonfinancial sector can result in bilateral asymmetries if the level of coverage of counterpart economies is different.

The collection of direct investment data from the nonfinancial private sector could be based on a sample or a census. Based on the responses to the CDIS metadata questionnaire, most of the economies collect data from a sample of the target population for the nonfinancial corporations with direct investment positions. Data are collected from a census of enterprises by 33 percent of the reporting economies for inward and 20 percent for outward. The response rate to the surveys could also vary and produce bilateral asymmetries.

The absence of a legal mandate to collect data from nonfinancial corporations for statistical purposes continues to be the main reason for low response rate to surveys (see Philippines’ Experience (BOPCOM 16/27)). Low response rate may result because survey participation is voluntary. However, legal mandate itself may not suffice to significantly improve response rate and should be combined with other measures. Online survey questionnaires that are user friendly, consultative meetings with industry associations, and company visits to nonresponding corporations are effective in encouraging reporting.

Real Estate

Investments in real estate by nonresidents could be significant for some recipient/investor countries, however, they may not be covered. This can cause asymmetries with counterpart investing economies that do cover such investments. According to the responses to the CDIS metadata, around 40 percent of CDIS reporting economies include data on real estate in their direct investment. The level of coverage could vary.

Most of the positions on real estate involve the household sector which is the most difficult to cover because household surveys are costly and cannot be conducted with the required periodicity. Compilers need to collect these data mainly through public registers. In some countries, real estate data are covered, but the counterpart economy allocation may not be accurate or it is not provided (e.g., Austria). In Russia, direct investments of individuals in real estate are included in CDIS. The main data source is banks’ reports on personal transfers for real estate purchases (international transaction reporting system). However, counterpart economy breakdown is not always correct for this kind of data, as often the receiving country is not the destination, but a place of transit. For example, the main destination country for real estate purchase transfers is Switzerland from where the funds are further transferred to the countries in which the real estate is ultimately purchased (Central Bank of the Russian Federation BOPCOM 15/10.b)

Figure 7.
Figure 7.

Number of economies including real estate data in the GDIS

Citation: IMF Working Papers 2017, 261; 10.5089/9781484330036.001.A001

Source IMF’s CDIS, 2016 release

C. Different Geographic Allocation: Immediate vs. Ultimate Investing Economy

CDIS guidelines establish that direct investment data be reported by immediate (first) counterpart economy for both inward and outward direct investment positions. However, some countries rather allocate inward direct investment positions to the ultimate investing economy (the economy of the ultimate investor in the chain of ownership). It has also been observed that if the first enterprise is a nonresident SPE, few countries determine the counterpart economy looking through the nonresident SPE up to the next enterprise in the ownership chain. In this case, differences in the geographic allocation of direct investment positions between counterpart economies could produce bilateral asymmetries at country level, but not at global level. Inward direct investment positions by ultimate investing economy are also useful for analysis and some economies compile this additional breakdown.

The ultimate investor, for this purpose, is the enterprise that has control over the investment decision to have a direct investment position in the direct investment enterprise. The country in which the ultimate investor is resident is the ultimate investing country for the investment in the direct investment enterprise. It is possible that the ultimate investor is a resident of the same economy as the direct investment enterprise (Paragraphs 610 and 611 of BD4).

D. Unallocated and Confidential Data

When economies are unable to identify the counterpart country or this information is confidential, data are reported under the CDIS component: “Not specified (including confidential)”. The reporting of confidential or unallocated data by counterpart economy does not affect global asymmetries, but prevents from conducting accurate bilateral comparisons and may ultimately affect the analytical value of the data if the weight of the not specified data is relevant.

Figure 8 shows the weights of “Not Specified (including confidential)” over total direct investment along the periods covered by the CDIS. The values are quite steady over the years, except for the increase shown in 2012 for inward and in 2012/2013 in outward positions. At global level, data reported as not allocated or confidential represent around 6 percent for inward and 7 percent for outward and have decreased in the last two years. For some economies, there are relevant values under “Not specified” compared with their total direct investment.

Figure 8.
Figure 8.

