6. Experience in Conducting the 1997 CPIS
- International Monetary Fund
- Published Date:
- May 2002
A. Choice of Collection Method
6.1 Countries’ choices of collection method were influenced by a number of factors, such as availability of the data, ease of collection, respondent burden, availability of resources, existing surveys for collecting data for the international investment position (IIP), a country’s national circumstances, and its institutional arrangements. What is important is that the choice maximizes the coverage (without double counting) while minimizing the costs to the respondents and remaining within the resources of the compiler.
6.2Table 6.1 shows the choices made by the 29 jurisdictions that participated in the 1997 CPIS. The aggregate/end-investor-only approach was chosen by 4 countries; 1 country used an aggregate/end-investor and investment managers approach; the aggregate/combination of end-investor and custodian was chosen by 12; the security-by-security/end-investor approach was the choice of 1 only; and the security-by-security/combination of end-investor and custodian was used by 6 countries. Four countries used other combinations, mixing the aggregate approach with the security-by-security approach. Aggregate/custodian-only and security-by-security/custodian-only approaches were not used by any country. One country did not use any of these approaches because it used administrative records.
|Aggregate Approach||Security-by-Security Approach|
Mixed, depending on the available information.
Also used investment managers.
Mixed, depending on the available information.
Also used investment managers.
6.3 In its preliminary investigations, the Austrian central bank (Oesterreichische Nationalbank, ONB) found that domestic residents held more than 90 percent of their securities issued by nonresidents with domestic custodians. As a result, the ONB introduced a survey primarily targeted at custodians, requiring that they report on a monthly frequency their own holdings of securities issued by nonresidents and those of their clients.
6.4 The reporting population consists of local (or sub-) custodians (those closest to the end-investors, not global custodians) to avoid double counting. In this new survey, domestic custodians have no difficulty in identifying holdings of securities by residents. End-investors report data only if their holdings of securities issued by nonresidents are entrusted directly to nonresident institutions and have a value exceeding 1 million Austrian schillings on the reporting date. As direct investors also use the services of domestic custodians and as domestic custodians cannot determine whether securities are issued by related enterprises abroad, these holdings (which are appropriate for direct rather than portfolio investment) are estimated from data provided by direct investors (Section II, below). Data on nonresident ownership of securities issued by residents are also collected from domestic custodians to measure portfolio investment liabilities.
6.5 In its preliminary investigations, Bermuda determined that the bulk of cross-border portfolio investment is held by the financial sector (including the international financial sector) and that most of these holdings are likely to be placed with nonresident custodians. From discussions with the relevant industry associations, it was determined that banks, insurance companies, and mutual funds could provide the required information on an aggregated basis. For these reasons, the aggregate end-investor approach was selected. Reporting by custodians on an aggregated basis was also included to provide a supplementary data source.
6.6 In its preliminary investigations, the Bank of Israel concluded that, because residents have to report their cross-border portfolio investment under exchange control regulations, data sources based on these regulations ensure a complete coverage of cross-border portfolio investment and can also be compiled with a high frequency. Data were reported by resident custodians (using the security-by-security approach to facilitate quality control and checking on an instrument basis against other data sources) and by end-investors (on an aggregate basis). Reporting by end-investors comprises those residents who use the services of nonresident custodians or hold securities in self-custody, and includes businesses and households.
6.7 Because this approach was developed to support the compilation of an IIP statement for portfolio investment assets and liabilities, the resulting database compiled by the Bank of Israel’s Foreign Exchange Control Department (FECD) covers residents’ holdings of cross-border portfolio assets and nonresidents’ holdings of securities issued by residents. Resident custodians report the value of securities traded in Israel that are held by nonresidents. Residents’ holdings of shares issued by Israeli companies abroad are compiled from information supplied by the issuers. This information is used to derive estimates of nonresidents’ holdings of shares issued by Israeli entities abroad.
6.8Figure 6.1 illustrates the framework of data collection in Japan. In its preliminary investigations, the Bank of Japan concluded that institutional investors are the only sector likely to use the services of non-resident custodians. Both the nonfinancial corporate sector and households were believed to prefer the services of domestic custodians. For this reason, reporting by resident custodians and institutional investors was concluded to be sufficient to provide a complete coverage for the IIP and CPIS.
Figure 6.1:Framework of Data Collection in Japan1
1To avoid double counting, only the data in the shaded areas are aggregated to produce total Japanese portfolio investments.
2Disaggregated into public sector, banking sector, and other sectors.
6.9 The Foreign Exchange and Foreign Trade Law and related ordinances provide the legal framework for collecting portfolio investment assets and liabilities as part of the IIP. This legal framework (which was in place before the 1997 CPIS) required that a summary breakdown of portfolio investment assets, by the country of residence of the issuer, be reported, and specified that data be reported on an aggregate basis. Banks were required to report both in their capacity as end-investors and custodians. Hence, the legal framework met the requirements of the CPIS.
6.10 For the IIP and CPIS, the Bank of Japan relied mainly on reports submitted by banks (as end-investors and custodians), securities companies, and institutional investors under the framework prescribed by the Foreign Exchange and Foreign Trade Law. This authority to collect statistics has remained in place notwithstanding recent financial deregulation.
6.11 The most important reason for choosing the aggregate approach was to build on the existing reporting system for IIP purposes, and to avoid any additional burden (that would be seen to result from any change of the basis of reporting) on reporters at a time of financial deregulation.
6.12 The Netherlands Central Bank has been conducting an annual survey of portfolio investment assets and liabilities since 1986. From its experience over the years, the national central bank has concluded that, with the exception of institutional investors, residents use the services of domestic custodians. From discussions with custodians, it was concluded that custodians could be relied upon to provide accurately the required information on an aggregate basis. Hence, a combined custodian/end-investor reporting system was selected, with the latter confined to institutional investors and a few important investment companies. For pension funds and insurance companies, total holdings as reported in the CPIS end-investor survey could be checked against data reported in their annual balance sheets. Any differences between these totals and those reported to be held by resident custodians on behalf of pension funds and insurance companies are taken to be securities held with nonresident custodians.
