International Technical Meeting on Measuring Remittances World Bank, Washington DC, June 11-12, 2009 Improving the quality of remittance statistics through the analysis of bilateral asymmetries by Giovanni Giuseppe Ortolani (Banca d Italia) Evis Rucaj (World Bank) 1 Summary In the context of current initiatives for the improvement of the quality of statistics on remittances, the paper discusses the potentialities of the analysis of bilateral asymmetries, i.e. differences between the figures reported by remittance sending and recipient countries. To this end, the note highlights the current high and increasing level of global discrepancies on remittances, both in absolute and relative terms, as measured in the world balance of payments. Subsequently, as a case study, it reports about a bilateral comparison exercise on the Italy-Albania remittance corridor. The paper concludes that the analysis of asymmetries, through the in-depth analysis of figures and data collection methodologies, carried out cooperatively by the national compilers involved, possibly with the support of international agencies, can contribute to improve the overall quality of statistics on remittances. 1. Introduction Policy and analytical interest on remittances 2 has risen dramatically since the G8 Heads of State meeting at Sea Island in 2004, which stressed their importance as a source of external financing for 1 The views expressed in this paper are those of the authors and do not necessarily reflect those of the Bank of Italy or the World Bank. Although the paper should be considered the outcome of the two authors combined efforts, paragraphs 1-3 have been drafted by G.G. Ortolani and par. 4-5 by E. Rucaj.
developing countries. Following the subsequent meeting of G7 Finance Ministers, which prompted the review of the quality of existing remittance data and the development of a work program to make improvements, the community of balance of payments (BoP) compilers has had to pay increasing attention to remittance-related items, 3 until then fairly neglected. The call from G7 countries resulted in three important outcomes: the improvement of concepts and definitions of remittances in the balance of payments framework, with the elaboration of the four new remittance-related concepts, 4 implemented in the new edition of the BoP Manual (BPM6); 5 the publication of a specific compilation guide, authored by the Luxembourg Group on Remittances, to assist countries about data collection issues related to remittances; 6 the overall improvement of the quality of remittance data through the revision of data collection and compilation methods in many countries. Currently, priority is given to the third point, i.e. the need to pursue international efforts to improve remittances data quality. Following the indications G8 Summit in Hokkaido Toyako of July 2008, the World Bank is coordinating a new Global Remittances Working Group (GRWG), whose mandate includes improving data as one of its four work sub-themes. This paper deals with one of the possible strategies, apparently not sufficiently exploited so far, to improve data quality of remittances,. It is the analysis of bilateral asymmetries, i.e. the discrepancy between the figures of the remittance sending countries vis-à-vis those of the remittance recipient countries, carried out cooperatively by compilers of the two countries involved (so called bottom-up approach). 2 In the most typical form, remittances are the transfers of funds, without a quid pro quo, carried out by migrants, while residing in the new economies, to physical persons - usually their relatives - resident in the economy of origin of the migrants themselves. 3 The most important are: workers remittances and compensation of employees. Workers remittances are broadly the transfers of funds mentioned in previous footnote. Compensation of employees are the wages and salaries paid to shortterm workers (working for less than one year) non resident in the economy where they are employed. Some analysts also include migrants transfers, which arise when individuals move their residence from one economy to another; they essentially cover the movement of personal effects and financial assets between countries and changes in stock position of personal investments and debt arising from the change in residence status; however, considering that this third item is less directly related to remittance flows, is of minor importance in terms of size and is lacking comparable and reliable statistics, it will not be taken into account in the rest of the paper. 4 Personal transfers, personal remittances, total remittances, total remittances and transfers to Non Profit Institutions Serving Households. 5 See IMF(2008a). 6 See IMF(2008c). 2