Not Specified (including confidential) over Total DI in the CDIS

(US Dollars, billions)

Citation: IMF Working Papers 2017, 261; 10.5089/9781484330036.001.A001

Source IMF’s CDIS, 2016 release

Annex VI shows the top economies reporting unallocated or confidential data in percentage to their total inward or outward direct investment positions and representing more than 15 percent. For inward, there are 16 economies with a percentage larger than 15 percent and for outward there are 10 economies. The large amounts of not specified data provided by these economies hamper bilateral comparisons with their counterpart economies. However, it can be concluded that for most other CDIS reporting economies, the amounts reported as not specified are not significant.

When datasets provide more granular information with additional breakdowns and details, there is a risk of increasing the values reported as confidential; this is, the more granular the data, the higher the risk of confidentiality. Confidentiality issues can play a role for specific counterpart economies, investments or statistical units. For example, in some economies SPEs are covered in CDIS, but due to their limited number there are confidentiality restrictions and their data are not allocated to a specific counterpart (reported under “Not specified (including confidential)”). 5

E. Methodological Issues

Valuation Methods

As direct investment is mainly concentrated in equity, representing around 80 percent of the total direct investment, the use of different valuation methods, mainly for unlisted equity, could produce large bilateral asymmetries between counterpart countries. The exchange of data and metadata between compilers could help to understand and address some of the main bilateral asymmetries.

Valuation Methods for Direct Investment Equity in the CDIS

This section analyzes the bilateral asymmetries that could be produced by applying different valuation methods for estimating direct investment positions on equity. International standards set that all financial positions should be valued at market value.

Positions of financial assets and liabilities should, in general, be valued as if they were acquired in market transactions on the balance sheet reporting date (BPM6, paragraph 3.84).

Direct investment equity positions included in the CDIS consist of listed equity, unlisted equity, and other equity. In principle, listed equity positions can be valued at market prices, however, there may be no observable market prices for positions in unlisted or other equity. When actual market values are not available, equity positions, in the CDIS, should be valued at own funds at book value (OFBV6) as stated in paragraph 3.12 of the CDIS Guide. The more frequent the revaluation of assets and liabilities, the closer the approximation to market values; data that are not revalued for several years may be a poor reflection of market values.

The CDIS Guide requires that, when there are no observable market prices, information be collected from the books of the direct investment enterprise, as they are more likely to consider the activities of the direct investment enterprise than are the books of the direct investor, and because using this valuation helps to promote comparability and consistency of the information collected between economies. The application of OFBV based on the books of the direct investment enterprise would avoid any bilateral asymmetries arising from valuation. However, compilers may face difficulties in having access to this information to compile outward data because, in this case, they may have no access to the books of the nonresident direct investment enterprise. Also, the assets and liabilities reflected in the books should include revaluations to provide an accurate proxy of market values.

The CDIS metadata reveals that not all countries follow the international valuation standards to compile direct investment equity positions, which may give rise to bilateral asymmetries.

Figure 9 reflects the different valuation methods used by CDIS reporting economies based on their replies to the CDIS metadata questionnaire. It shows that most of the CDIS reporting economies follow the international standards and value listed equity at market value and unlisted equity at OFBV. However, a significant number of economies use OFBV to value listed equity. This deviation of the standards may not cause significant asymmetries when the books of the direct investment enterprise are frequently revaluated and reflect the current market value of the company. On the other hand, it could produce large asymmetries when the values recorded in the books of an enterprise are not revalued for several years and are a poor reflection of market values.

Figure 9.
Figure 9.

Primary valuation principle for Listed Equity (%)

Citation: IMF Working Papers 2017, 261; 10.5089/9781484330036.001.A001

Source IMF’s CDIS, 2016 release

Also, significant asymmetries may arise when economies apply other valuation methods, different than market value or OFBV, such as the historic cost (see Box 1). “Historic cost, in its strict sense, reflects the cost at the time of acquisition, but sometimes it also may reflect occasional revaluations.” (BPM6, paragraph 3.88 (f)).