6.13 In its preliminary investigations, the Office for National Statistics (ONS) concluded that the existing end-investor surveys used to compile portfolio investment assets in the IIP statement could also be used for the CPIS. For this reason, an aggregate end-investor approach was selected. Questions were added to seek information on the country of residence of the issuer.
6.14 This approach had a number of attractions. First, end-investors were already familiar with the enquiry forms and the subject matter. The contact who has responsibility for the total stock of portfolio investment assets would be the best person to complete the geographical breakdown of those assets. Second, the enquiry team in the ONS was already experienced in the subject matter, had established relationships with end-investors, and was in a good position both to encourage the respondents and guide them through the requirements of the survey. Third, the legal basis of the CPIS data collection could be linked to the existing survey, thus avoiding the need to introduce a new legal provision to collect the information. Fourth, the costs to both the end-investor and the ONS could be minimized in that both already had established systems for generating the total stock of portfolio assets. Other economies resulted from using the existing survey design. Fifth, linking the data to the balance sheet returns provided an in-built consistency check. Finally, potential problems of double counting were minimized in that end-investors would only be reporting their own holdings of portfolio assets.
6.15 From its experience in 1994 from conducting a survey similar to the CPIS, the U.S. Treasury concluded that the bulk of residents’ holdings of cross-border portfolio investment assets is entrusted with resident custodians. Reporting by custodians was preferred because there are a comparatively small number of custodians as against the very large number of enterprises that are likely to hold cross-border portfolio investment, and because reporting by custodians made it possible to include the household sector and small businesses. The sampling of enterprises was considered but not thought practical, given the prospect of complete reporting by custodians. For institutional investors likely to use the services of nonresident custodians, an end-investor survey was used. The security-by-security approach was used for custodians because experience had shown that initial data received from custodians were frequently inaccurate as regards the country of residence of the issue and the price of securities. It was found that errors could most readily be detected and resolved only by collecting security-level data. Also, by collecting security-level data, more detailed information would be available for analysis.
6.16 Primary custodians who employ the services of domestic global custodians are required only to report the custodian’s name and the amount entrusted, and these data are used in the same manner as those of end-investors employing domestic custodians. End-investors are required to report their foreign holdings security-by-security only when they do not utilize a resident custodian. If they use a resident custodian, they are required to report only the custodian’s name and the amount entrusted. These amounts are not added to the survey totals, since this would cause double counting, but they are used to verify the accuracy of the data reported by custodians.
B. Limitations in Approaches to Coverage
6.17 Each of the systems faces potential problems of coverage. In the primarily custodian-based systems, such as Austria, the Netherlands, and the United States, the potential weakness was (and in some cases, where there are ongoing surveys, is) that holdings may be missed if residents do not use the services of resident custodians. The Netherlands compilers cover this problem through the use of published annual accounts of the major investing institutions, although they acknowledge problem areas among nonfinancial companies and the household sector. Austria and the United States require large investors who entrust securities directly with foreign-based financial institutions to report their holdings directly to the compiler. Both Austria and the United States consider aggregate holdings by small investors (including the household sector) who hold securities in self-custody or entrust them with nonresident custodians to be relatively small.
6.18 For all of these collection approaches, the risk of missing direct holdings abroad (especially of households) increases as more use is made of on-line trading in securities issued by nonresidents directly with nonresident brokers. There is, as yet, no satisfactory means of addressing this problem.
6.19 In its preliminary investigations, the ONB found that banks identify securities by using the security codes of either the Austrian numbering agency (Oesterreichische Kontrolbank) or the German numbering agency (Wertpapiersammelstelle Frankfurt). These codes are linked to the International Securities Identification Number (ISIN) code system.60 By requiring custodians to report a security identification code and the par value for each security held, the Austrian compilers have been able to ensure, with the use of commercial security databases (see Chapter 4), that securities are correctly allocated geographically and revalued to end-period market prices. Also, because the ONB collects security-by-security information on portfolio investment stocks and flows, it can reconcile stocks and transaction data, recalculate historical time-series aggregates, and ensure that data are consistent with international concepts and principles.
6.20 Although the purchase of commercial security databases is costly, the Austrian compilers have found that the benefits of reliable and frequent data outweigh the costs. Commercial databases are mainly used as the preferred source for market prices of securities. Custodians in Austria prefer security-by-security reporting because the central bank bears the costs of the compilation of aggregates and the costs of developing and running the commercial securities database.
6.21 The Bank of Japan considers the coverage and quality of data reported annually by end-investors and custodians on their holdings of portfolio investment to be high. The reported data are checked against other information, and any apparent irregularities are followed up with the respondent.
6.22 Although experience from other countries suggests that it may be very difficult to collect accurate aggregate data from resident custodians, Netherlands compilers have extensive and in-depth experience in collecting data by this method and are convinced that their data are of good quality. By comparing data reported by resident custodians against data reported in the annual reports of institutional investors, an independent check is being provided for these investors, assuming that the definitional and conceptual approaches are consistent.
6.23 Because the reporting of aggregate data by resident end-investors would result in a heavy response burden, involve an unmanageably large number of respondents in the survey, and also provide potentially unreliable data on the attributes of the issue and the issuer that are needed for the CPIS, data were collected from resident custodians on a security-by-security basis to facilitate checking and verification against data for holdings reported by resident custodians. Commercial data sources were the preferred source for information on individual securities issued by nonresidents required for the CPIS (such as end-of-period market prices and the country of residence of the issuer). To ensure that the securities were correctly valued and allocated geographically, the survey respondents provided a security identification code number for each security reported. Any of the 33 most commonly used coding systems—for instance, ISIN, CUSIP, SEDOL—were used (see the list in Appendix I). It is not unusual for commonly traded securities to have codes allocated by ten different coding systems. Survey respondents were encouraged to use ISIN codes to identify securities, but in the 1997 survey only 26 percent of all securities were so identified.