The paper is structured as follows. Paragraph 2 discusses BoP asymmetries in general, paragraph 3 explains why analysis of asymmetries in remittances may be useful, paragraph 4 gives some evidence from an ad-hoc comparison of bilateral data on the Italy-Albania corridor and paragraph 5 provides some conclusive remarks. 2. Asymmetries in balance of payments A typical feature of BoP statistics, in common with other domains where a country of origin/destination breakdown is meaningful (e.g. foreign trade, migration statistics), is that two measures are in principle available for a given phenomenon under observation. For example, the sale of a service by country A to country B is measured both from country A, as an export (BOP credits), and from country B, as an import (BOP debits). The two measures are rarely identical or even similar. Following Eurostat (2006), this is caused by differences in the two countries as regards: o data collection system (sources, methods, etc.) o classification of transactions in the various BoP items o time of recording of transactions o identification of the country of residence of the counterpart o treatment of complex transactions, e.g. when more than two transactors are involved. At the world level this implies that global BOP credits are different from global BOP debits. This is usually indicated as a global discrepancy and the difference between the figures of the two countries involved is called bilateral asymmetry. Large asymmetries legitimately raise doubts about the quality - the accuracy, using current terminology or even the credibility, of the underlying statistics. Therefore, it is normally considered appropriate to find ways to reduce these discrepancies. Since the beginning of the decade, much work has been carried out in Europe, in the analysis of bilateral asymmetries in balance of payments. 7 This is largely due to the fact that, in this region of the world, a consolidated at the EU or euro area level - BOP is legally needed and it is produced 7 See Eurostat (2002). 3
through consolidation of individual EU Member States BOPs; the evidence of large asymmetries seriously hampers the confidence about the quality of consolidated economic statements. Two approaches can be followed to reduce asymmetries: - the top-down approach, which consists in transforming original figures with a mathematical / statistical model, more or less sophisticated, e.g. constrained optimization; with this method, normally applied by the central region institutions (e.g. Eurostat or the European Central Bank), the emphasis is about finding the right model; - the bottom-up approach, which consists in the compilers of the two countries involved comparing the respective data collection methods, definitions and, sometimes, microdata and agreeing a correction of one or both country(ies) statistics; with this method the emphasis is about analysing the situation and then negotiating the best bilateral figures. In Europe there have been positive experiences with both approaches. None of them has clearly prevailed in terms of consensus. The top-down approach is generally regarded as cost-effective, but it is often criticised because it artificially changes individual country figures, causing a loss of coherence between countries and regional figures. Conversely, the bottom-up approach is appreciated because appears more evidence-based and does not imply to lose consistency between national and European aggregates; however, it is regarded as time consuming and resource intensive. This paper, consistently with the objective of improving the quality of remittance data, favours the bottom-up approach. The main rationale behind this preference is that asymmetries for remittances are currently so large in magnitude, as it will be illustrated in next paragraph, that a top-down algorithm would be totally inappropriate (as it would mainly produce fake figures). Secondly, the bottom-up approach tends to correct bilateral discrepancies in a permanent way, as it try to correct the problem at its source, i.e. concepts, definitions, data collection and compilation methods. Thirdly, the bottom-up approach appears of more general use, as the top-down approach seems more suitable for highly formalised economic regions (such as the European Union or the euro zone) where consolidated macroeconomic statements (BoP, National Accounts, etc.) are mandatory. 4
3. Why analysing asymmetries on remittances? This paragraph tries to explain the reasons why analysing bilateral asymmetries can help the improvement of the quality of statistics on remittances. First, it should be mentioned that remittances are inherently and notoriously difficult to measure, mainly because of three factors: - the huge quantity of small transactions that characterise them, carried out by a large number of transactors, often through complex payment schemes; - the relevant role played by the so-called informal remittance service providers, unauthorized and/or unregistered entities, acting along with, to a large extent in competition with, the formal providers - money transfer operators, banks, etc. - officially authorized to operate in the money transfer business; - the unavailability of accurate statistics on the population of migrants, because of the relevant presence of hard-to-count subjects (illegal immigrants, refugees, etc.), which makes the measure of the phenomenon through sample household surveys particularly thorny. Therefore, the analysis of asymmetries is a useful complementary tool to monitor data quality and can represent an important driver to improve countries data collection systems. Second, the evidence shows that asymmetries are currently very large at the global level. As reported in Table 1, at the world level, the IMF, for the year 2007, estimated 336.3 billion USD as the total workers remittances (WR) and compensation of employees (CoE) credits; however, the mirror debits are far from being equal, being them only 249.7 billion USD. The global discrepancy, or absolute asymmetry is therefore 87 billion USD, or 29.6% in relative terms. 