A Case of Valuation at Historic Cost: The United States

In general, equity investment valued at historic cost will produce smaller estimates than valued at market value or OFBV because this valuation method, in principle, does not include revaluations. One of the countries using the historic cost is the United States. All CDIS reporting economies which have provided outward equity positions in the United States (different than null or confidential), when the United States has also provided inward equity positions received from them (different than null or confidential) during the last five years, have been selected. A total of 22 economies meet these criteria.

Table 8 shows the aggregated value reported by the United States against the selected 22 economies as inward equity positions and the aggregated value reported by the 22 Unites States’ counterpart economies against the United States as outward equity positions. It shows that the asymmetries are always negative, that is, the total amount reported by the US is smaller than the total amount reported by its counterpart economies for the all the period.

Table 8.

A case of valuation at Historic Cost

(US dollars, billions)

article image

However, in some cases, the bilateral asymmetries at individual economy level are positive, in particular, the asymmetries with two of the selected countries (Belgium and the United Kingdom) are positive for the five years covered in this analysis. These positive asymmetries could be explained by the valuation method used by the counterpart economies (Belgium and the United Kingdom use OFBV to value listed equity instead of market value) or by other reasons described in this paper (e.g., coverage or geographical allocation).

Compilers should address these asymmetries and try to estimate the market or fair value of direct investment positions on equity.

Valuation Methods for Debt Instruments in the CDIS

Debt instruments consist of: (i) debt securities, which should be valued at market prices and (ii) other debt instruments—that is, loans, deposits, insurance, pension, and standardized guarantee schemes, trade credit and advances, and other accounts payable/receivable—which should be valued at nominal value. Nominal value is defined as the amount the debtor owes to the creditor, which comprises the outstanding amount of principal and any accrued interest not yet paid.

Figure 10 shows the valuation methods used by CDIS reporting economies to value debt instruments as indicated in the CDIS metadata questionnaire. More than 40 percent of the CDIS reporting economies follow the international standards and value debt securities at market value and other debt instruments at nominal value. In principle, the bilateral asymmetries produced on direct investment debt instruments due to the application of different valuation methods by counterpart economies are not expected to be relevant, as market value is not likely to significantly differ from nominal value.

Figure 10.
Figure 10.

Primary valuation principle for debt securities (%)

Citation: IMF Working Papers 2017, 261; 10.5089/9781484330036.001.A001

Source IMF’s CDIS, 2016 release

As indicated in Section II, the total amounts reported under inward debt instruments are usually larger the total amounts reported under outward debt instruments at the global level. This behavior is also observed at individual country level. In 2015, 80 percent of the values reported under inward debt instrument positions by individual economies are larger than the corresponding values reported under outward debt instruments positions by their counterpart economies.

These positive asymmetries could be produced by overestimated inward debt data and /or by underestimated outward debt data mainly due to different data coverage, and to a less extent to the valuation method applied. Compilers usually have access to more accurate data to compile inward direct investment (that is, direct investment in the reporting economy), than to compile direct investment outward (that is, direct investment abroad). Other compilation issues may take into play; for example, debt instrument liabilities (including in the form of intercompany debt) are regularly more closely monitored in the context of external debt than debt instruments assets. In addition, the methods to compile intercompany debt may focus on initial debt disbursements, but may not properly track any debt repayments to the direct investors, or the debt repayments are accounted for through other means or registered in other items (repatriating benefits, under invoicing, etc.). Therefore, in principle, outward debt instruments tend to be underestimated.

Debt Between Affiliated Financial Corporations

Debt between selected affiliated financial corporations (deposit-taking corporations, investment funds, and other financial intermediaries except insurance corporations and pension funds) should be excluded from direct investment, because it is not considered to be so strongly connected to the direct investment relationship. This is, when both the resident and the nonresident affiliated enterprises are selected financial corporations, their intercompany debt is excluded from direct investment and included instead in portfolio or other investments, as appropriate. For some economies, data sources may lack of sufficient detail to identify these debt positions between selected affiliated financial corporations, specially to identify the institutional sector of the nonresident enterprise. These positions would therefore not be excluded from direct investment, leading to asymmetries when counterpart economies follow the recommendation and exclude them.