6.24 To assist in the process of identifying and valuing securities, the U.S. compilers purchased commercial security databases that covered all the commonly used coding systems and provided pricing information on all securities that had traded within the previous year on any of the world’s 87 largest non-U.S. securities exchanges. (See Chapter 4 for a discussion of securities databases.) These databases provided key information that was not always supplied by survey respondents; they also helped to resolve conflicting information, and confirmed the information reported. These included: ISIN and other security identification codes; security type; registered or bearer; issue date; maturity date; coupon rate; issuer name; security description; country of issuer; country of guarantor; country where issued; currency; industry sector; price, price type (current, old, estimated); pricing date; default status; American depository receipts (ADRs); ratio of number of ordinary shares per ADR; dividends per share (past twelve months); and nominal amount outstanding for bonds or the number of shares on issue.
6.25 Data from commercial databases are used extensively to help verify and supplement data received from respondents. Verification programs map reported data to the commercial database based on the security identification code. The reported data are then compared field-by-field with the commercial data, and mismatches are specified in an exception report. In addition, data fields left blank in the reported data can frequently be obtained from the commercial database. This process can also help evaluate the overall accuracy of a given reporter’s submission on the assumption that a greater or lesser number of errors in items that can be checked is indicative of the accuracy of data items that cannot be so easily verified.
6.26 The use of commercial databases also reduces reporter burden because some data items that may be required or are highly desirable for analytical purposes (such as amount outstanding, a single security identification system, dividends paid, coupon rate, industry sector, and country of guarantor) need not be obtained from the survey respondents.
6.27 Nonetheless, a major problem for the compilers was revaluing to end-of-period market prices those rarely traded securities, particularly equities, that were reported as unpriced by custodians. However, these securities accounted for only a small proportion of the total number of securities reported.
II. Separating Direct Investment from Portfolio Investment
6.28 One of the major statistical concerns for the balance of payments, IIP, and CPIS is the separation of financial investment into functional categories. In particular, within the context of this survey, direct investment should be separated from portfolio investment. However, if data are collected from data sources that are either unaware of this distinction or unable to make the separation (because they have insufficient information), compilers need to have in place mechanisms that will ensure that the data remain as cleanly separate as possible.
6.29 Because resident custodians are unlikely to be able to identify whether securities are issued by parties (direct investment enterprises) related to the issuer (direct investor), other sources of data will be needed to determine such holdings of securities, which should be included under direct investment abroad. It is frequently the case that direct investors keep such securities in self-custody or use the services of nonresident custodians, but practice varies across countries. Most countries conduct end-investor direct investment surveys in conjunction with portfolio investment surveys, and design the surveys to avoid double counting.
6.30 Australia has an ongoing, integrated collection system, the Survey of International Investment (SII), that collects, inter alia, portfolio and direct investment and integrates flows and balances. Throughout the SII survey forms, respondents are asked to record separately their liabilities to and claims on (i) nonresident direct investors, (ii) direct investment groups abroad, and (iii) other nonresident investors. Only data collected for the third of these categories are relevant to the CPIS. The definition of direct investment is in line with BPM5, and an explanatory diagram is drawn from the IMF’s Balance of Payments Textbook (Washington, 1996). For each instrument, respondents are generally required to record separately particulars for nonresident direct investors, direct investment groups abroad, and other nonresident investors on the same page of the survey forms. This helps to prevent misreporting among these three categories.
6.31 A sample of resident direct investors (for outward portfolio investment) was taken to determine the proportion of equity held (i) in custody with resident banks, (ii) in custody with nonresident banks, and (iii) elsewhere. The information was checked against security registers in all cases where the value in custody was more than 10 percent of the total outstanding on the company register. A similar approach was undertaken on the liability side by sampling direct investment enterprises. This permitted an estimate of the direct investment component of all reported investment abroad in equities and of all reported inward investment in equities. See Box 3.3.
6.32 For inward investment, the differentiation between direct and portfolio investment was derived from information provided by the Tel Aviv Stock Exchange and the Israel Securities Authority (ISA).61
6.33 For outward investment, the differentiation between direct and portfolio investment was made by comparing holdings of a particular security against total issues outstanding of that security (as reported in FECD’s securities database).
6.34 CPIS data were checked against the results of a survey of outward direct investment, administrated by the central bank. Because this survey was not fully comparable with the CPIS, the definition of direct investment differed. The outward direct investment survey gave an indication of which direct investment enterprises may have had securities that were reported in the CPIS that should have been treated as direct investment, and therefore provided an indication for further investigation. This approach helped to detect some cases of inaccurate reporting, but its usefulness was limited by lack of supplementary information (such as balance sheets for nonfinancial enterprises) that could confirm possible misreporting.
6.35 The exclusion of direct investment securities was accomplished by a twofold procedure. First, the guidance notes to the CPIS clearly defined direct and portfolio investment and gave precise instructions to respondents on how to exclude direct investment positions. Second, information was taken from results of the direct investment survey that was conducted at the same time as the CPIS to ensure that there was no double counting.
III. How Market Price Valuation Was Addressed
6.36 One of the central principles of the CPIS (as with the IIP and the balance of payments) is the use of market price. For regularly traded equities and debt securities, this should not present any problems except that end-investors may not value all their holdings of securities at market prices and resident custodians may not keep records of the market prices of the securities they are holding. End-investors may follow alternative valuation principles, such as par value, acquisition cost, amortized value, or keep separate books for different portfolios of securities. For example, banks may maintain separate trading books (at current market prices) and investment books (at acquisition cost) for debt securities. (On the other hand, institutional investors and collective investment schemes are likely to use market prices for all their holdings of securities to price them on a regular and frequent basis.) Other problems arise in the case of infrequently traded securities. Lastly, where market sources are used for prices, issues arise concerning the treatment of interest accrued, the use of bid or ask prices, and the impact of time zones.
6.37 Following is description of the experience of four countries (Australia, France, Portugal, and the United States) in seeking to ensure for the 1997 CPIS (and any ongoing survey of portfolio investment) that data were either reported on a market price basis or where there were mechanisms in place by the compiler to make adjustments, where necessary. Securities not traded actively or denominated in foreign currencies are particular problems. Not all of these problems were overcome.