8 The relative asymmetry is even higher for the individual components, as they have opposite signs; the asymmetry for CoE is -14.1%, whereas that of WR is positive, with a staggering +54.5%. The table also witnesses the fast growth pace of the phenomenon. However, and this of course still the effect of the discrepancy, the data give users two truths about the average annual growth rate of total CoE + WR over the period 2001-2007: 18% if measured from the credit side and a still respectable but significantly lower 13.2% from the debits side; credits appear to have increased three times during the period, whereas total debits only seem to have increased twice. 8 Relative asymmetry, in line with common use, is calculated by dividing the absolute asymmetry figure, with its sign, by the average of gross flow figures (credits and debits). 5
Table 1 - World balance of payments. Items "compensation of employees" and "workers remittances". Amounts, absolute and relative asymmetries. 2001-2007. 2001 2002 2003 2004 2005 2006 2007 Average Annual Growth Rate (2001-2007) Compensation of employees Credit - Amount (million USD) C 43,517 49,475 60,096 70,714 76,460 82,602 98,934 Credit - % change on prev. year 5.4% 13.7% 21.5% 17.7% 8.1% 8.0% 19.8% 14.7% Debit - Amount (million USD) D 51,030 57,312 67,926 76,463 84,883 95,144 113,975 Debit - % change on prev. year 7.0% 12.3% 18.5% 12.6% 11.0% 12.1% 19.8% 14.3% Absolute asymmetry AA=C-D -7,513-7,837-7,830-5,749-8,423-12,542-15,041 Relative asymmetry RA=AA/((C+D)/2) -15.9% -14.7% -12.2% -7.8% -10.4% -14.1% -14.1% Workers remittances Credit - Amount (million USD) C 81,168 93,410 112,693 126,007 153,009 180,715 237,396 Credit - % change on prev. year 12.3% 15.1% 20.6% 11.8% 21.4% 18.1% 31.4% 19.6% Debit - Amount (million USD) D 67,612 77,829 81,465 91,780 99,853 117,259 135,748 Debit - % change on prev. year 8.9% 15.1% 4.7% 12.7% 8.8% 17.4% 15.8% 12.3% Absolute asymmetry AA=C-D 13,556 15,581 31,228 34,227 53,156 63,456 101,648 Relative asymmetry RA=AA/((C+D)/2) 18.2% 18.2% 32.2% 31.4% 42.0% 42.6% 54.5% Sum of compensation of employees and workers remittances Credit - Amount (million USD) C 124,686 142,885 172,789 196,721 229,469 263,318 336,330 Credit - % change on prev. year 9.8% 14.6% 20.9% 13.9% 16.6% 14.8% 27.7% 18.0% Debit - Amount (million USD) D 118,642 135,141 149,391 168,242 184,736 212,403 249,722 Debit - % change on prev. year 8.1% 13.9% 10.5% 12.6% 9.8% 15.0% 17.6% 13.2% Absolute asymmetry AA=C-D 6,044 7,744 23,398 28,479 44,733 50,915 86,608 Relative asymmetry RA=AA/((C+D)/2) 5.0% 5.6% 14.5% 15.6% 21.6% 21.4% 29.6% Source: Author' elaboration from data in IMF (2008c). 6
Table 2 - World balance of payments. Global discrepancies by main current account aggregates. Amounts (in billion USD), absolute and relative asymmetry. 2007. Credit Debit Absolute asymmetry Relative asymmetry CURRENT ACCOUNT 22,041 21,724 317 1.4% GOODS 13,867 13,656 211 1.5% SERVICES 3,386 3,218 168 5.1% of which: - Transportation 741 891-150 -18.4% - Travel 850 767 83 10.3% INCOME 3,915 3,986-71 -1.8% of which: - Compensation of employees 99 114-15 -14.1% CURRENT TRANSFERS 873 864 9 1.0% of which: - Workers' remittances 237 136 102 54.5% Source: Authors' elaboration of IMF (2008b). 7
An additional worrying aspect is the clear tendency of asymmetries of WR to increase over time with a steep trend (the relative asymmetry was only 5.0% in 2001). Moreover, as it is often the case for BoP asymmetries, the discrepancies are structural, e.g. they are consistently negative for CoE and positive for WR. The asymmetry for remittances appears serious even in comparison with other BoP current account items. As shown in Table 2, the global discrepancy in 2007 is quite small for the account as a whole and for its main headings ( goods, services, income, current transfers ), with values in the range -2% to +2%. Transportation and travel represent an exception, respectively with -18% and +10%. While CoE asymmetry is at a similar level, the discrepancy for WR looks quite exceptional. Similar evidence is found for the European Union. Eurostat (2007) reports that for both 2005 and 2006, intra-eu WR credits were more than twice intra-eu debits. Intra-EU asymmetries were 8.5 billion or 55% of recorded credits in 2006, partly compensated by net CoE - calculated as CoE minus taxes, social contributions, transport and travel expenditures related to short-term employment whose intra-eu asymmetry was - 3.2 billion or 26% of recorded credits. Apparently, at present there is no clear consensus on a set of methodological explanations for the systematic positive asymmetry on WR. However, as noted in World Bank (2007), it may partially result from the fact that net remittance receiving countries have recently invested more resources into improving data than net sending countries, assuming that this more intense effort leads to an increase of measured remittances. A stronger commitment about data quality in top net recipient countries, especially where inflows represent a large part of the country s GDP, seems both legitimate and rationale, given the effects of remittances on macroeconomic aggregates. 