The CDIS metadata shows that most CDIS reporters exclude this debt from direct investment; therefore, in principle, this exclusion should not generate significant bilateral asymmetries.

F. Data Sources and Estimation Techniques

The use of different data sources to collect direct investment positions, could also give rise to bilateral asymmetries between economies. Compilers collect direct investment positions from diverse and complementary data sources that are combined trying to avoid duplications or omissions.7

The main data sources used to collect direct investment positions are enterprise surveys and the financial statements of the companies; they are the most adequate and accurate data sources to compile direct investment positions and should result in consistent data across economies. These data sources could be complemented with additional information. Some countries use complementary data sources to fill data gaps or until the results of the surveys are available.

Figure 11.
Figure 11.

Data Sources used to collect CDIS data (%)

Citation: IMF Working Papers 2017, 261; 10.5089/9781484330036.001.A001

Source IMF’s CDIS, 2016 release

Additionally, some countries estimate the data for nonresponding units. The methods used are included in Figure 12 based on the replies to the CDIS metadata questionnaire.

Figure 12.
Figure 12.

Methods to estimate data for non responding units (%)

Citation: IMF Working Papers 2017, 261; 10.5089/9781484330036.001.A001

Source IMF’s CDIS, 2016 release

Most of the reporting economies (around 70 percent) apply some estimates to fill the gap of nonresponding companies. Also, most of the compilers indicate in the metadata questionnaire that the response rate to the surveys is quite high, therefore, in most cases, the estimated amounts are not relevant for the economy.

The use of other data sources different than enterprises surveys or the financial statements of the direct investment enterprises and the use of different methods to estimate data for nonresponding units could produce asymmetries between counterpart economies, but they are not expected to be relevant provided information related to key companies is compiled.

Accumulation of Transactions

As shown in Figure 11, few economies use accumulation of transactions as data source to estimate direct investment positions in the CDIS. Also, some economies use accumulation of transactions to estimate data for nonresponding units.

In principle, compilers may estimate the value of debt instruments positions, different than debt securities, by accumulating transactions with adjustments for exchange rate changes. However, positions on equity and debt securities are more likely to be affected by price changes; therefore, the estimation of positions from balance of payments transactions should take in consideration adjustments for exchange rate and price changes, which may be more difficult to estimate for each negotiable instrument and could lead to bilateral asymmetries.

G. Criteria for Identifying Entities in a Direct Investment Relationship

In order to compile direct investment statistics, it is important to identify the resident institutional units that are in a direct investment relationship with nonresidents. Direct investment positions could be misclassified, under portfolio or other investment, if the direct investment relationship has not been identified. These misclassifications could also lead to bilateral and global asymmetries.

Economies should use the framework for direct investment relationships (FDIR) to identify entities in a direct investment relationship. However, some economies follow other criteria and use more simplified methods, such as the direct influence/indirect control, or the participation multiplication method. This results in the classification of direct investment positions under other functional categories—portfolio investment or other investment.8

Figure 13.
Figure 13.

Methods used to determine direct investment relationships (%)

Citation: IMF Working Papers 2017, 261; 10.5089/9781484330036.001.A001

Source IMF’s CDIS, 2016 release

More than 50 percent of the CDIS reporters apply the recommended standards and identify direct investment relationships based on the FDIR for both inward and outward positions. Positions between related companies could be misclassified for economies using other method.

VI. Addressing Bilateral Asymmetries

The previous section examined the reasons for asymmetries. This section goes one step further proposing ways for addressing them; however, it also recognizes the challenges in the reduction of asymmetries. This section concludes with specific planned actions to reduce asymmetries, including initiatives led by international organizations.