6.38 Respondents to the SII were requested to report stocks at market value and the explanatory notes provided guidelines on possible valuation methods. In measuring the value of equity securities, respondents were asked to adhere to the following principles: (i) For listed enterprises, the market value of the equity positions should be reported using a recent transaction share price; if recent transaction prices were not available, the midpoint of the quoted buy and sell prices of the shares on their main stock exchange on the reference date specified provided a useful approximation. (ii) For unlisted enterprises, if a market value of the shares was not available, the respondent was asked to estimate the market value by one of the following methods (in descending order of preference): a recent transaction price, a director’s valuation, or net asset value. For debt securities, survey respondents were asked to report traded price on the date specified. If that value was not available, they were asked to report using, in order of preference, one of the following methods: yield to maturity, discounted present value, face value less written-down value of discount, issue price plus amortization of discount (less amortization of premium), or another mark-to-market basis. The bases upon which debt securities were actually valued are not known.
6.39 Valuations were reported at market prices except for the nontrading book of resident banks, which were reported at book values. The latter accounted for about 25 percent of total holdings of long-term debt securities issued by nonresidents. No attempt was made to adjust the latter to market prices. Security-by-security portfolio investment survey data and banking trading portfolio data were collected through separate data-collecting systems.
6.40 It was not possible to adjust fully to market prices because of a lack of information and the practice of using unit values and of reporting unit value as a percentage of the nominal value. This led to further discussion with respondents. The Lisbon Stock Exchange’s publications were consulted to perform quality checks on unit values of securities reported in the CPIS and to perform consistency checks against the total value of each investment fund. For each investment fund, published data provided information on the portfolio composition (quantity, market price, and currency) for relevant dates. In cases where respondents could not identify a market price, they were asked to use the alternative methods suggested in paragraphs 3.33–3.42.
6.41 Information on market prices was readily available on large frequently traded issues, which make up the majority of debt issues and the vast majority of equity issues. In the event of missing prices, information available from both commercial sources and other reporters were used. When all such sources were exhausted and failed, estimated prices were used for debt securities. For equities, estimated prices were also used, but these securities would probably be undervalued at very low levels and, in any event, represented only a very small percentage of the total.
IV. Treatment of Accrued Interest
6.42 The 1997 CPIS sought, as the 2001 CPIS seeks, market prices of debt securities to include interest accrued but not yet paid (the so-called dirty price).
6.43 In many countries, long-term debt securities are priced on a “clean” basis—that is, excluding accrual of interest; however, some markets include accrual of interest in “dirty” prices. For long-term debt securities held in investment accounts (that is, they are not held for trading purposes—they are usually held for prudential or reserve capital requirements), the valuation is often that of par or acquisition price, and the valuation does not change with market conditions. Where prices for bonds exclude accrued interest, usually the difference between clean and dirty prices is not substantial when coupons are paid frequently. Even though the CPIS seeks information on market prices, respondents frequently will not revalue their “investment book” (as opposed to their “trading book,” which is usually marked to market on a frequent basis—daily, weekly, or monthly). In the same manner, they are unlikely to record accrued interest, and it is often difficult for compilers to obtain the data on the required basis.
6.44 However, for zero-coupon or deep-discount bonds, the difference can be substantial because the amount of interest accruing over five years or more can amount to a substantial proportion of the initial principal lent/borrowed, especially when interest rates are high. In these instances, if respondents are reporting on an aggregate basis, it is important to have respondents report their holdings on a dirty price basis as far as possible. If their holdings are reported on a security-by-security basis, the compiler should be able to make the adjustments, especially if a securities database is available.
6.45 In the 1997 CPIS, short-term debt securities were included only on a voluntary basis, so less experience was reported. Most short-term debt securities are issued at a discount—i.e., they are zero-coupon instruments. There are a number of methods used to record money market instruments: at acquisition cost, at amortized value, at par value, or at market price. Respondents may be able to report at market price those securities held for trading purposes, but, as with bonds held for investment purposes, some respondents may use one of the other methods and be unable or unprepared to report on a different basis. However, because of the shortness of time in which interest can accrue, the issue is less important than for zero-coupon, or deep-discount, bonds. Even so, whenever possible, respondents should be encouraged to include accrued interest.
6.46 The experiences of Ireland, Israel, and Portugal in the 1997 CPIS (and any ongoing survey of portfolio investment) are described below.
6.47 Respondents were asked to provide positions on a clean price market-value basis, market-priced transactions, and valuation changes. In addition, details of outstanding interest (positions) and movements in interest (flows) were required. For the CPIS, the clean price positions were combined with outstanding interest to obtain the dirty price market values of the stocks.
6.48 Tel Aviv Stock Exchange market prices include the element of accrued interest (dirty price), and these figures were used for CPIS liabilities in the local market. Israeli debt issued abroad, however, is based on the nominal value of the debt security and accrued interest.
6.49 Respondents were requested to include accrued interest in the market price valuation. Where custodians and investors were unable to do so, no adjustments were made. For investment funds that were checked through the publications of the Lisbon Stock Exchange, accrued interest was added to the total stock value.
V. Treatment of Nominee Accounts
6.50 Nominee accounts pose a problem for compilers of the CPIS. Nominee accounts are a means through which securities are held by one party as a nominee on behalf of another party, usually for reasons of administrative convenience or confidentiality. (See the discussion on nominee accounts in Chapter 3.) For the purposes of the CPIS (and for the balance of payments and the IIP), these holdings should be “looked through,” that is, treated as though the nominee account had not been used and the beneficial owner is reported as the holder. However, when nonresident nominee accounts and custodians are used, it is not always possible to identify the beneficial owner. For example, if a resident of country A holds securities issued by a resident of country B and uses a nominee account in country C, and the securities are kept in custody in country C, the custodian in country C may not be aware that the ultimate owner is in Country A. In such cases, efforts should be made to encourage the custodian to obtain the residence of the beneficial owner from the nominee. In some countries, this information may be required by the authorities to restrict tax evasion.
6.51 The approaches by Israel, Singapore, and the United Kingdom in the 1997 CPIS (and any ongoing survey of portfolio investment) are described below.
6.52 Although residents may hold nominee accounts with resident custodians, the names of the beneficial owners must be provided to the custodian, Hence, resident custodians had no difficulty in identifying accounts held by residents. On the liability side, it was not possible to identify the beneficial owner of nominee accounts.
6.53 Custodians reported separately securities held in nominee or similar accounts on behalf of nonresidents. They were, therefore, excluded from the CPIS.
6.54 The ONS conducts periodic sample surveys of all companies listed on the U.K. stock exchange to determine the ownership of their share issues, based on the share registers maintained by companies. Holders were classified according to their institutional sector. About one third of all ordinary traded shares were estimated to be held through nominees (mainly representing unit trusts and pension funds, offshore entities, and other overseas investors). For the most part, it was possible to identify the investor, since this information was known to the reporting enterprise (knowledge of the beneficial owner is required by company law). A residual of undesignated nominee accounts was attributed to beneficial owners, based on the proportions that obtained for nominee holdings for which beneficial owners were identified.
VI. Treatment of Collective Investment Schemes
6.55 Units in collective investment schemes (mutual funds, investment trusts, unit trusts) are classified as equity, regardless of the type of fund or assets that the fund acquires. Thus, for example, if a resident in country A owns units in a collective investment scheme domiciled in country B that holds only bonds issued by the government of country C, the resident of country A should report that he or she holds an equity claim on country B, not a holding in government bonds from country C. It is important that respondents are aware of this classification because they may classify their portfolios on the basis of ultimate risk and may, therefore, “look through” the collective investment scheme—and report holdings of government bonds of country C, as if they did not have units in the mutual fund in country B.
6.56 The residence of a collective investment scheme is determined by its legal domicile. Mutual funds may be legally domiciled in one economy (e.g., a SEIFiC), be managed in a second, be administered in a third, and be listed in a number of jurisdictions. An investor may be interested in where and how funds are invested, and possibly in the ownership of the mutual fund (such as the parent company), but less so in its legal domicile. For this reason, care will need to be taken to identify the legal domicile of mutual funds in an end-investor reporting system, especially if they are registered, listed, managed, and administered in different jurisdictions. It is also worth emphasizing that units in collective investment schemes are unlikely to be held with custodians for safekeeping. Hence, the investing country will have to rely on end-investor reporting.
6.57 How Bermuda, Ireland, Israel, Sweden, and the United States treated collective investment schemes in the 1997 CPIS (and any ongoing survey of portfolio investment) is described below.
6.58 An end-investor reporting system was introduced for all legally domiciled mutual funds. The respondents included mutual funds that were legally domiciled in Bermuda and also administered in Bermuda and a number that were legally domiciled in Bermuda but administered elsewhere—which meant that time had to be allowed for the relevant accounting centers to be contacted. Although reporting by resident custodians was considered, this did not prove practical because Bermuda’s mutual funds tended to use the services of nonresident custodians. Respondents were asked to report their gross holdings of cross-border portfolio investment. These could differ from shareholders’ values (net asset values) as a result of leveraging by the managers of the funds (such as by borrowing, the use of repos, or the use of financial derivative instruments). However, in the Bermuda experience, leveraging was not a critical issue. On the asset side, only portfolio investments were included (e.g., any marked-to-market holdings of financial derivative instruments on their balance sheets would be excluded). Holdings of units issued by other mutual funds were reported as issued by the country where such funds were legally domiciled.
6.59 Because the survey was conducted on a voluntary basis, a means for grossing up for non-response was needed. This was done by using the net asset values of respondents as a benchmark—which were required to be reported for regulatory purposes.
6.60 Data are collected from mutual funds that are legally domiciled and/or administered in the Dublin International Financial Services Center. However, for the purposes of the CPIS, only the cross-border portfolio investment assets of mutual funds legally domiciled in Ireland were included.
Israel and Sweden
6.61 Holdings in mutual funds are categorized as equity investments regardless of the type of fund or assets held by the fund.
6.62 If U.S. residents owned shares in a foreign mutual fund, these were counted as equity holdings in the country in which the mutual fund is legally domiciled. This was so regardless of what type of security the mutual fund purchases or where the mutual fund managers physically reside. U.S.-based mutual funds (or their U.S.-based custodian) reported in detail the securities they owned that were issued by unrelated nonresidents.
VII. Direct Holdings Abroad
6.63 One of the major problem areas identified by the 1997 CPIS was determining the extent to which residents used the services of custodians abroad or held securities in self-custody. For some countries, their inability to find reliable data sources for such holdings by households and small companies (which could not be covered by end-investor surveys) was of serious concern. For others, such investments were considered to be small, and their omission not likely to much affect the results of the survey. The full extent of the problem could not be identified,62 but it was believed that some countries had more severe problems than others. For example, in countries where there is a history of bypassing resident financial institutions and investing directly abroad, countries did not have mechanisms to collect the information because a direct survey of households is not a feasible proposition. Quite apart from the cost and difficulty of deriving a representative sample (with a bias toward high-wealth individuals), there is the problem of response rate. If the capital has been placed directly abroad to avoid the attention of the country’s authorities, it is probable that those individuals will be disinclined to provide the information for a statistical survey, regardless of assurances of confidentiality. The difficulties in obtaining this information and the perception of its importance is related to how information may be obtained from third-party custodians (that is, custodians who hold securities issued by nonresidents on behalf of nonresidents). See Chapter 4, Section V, for discussion of the issues related to third-party holdings.
6.64 The experiences of Argentina, Australia, Israel, and the Netherlands in addressing the issue of direct holdings abroad in the 1997 CPIS (and any ongoing survey of portfolio investment) are outlined below.
6.65 The scale of direct holdings abroad by the private sector (households, financial corporations, and nonfinancial corporations) was thought to be significant. Regulatory sources were available for cross-border portfolio investment, including direct holdings abroad, for selected categories of end-investors (banks, insurance companies, pension funds, and mutual funds). For households, an attempt was made to approach nonresident custodians and other intermediaries but this proved unsuccessful. Some attempt was made to estimate households’ cross-border portfolio investment through the use of BIS data and partner-country sources (such as the United States).