To reduce bilateral asymmetries, it is essential to comply with international guidelines for recording CDIS data by applying uniform standards in terms of coverage, geographical counterparty allocation, valuation, and criteria for identifying entities in a direct investment relationship. In addition, reducing the reporting of confidential and unallocated data by doing best efforts to allocate investments to specific counterpart economies can also contribute to reduce bilateral asymmetries. Finally, bilateral data exchange at enterprises’ level has proved to be an efficient way to identify the reasons for asymmetries and ultimately address them, but may be limited due to confidentiality constraints. Data exchange at more aggregated levels and metadata exchange do not pose confidentiality issues and are useful in understanding asymmetries, but may be less likely to allow addressing and reducing asymmetries. The CDIS provides very valuable information on more aggregated data levels and metadata.

The reduction of asymmetries has always been a challenge for national compilers. Economies involved in bilateral asymmetries exercises conducted by STA have reported specific planned actions to reduce asymmetries. In addition, there are various initiatives led by international organizations, including the IMF, to facilitate bilateral meetings between countries interested in discussing asymmetries in their direct investment statistics. The steps taken to increasing awareness of the asymmetries and understanding the reasons that explain them are, per se, an important achievement, even if it is not feasible to reduce them.

A. Compliance with Recommended Guidelines for Recording CDIS Data

Compliance with recommended guidelines is a key factor that would reduce asymmetries. Uniform standards in terms of coverage, geographical counterparty allocation, valuation, and criteria for identifying entities in a direct investment relationship are essential.

In terms of coverage, direct investment data should include investments in or from resident SPEs, positions between fellow enterprises, direct investment involving nonfinancial corporations, and real estate. However, in some economies SPEs data may be excluded or only partially covered; fellow enterprises’ data may not be identified and therefore classified in data categories other than direct investment; nonfinancial corporations may be poorly covered due to low response rate and lack of the legal mandate to collect data for statistical purposes; and investments in real estate may be out of the scope of the reporting system.

In terms of geographical allocation, the breakdown of positions by counterpart economy should be based on the immediate (first) counterpart economy. However, for inward, some economies report this breakdown based on the ultimate investing economy—that is, the economy in which the ultimate investor is resident—or if the first enterprise is a nonresident SPE, they may determine the counterpart economy looking through the nonresident SPE until the next enterprise in the ownership chain.

In terms of valuation, market value is the recommended basis for the valuation of listed equity and debt securities. Unlisted equity and other equity should be valued using the concept of OFBV, as proxy for market value, obtained from the books of the direct investment enterprise to ensure data comparability and provided frequent revaluations of assets and liabilities are considered. Debt instruments other than debt securities are to be valued at nominal value (amount the debtor owes to the creditor, which comprises the outstanding principal amount including any accrued interest not yet paid). Most of the CDIS reporting economies apply these standards, however, some economies do not follow these recommendations mainly due to limitations in their data sources or compilation process.

In addition, debt between selected affiliated financial corporations should be excluded from direct investment, but in some economies data sources may lack of sufficient detail to identify and exclude this debt from direct investment.

In terms of data sources, position (stock) source data should be used based on enterprise surveys or financial statements of the direct investment enterprises. However, in the absence of position source data, few economies estimate positions from balance of payments transactions, or may use this estimation method for certain sectors/reporters for which position data are unavailable. This can result in asymmetries, particularly if data are not adjusted from changes in positions caused by reasons other than balance of payments transactions (price and exchange rate changes and other flows).

In addition, statistical techniques related to the survey frame need to be appropriate, otherwise, they can lead to inaccurate data. Some economies may use sample surveys, may target enterprises or investments above a certain threshold in their surveys, or may have low response rate to surveys. In these cases, estimation techniques could be used for grossing up the sample to the universe of enterprises, accounting for positions below the threshold, and addressing non-response.

In terms of the criteria for identifying entities that are in a direct investment relationship, the FDIR is conceptually the preferred method. Economies should aim to evolve from more simplified methods to the FDIR.

B. Reducing the Reporting of Confidential and Unallocated Data by Counterpart Economy

Economies should do best efforts to allocate investments to specific counterpart economies and reduce, to the extent possible, the use of confidential/unallocated counterparts when reporting direct investment data. Whereas higher granularity poses confidentiality issues, the rules to determine when data are confidential may need to be revised and applied only when there is potential risk for identifying the reporting enterprise. Likewise, whereas it may not be feasible to allocate the counterpart economy for certain positions or investments, CDIS reporters should do best efforts to estimate the breakdown by specific counterpart economy.