6.66 Direct holdings abroad by Australian enterprises are directly covered in the SII collection. The SII targeted all Australian enterprises that were known to have any international investment activity. Although most enterprises channel such activity through investment (fund) managers or the custody system, those enterprises that have direct holdings abroad are asked to report directly on these holdings. Direct holdings abroad by Australian households, on the other hand, are not directly covered by the SII. Households may invest abroad through one of three routes: (i) direct personal holdings of securities issued by nonresidents, (ii) enterprises and legal entities set up by individuals and families to manage their financial affairs (for example, for tax-efficiency reasons), or (iii) superannuation (pension fund) and other collectively managed investment vehicles. While individuals may deal directly with foreign-based brokers, nominee companies, or foreign-based fund managers, such direct holdings are regarded as being relatively insignificant in the Australian context and are not measured in the SII. Since it is considered that the great majority of investment abroad by households passes through either the second or third routes, holdings by households of securities issued by nonresidents were assumed to be adequately covered in Australian statistics.
6.67 Aggregate data for holdings abroad are compiled quarterly from entities (businesses, nonprofit organizations, and households) with holdings of at least $5 million (for nonprofit organizations the threshold was $0.5 million). Additionally, large companies (selected by various criteria) and households holding investment portfolios abroad of at least $0.5 million are required to report monthly and yearly, respectively, to the FECD.
6.68 Resident institutional investors, such as pension funds and insurance companies and, to a lesser extent, resident investment institutions, such as mutual funds and unit trusts, are likely to have direct holdings abroad. For institutional investors, total holdings of foreign shares and bonds are known from balance sheet information. In contrast to the figures derived from the custodian survey, they are not broken down by country. As a consequence, direct holdings may be estimated as the difference between these sources, but without a geographical breakdown. Additionally, the most important institutional investors are required to report their direct holdings annually, broken down by country of residence of the issuer. A similar method is used for investment institutions. For the nonfinancial sector, no information on direct holdings is available. It is assumed that households mainly prefer to use the services of resident custodians rather than invest directly abroad, and that the size of direct holdings abroad by households is, therefore, likely to be small.
VIII. Quality Control
6.69 Data received by the compiler should be verified because there is usually no prima facie assumption that respondents will provide the information on the basis required for the survey. Without good-quality data, the usefulness of the CPIS is seriously reduced. These checks can take a variety of forms. Following is an indication of some of the major verification checks that were used by Australia, Belgium, and Ireland (for the aggregate approach) and by Portugal (for the security-by-security approach). Many of the checks used are applicable to whichever approach is used to collect data for the survey. See also Chapter 5, Section VI.
A. Aggregate Approach
6.70Through survey forms. In Australia, the international investment reporting forms are structured so that there is a full reconciliation for each item between the opening and closing levels of investment and for transactions during the period as illustrated in Figure 6.2.
Figure 6.2:Reconciliation of Reporting of IIP and Balance of Payments Data in Australia
6.71 In the Australian experience, this approach both educates respondents about the relationships between individual data items and forces them to consider the consistency of the data reported. The Australian Bureau of Statistics (ABS) has found that full data on transactions and other reasons for changes in stocks are usually available when survey respondents provide data on their own accounts or client activity—for example, pension funds investing and managing their own assets, investment (fund) managers managing assets on behalf of other entities, and trading entities managing their own portfolios. These entities will collect all relevant stocks and transactions data at market prices.
6.72Through other reported data. For the nonfinancial corporations sector, independent sources (for instance, published stock or bond price indices or exchange rate movements over the period) are used to judge whether the reported data are accurate. For example, if a 20 percent change in the value of a U.K. asset in Australian dollar terms is explained by a negative entry in the “exchange rate variations” column (implying a decline in the value of sterling relative to the Australian dollar) and it is known that the bilateral rate had not moved in that direction or by that approximate magnitude over the period, then the ABS questions the respondent. Information from share price indices provides a useful check on entries in the “other factors”63 box, for equity. Anecdotal evidence, such as that derived from market or media comment or contact with other data providers, is also used to confirm data. For example, if there has been evidence that Australian enterprises in general were selling securities in a specific market, but a particular respondent reported a large increase in assets in that market owing to positive net transactions, the respondent is questioned. Comparisons are also made with the general magnitude, type, and direction of data reported by similar entities to see if they are behaving similarly—and providing consistent explanations for movements.
6.73Through analytical checks. Within the ABS, individual analysts are responsible for confirming the reported data. They produce editorial notes that identify and explain all large movements at the aggregate level by reference to the entities contributing significantly to those movements. The computer system edits and checks the arithmetic and compares large changes in levels with the “reconciliation” data at the individual respondent level. All large amounts (approximately $A 100 million or more) reported in the transactions, exchange rate variations, and “other factors” boxes are validated with respondents.
6.74Other edit checks. Survey outputs are verified through (i) data confrontation with results obtained from the ABS’s Survey of Financial Information, which collects data on institutional units’ balance sheets, transactions, and “other changes,” not just transactions and balances with nonresidents (the Survey of Financial Information does not provide geographical detail); (ii) an examination of the comparability of balance of payments and financial account balancing items; (iii) data confrontation with annual reports and Australian Stock Exchange reports (especially for dividends and reinvested earnings); and (d) checking details provided against press clippings and unit profiling checks.
6.75 The 1997 CPIS was the first such survey conducted in Belgium. The aggregate/combined end-investors and custodians approach was adopted.
6.76Units approached. The end-investors questioned comprised all institutional investors: credit institutions, investment enterprises, collective investment undertakings, holding companies, insurance enterprises, and private provident institutions as well as those other large enterprises that, according to their balance sheet data, had investments to the value of at least BF 200 million. Other end-investors, including the household sector, were not surveyed directly. Through the inclusion of custodians in the survey, securities issued by nonresidents and held by the household sector that had been deposited with resident custodians were incorporated in the results.
6.77Survey forms. There were separate forms for end-investors and custodians.
6.78 End-investors were required to make a return for each type of instrument, broken down by country of issuer and where held (with custodians abroad, with custodians in Belgium, or in own custody).
6.79 Custodians were required to complete a return, by type of instrument, of the securities that they held in custody for residents, distinguishing between securities held on behalf of other resident custodians and other residents. For the latter, this return was broken down by the country of resident of the issuer and where held.