Unallocated data may play a role in economies that participate in the CDIS for the first time, as they may not be able to allocate all positions by counterpart country in a first stage.

However, as their data collection systems are enhanced and delayed information received, they should revise back past time series and revisit unallocated figures. CDIS reporters should do best efforts to minimize the use of unallocated data by estimating counterpart economy breakdown. Estimations could be based on counterpart allocation of past data if available, transaction data, mirror information, etc.

C. Data and Metadata Exchange

Data exchange at bilateral economy level has proved to be an efficient way to identify the reasons for asymmetries. The most efficient way to identify and ultimately address asymmetries is through a detailed data reconciliation, desirably at the enterprises’ level. Data exchange at enterprise level can help identifying missing investments/reporters, deviations in valuation, timing differences, etc., but may be limited due to confidentiality constraints. Data exchange at more aggregated levels and metadata exchange are also useful in identifying and understanding the reasons for asymmetries and do not pose confidentiality issues, but may be less likely to allow addressing and reducing asymmetries. The CDIS provides very valuable information on more aggregated data levels and metadata.

Data Exchange

Enterprise level data are usually available to compilers through data report forms collected for statistical purposes. Some mechanisms are in place to successfully exchange micro level data at enterprise level bilaterally, such as the FDI Network launched under the auspices of Eurostat and the ECB (see Box 2). Some countries that do not participate in such Network also conduct bilateral exchanges of data regularly, especially with their main counterpart economies.

Bilateral exchange of micro data between statisticians is considered a good practice to reduce asymmetries and to identify the reasons behind them, even if it does not allow for the reconciliation of all transactions and positions.

The FDI Network

The transmission and exchange of confidential data in the context of the FDI Network is foreseen exclusively for statistical purposes and in particular for increasing the consistency/quality of the EU/euro area balance of payments statistics. The transmission of confidential data is enabled pursuant to the specific regulations of the European Parliament and of the Council. The exchange of micro level data between direct investment compilers is based on a secure data exchange between countries which are parties to the financial transaction/position. The concerned countries will then try to reconcile their data.

(ECB/Eurostat BOPCOM 14/20)

However, there are limitations to data exchange due to confidentiality constraints. Confidentiality both at microdata level (enterprise) or at higher levels of aggregation (for some types of investments or sectors) may hamper analyzing the bilateral asymmetries with counterpart economies. For some economies, the possibility of sharing micro data with third parties or even with statistical agencies of counterpart economies may not be feasible due to confidentiality or legal constraints. The envisaged increase in data flow with more granular information requires substantial work in several areas, including on confidentiality and legal issues. In this context, the new recommendation on promotion of data sharing of the Second Phase of the G-20 Data Gaps Initiative (DGI-2) (recommendation II.20) was welcomed by the G-20 economies to facilitate sharing the experiences with exchanges of granular data and identifying ways to overcome existing barriers as well as to strengthen the linkages and consistencies between datasets. Some economies conduct bilateral exchanges of micro data through bilateral agreements or contracts to preserve confidentiality. For this purpose, some economies request prior authorization from reporters.

Data exchange can also be done at more aggregated levels posing less confidentiality issues. In this regard, the granularity of CDIS data with breakdowns by instrument (net equity and net debt), further breakdowns of net debt instruments into gross assets and gross liabilities, and the separate identification of positions between fellow enterprises provides a wide range for analysis.9

Metadata Exchange

Metadata exchange has also proved to be an important and useful way for identifying and analyzing the reasons for asymmetries. Various economies collaborate with their main counterparts to enhance their understanding of each other’s direct investment compilation systems, by conducting professional exchanges with the officials responsible for the compilation and dissemination of direct investment statistics.