6.80Checks on the data collected. Provision was made for checks at the level of the individual documents, individual respondents, sectors of respondent, and overall. Checks at the level of individual documents consisted of mathematical checks on the accuracy of the forms completed. Checks at the level of individual respondents were made to exclude direct investment (by comparison with the “direct investment” survey that was conducted at the same time). The data on own portfolio investment reported by each credit institution were individually compared with the asset and liability statements that were filed with the National Bank of Belgium by these institutions for other purposes. The latter sources were also broken down by instrument and by the country of residence of the issuer.
6.81 An examination was carried out by custodian and by instrument to check the value of the securities that the end-investors declared that they had deposited with the custodian and the value that the custodian itself declared that it had received from resident end-investors. The latter value must at least be equal to the former. A positive balance was subsequently assigned to the end-investors that had not been surveyed (mainly the household sector).
6.82 A global sectoral check was also carried out for the following four categories of end-investors for which balance sheet data were available in time within the Belgian National Bank: credit institutions, collective investment undertakings, insurance enterprises, and private provident institutions. For investment enterprises, holding companies, and other large enterprises, there were no similar checks owing to the lack of comparably structured material. With a very high response rate (99.7 percent), the global survey results, including the estimated additions, were then compared, instrument by instrument, with the IIP.
6.83 With regard to quality, it was quite frequently necessary to contact respondents to ensure that the forms were completed correctly. The many individual and global checks built into the survey design and data processing system enabled the quality of the data to be ensured. For credit institutions (the largest sector), the checks were based on individual comparisons, and no noteworthy deviations were found.
6.84 Although the quality of the data collected was regarded as satisfactory, the main areas of concern were the correct attribution of market values for the reference date, and the inability of custodians to adjust the value of the securities placed in custody to exclude repos.
6.85 From a comparison of CPIS with IIP data, it was apparent that the CPIS data were lower, by some 15 percent, than the data for holdings of cross-border portfolio investment in the IIP statement. The larger figure in the IIP statement was attributed to the inclusion of cumulative balance of payment flows, for which a breakdown by the country of residence of the holder could not be provided. These flows were believed to include holdings of euro bonds and smaller enterprises that are retained in self-custody or with nonresident custodians. Custodians are not able to adjust the value of the securities placed in custody.
6.86 When compiling the positions data for the 1997 CPIS and other data requirements relating to portfolio investment, the Central Statistical Office (CSO) used a number of sources to validate the data for both balance of payments and IIP purposes. No new collection system was needed because the existing collections already required reporting by instrument and by country of residence of the issuer. Data were reported according to these headings: opening position, transactions increases, transactions decreases, valuation, other changes, and closing position, as well as residence of issuer.
6.87 The survey collects balance sheet data by functional category of the balance of payments (direct investment, portfolio investment, other investment). In addition to the balance of payments and the international investment position data, holdings by residents of Ireland are also collected (to provide a fuller and more comprehensive set of data).
6.88 This approach to data collection allowed a higher level of quality control on the data supplied by respondents. A primary quality check ensured that the balance sheet balances through the identity: equity capital (plus all reserves) equals total assets less total liabilities excluding equity. This check was applied to both the positions at the beginning and the end of the period. Interperiod checks at the level of instrument and country, by respondent, were also carried out (e.g., that the closing position in Bonds & Notes for the United Kingdom in period 1 was the same as the opening position for Bonds & Notes in the United Kingdom in period 2).
6.89 The second level of validation checks was applied at the survey level. Ten surveys cover different sectors of the economy, and five are particularly significant in terms of portfolio investment. These were (i) Portfolio Investment by Institutional Investors, (ii) Non-IFSC (International Financial Services Centre) Banks, (iii) IFSC Banks, (iv) Portfolio Investment by Irish Registered Collective Investment schemes at IFSC, and (v) Life and Non-Life Insurance Companies at IFSC. The CSO matched the survey results with available independent sources of data. The surveys are described below.
(i) Portfolio Investment by Institutional Investors. Institutional investors report position data in this survey. The CSO validates these survey data using the data produced annually by the Irish Association of Investment Managers (IAIM) on portfolio investment by residents. The CSO compares the aggregate and individual company data reported by IAIM with the CPIS data. The main strength of the data is the detail for individual holders. The results of the validation give an indication of the quality of the data, provide a basis for correcting for undercoverage or nonresponse in the CSO survey, and suggest areas for further investigation.
(ii) Non-IFSC Bank and (iii) IFSC Banks. These data are checked against Central Bank of Ireland data for the banking sector, which includes data on banks’ cross-border holdings of portfolio investment.
(iv) Portfolio Investment by Irish Registered Collective Investment Schemes at the IFSC. In the IFSC in Dublin, a large number of collective investment schemes invest predominantly in securities issued by nonresidents and whose units are held almost entirely by nonresident investors. Those collective investment schemes that are legally domiciled in Ireland are treated as resident for the purposes of the CPIS. The CSO validated the data against the summary data collected by the Central Bank of Ireland for regulatory purposes.
(v)Life and Non-Life Insurance Companies at the IFSC. The CSO uses data collected by the insurance regulator (the Department of Enterprise and Employment) to validate the data reported to the CSO. The analysis of the assets of life and general insurance companies is comprehensive and includes instrument detail.
B. Security-by-Security Approach
6.90 Through security-by-security forms, respondents were asked to provide the security identification number, quantity of securities held, and face or nominal value of each security. Methodological explanations and definitions, in line with those in the 1997 Survey Guide, were included in the instructions. Checks were also made to ensure that reported data contained an ISIN code, and that the information was consistent with the Ufficio Italiano dei Cambi (UIC) securities database.
6.91 For banks, data reported in the CPIS survey were compared with information reported in money and banking surveys. For investment companies, data reported in the CPIS survey were checked against data provided by the Lisbon Stock Exchange.
6.92 In the case of some nonfinancial corporations, the accounting reports were used to check the reported information.
6.93 Using a breakdown of portfolio assets by the currency in which issues are denominated, a reconciliation between flows and stocks was performed
6.94 In the event of misreporting (such as of market values or units of values), respondents were asked to provide corrected values. For some banks, the reported value was replaced by that derived from money and banking statistics sources.