The results of the CDIS metadata questionnaire (available at the CDIS website) provide detailed information on the collection and compilation practices adopted by CDIS reporting economies. They are an important source of information for assessing data quality and deviations from recommended standards and have been used to explain some of the bilateral asymmetries in this paper. The structured design of the metadata questionnaires allows easy cross-economy comparisons, facilitating the identification of methodological differences with main counterparts. The questionnaires also include compilers’ contact data to facilitate information exchange with main counterparts. Methodological comparability is improving over time with the implementation of the international standards set in the BPM6 and the 2015 CDIS Compilation Guide, but there are areas were practices vary (see Annex VII).

D. Specific Planned Actions to Reduce Asymmetries

The reduction of asymmetries has always been a challenge for national compilers. First, it is time consuming and resource intensive for economies to identify the specific reasons and even more to address the asymmetries. Second, solving asymmetries in one item may create new asymmetries in another one. Therefore, investigating topics that are closely linked and the consistent recording of external positions is essential to reduce asymmetries. It is also of importance ensuring consistency between external sector statistics and other statistical domains, mainly national accounts, and linking asymmetries arising from both data sets.

Planned by CDIS Participants

Economies involved in bilateral asymmetries exercises conducted by STA reported on their specific planned actions to reduce asymmetries. As part of the second exercise of the CDIS bilateral asymmetries10 conducted in 2016, STA consulted with 47 CDIS participating economies on any plans to implement specific actions or changes in the collection and/or compilation process to reduce their asymmetries. Annex VIII shows a summary of the main actions they reported. Main categories of planned actions include: continue analyzing the reasons for discrepancies on an aggregated level; more active bilateral consultations and interaction with counterparts; more active participation in the FDI Network (European countries); continued work to develop the European Group Register; revisions to CDIS data; and enhancements in the reporting systems and improvements in data sources.

Ongoing Initiatives by International Organizations

In addition, there are various initiatives led by international organizations to facilitate bilateral meetings between countries interested in discussing asymmetries in their direct investment statistics. OECD organizes bilateral asymmetry meetings adjacent to the Working Group on International Investment Statistics (WGIIS) meetings. Eurostat facilitates platforms for bilateral asymmetry meetings. In addition, the IMF, with the support of Bundesbank, will conduct a workshop for CDIS compilers aimed at reducing asymmetries in November 2017.

Work ahead to facilitate data exchange include following up on the possibility of Eurostat sharing the infrastructure of the FDI Network to expand it at world level (see IMF, BOPCOM 16/18).

Work ahead on the CDIS metadata questionnaire include:

  • Enhancements to the questionnaire through the addition of new specific questions on bilateral asymmetries, for example on main reasons that explain asymmetries for specific counterpart countries, instruments (equity/debt), type of investments, or sectors involved; future or current work to address/understand/reduce them; links to any related publication on the research conducted in this area by the compiling agency, etc.; and

  • Continued efforts to encourage economies to update and fine-tune metadata information as needed, to account for most up to date information on any possible methodological differences across countries that could explain asymmetries.

The steps taken to increasing awareness of the asymmetries and understanding the reasons that explain them are, per se, an important achievement, even if it is not feasible to reduce them.

Same Series

Other IMF Content

Other Publishers

Cover IMF Working Papers
Asymmetries in the Coordinated Direct Investment Survey: What Lies Behind?
Author: Emma Angulo and Alicia Hierro
Volume/Issue:
Volume 2017: Issue 261
Publisher:
International Monetary Fund
ISBN:
9781484330036
ISSN:
1018-5941
Pages:
49
DOI:
https://doi.org/10.5089/9781484330036.001

Global Asymmetries in CDIS

(US Dollars, billions)

Number of CDIS Reporting Economies

Outward vs Inward by Instrument in the CDIS

(US Dollars, billions)

Global Asymmetries in the IIP

(US Dollars, billions)

Number of IIP Reporting Economies

DI Assets vs DI Liabilities by Instrument in the IIP

(US Dollars, billions)

Number of economies including real estate data in the GDIS

Not Specified (including confidential) over Total DI in the CDIS

(US Dollars, billions)

Primary valuation principle for Listed Equity (%)

Primary valuation principle for debt securities (%)

Data Sources used to collect CDIS data (%)

Methods to estimate data for non responding units (%)

Methods used to determine direct investment relationships (%)