6.95 A comparison between the investment funds (aggregate values) and the data provided by the Lisbon Stock Exchange was made. Checks were also made at the aggregate and detailed level of data reported by each custodian. Double counting was minimized by the fact that the CPIS was conducted primarily on a custodian basis, and the requirement that end-investors reported only their holdings of securities held in self-custody or with nonresident custodians.
IX. Steps Taken to Address Low Coverage or Low Response Rates
6.96 In some instances, there may be a need to gross up results, either because the survey did not cover all units in the potential universe or because of a low response rate. The latter problem often arises in situations where a survey is not obligatory. The experiences of Bermuda, Belgium (for coverage of households), and the United Kingdom in the 1997 CPIS are summarized below.
6.97 Because the CPIS was conducted on a voluntary basis, there was extensive reliance on estimation procedures to account for assets held by nonrespondents. The techniques used differed, depending on the sector classification of respondents and the available information. For collective investment schemes and insurance companies, supervisory sources provided some indication of cross-border holdings of portfolio investment, but did not provide a breakdown by the country of residence of the issuer. The CPIS data provided the latter information, but there was a substantial element of nonresponse. The totals reported by each sector were aggregated, and a calculation was made for the securities held in each country as a percentage of the total assets reported. It was assumed that these percentages also held for the assets of nonrespondents. The respective percentages were applied to the totals reported for supervisory purposes to derive an estimate for portfolio investment in each country. These estimation techniques assumed that the breakdown by the country of residence of the issuer, as reported by CPIS respondents in that sector, was representative of all companies within that sector (i.e., including those that did not complete the CPIS questionnaire).
6.98 Bearer securities are important in Belgium and are frequently held with nonresident custodians or in self-custody. In order to mitigate this gap, an attempt was made to find other sources offering sufficient guarantees of accuracy and reliability. Where possible, these data were wholly or partly incorporated in the survey results.
6.99Table 6.2 summarizes the survey results prior to estimated additions. The figures in the households column result from the difference between the value of the securities that custodians declared to have received from residents and the amounts that end-investors declared to have entrusted to resident custodians. The difference is attributed to the household sector.
|Debt securities, <1 year||288||226||110||85||398||311|
|Debt securities, >1 year||2,606||3,046||392||492||2,998||3,538|
6.100 Adjustments based on global comparisons with other sources could not be used owing to their limitations, such as the choice of breakdown, and the level of detail. Adjustments based on information obtained on Belgian holdings of securities issued in Luxembourg seemed to be very reliable and contained the necessary breakdowns for the survey. These data were added only where it was sufficiently certain that the securities concerned were held by residents. The adjustments were confined to units of a société d’investissement à capital variable (SICAVs), insurance notes, and bank cash certificates of Luxembourg issuers, which, according to external sources, are the property of Belgian residents and were added to the basic holdings.
6.101 Units of SICAVs were added to the equity securities issued by Luxembourg issuers insofar as they were not yet included in the declarations of end-investors and custodians, on the assumption that all equity securities mentioned under the country “Luxembourg” were SICAVs. This represented a very cautious approximation, but is clearly an underestimate because the survey results presumably also include investments in Luxembourg shares other than SICAVs.
6.102 Insurance bonds and bank cash certificates were added in full to the long-term debt instruments issued by Luxembourg issuers, because these, unlike SICAVs, always give entitlement to coupons and will consequently, for tax reasons, be less quickly deposited in custody with resident custodians.
6.103 The share of the household sector in the CPIS results was about 40 percent of the adjusted total. The underlying figure could well be higher, on the assumption that cross-border holdings of portfolio investment assets compiled on the basis of cumulative flows, as reported in the IIP, mostly comprises portfolio investment by households.
6.104 An aggregate end-investor survey was conducted for banks by the Bank of England and a separate survey for all other institutions by the ONS. For nonbanks, the CPIS was conducted on a statutory basis. However, as this was a one-off survey, it was not possible to force any respondent to respond. In the event of nonresponse, respondents were reminded by telephone and written reminder. For banks, a specific CPIS survey form was not required because existing survey forms were used. Nonresponse was not a problem for banks.
6.105 In the ONS survey, nonrespondents were assumed to have a pattern of holdings that was the same for those that did respond and therefore were included in the grossing up. The method of grossing was ratio estimation, using the same auxiliary value as used for the other surveys to financial institutions. Since the ONS had separate samples for insurance companies, pension funds, trusts, securities dealers, and nonfinancial companies, grossing up was done for each group separately.
X. Sectoral Information
6.106 Sectoral information about the holder is valuable additional information, but it is only an encouraged item in the CPIS. Information regarding the sector of the holder can be very useful, especially for counterpart liability analysis, because holders of portfolio investments may behave in different ways, especially during periods of financial crisis. However, obtaining this information is not always easy. Sector of holder is straightforward for an end-investor survey (the sectoral information is self-defining), but for a custodian survey it may be very difficult and expensive for the custodian to identify his or her customers’ sectors. However, the experience in this respect of Portugal and the United States is useful in having custodians identify the sector for which they hold the securities.
6.107 Identification of the sector of issuer is usually straightforward when using a security-by-security approach.
6.108 The CPIS included appropriate fields in order to obtain the institutional sector of the holder and the issuer. The latter information was provided by custodians, and the former obtained indirectly through other data available to the Banco de Portugal.
6.109 Sectoral information was collected with respect to both the sector of the domestic investor and the sector of the issuer. For the holder, custodians were asked to report record-by-record which of the following sectors they were reporting for: (i) own portfolio, (ii) custody for mutual fund, (iii) custody for pension fund, (iv) custody for insurance company, (v) custody for other. However, determining the sector for many investments proved to be difficult.
Parties at interest are required to file a report with the ISA.
An indication of the problem may be persistently large negative net errors and omissions in the balance of payments. Although there are many reasons for errors and omissions, and a low figure may only indicate offsetting errors and omissions, if the number is large and has been growing strongly since any liberalization of capital markets, there is a possibility that measured outflows in portfolio investment are undercounted. In some countries, estimates of cross-border portfolio investment are made on the basis of cumulative flows—but not included in the CPIS because there is no breakdown by the country of residence of the issuer.
On the ABS form.