4. Bilateral comparison on the Italy-Albania remittance corridor The objective of this paragraph is to provide hints about the evidence that the bottom-up approach to bilateral asymmetries can bring to compilers. To this end, the Italy-Albania remittance corridor was chosen as a case study. The exercise is absolutely not meant to represent an exhaustive study of the topic, which would deserve resources far beyond the possibility of this research. The analysis is limited to workers remittances, as CoE is far less significant in the case considered. Therefore, in the exercise remittances and workers remittances are used as synonyms. 8
We start providing a few descriptive notes about the corridor. Italy, after having been a net remittance-recipient country until the end of the Nineties, is at present one of the main remittancesender countries worldwide, with 6.4 billion of remittance outflows in 2008. In terms of value of formal remittance transfers, Albania ranks at the ninth place in the list of destination countries. Workers remittances are vital to Albanians and they have turned into a typical phenomenon in their economy. Over the last years, remittances have not only been considered as a source of income for nonproductive purposes or as a way to meet the current vital needs, which relate to the improvement of living conditions, but they have also been a financial source for productive purposes, in the form of short and long-term investment. It is estimated that, out of a total population of 3.2 million people in Albania, 1 million migrated to Greece and Italy (Lianos and Glytsos, 2004). According to Bank of Albania (BoA), remittances inflows are estimated as 952 million in 2007 and represent 12 percent of Albania s GDP. Remittances total value is twice the inflows of Foreign Direct Investment to the Albanian economy. Approximately 43% of remittance inflows were from Italy, 42% from Greece and the rest from the UK and the US. These inflows covered the high trade deficit by 45%, leading to the improvement of the current account and, consequently, the international investment position. The Italy-Albania corridor is dominated by the physical transfer of cash. The bulk of remittances to Albania were channeled through informal channel. The BoA estimates that, almost 60% of the total remittances received from Italy are in cash and take place through informal channels. The remaining 40% of transfers take place mainly through money transfer companies and to a limited extent by banks. Albanian migrants residing in Greece and Italy largely prefer to resort to informal channels partly because of the geographic vicinity between Albania and these countries. The application of the bottom-up approach requires a preliminary consideration of the data collection method of the two countries. A short summary is provided in the following. In Italy, statistics on remittances, in the framework of BoP, are produced by the central bank. Since 2006, data are compiled on the basis of direct reports of Money Transfer Operators (MTOs). This source replaced the information from the International Transaction Reporting system, heavily 9
biased both in the level and in the geographical breakdown. 9 The survey, in principle, collects data on person-to-person cross-border transfers carried out by all MTOs resident in Italy (around 30 agents). Data, reported monthly, concern the number and the amount of transfers, from Italy to abroad and vice versa. For outflows data are broken down by Italian province of residence of the sender and country of destination of the recipient. Despite the collection of data from MTOs has significantly improved the quality of statistics on remittances, the system bears the following three main weaknesses: - remittance through channels different from MTOs are not covered; in particular, no information is collected about informal channels. No official information is available about the share of the various remittance channels; however, there is clear evidence that MTOs currently account in Italy for the a very high part of formal outward remittances. There is consensus among Italian analysts that informal remittances represent an important component; - despite MTOs are instructed to report only person-to-person transfers, business-to-persons or business-to-business payments can sometimes be mistakenly included; - MTOs cannot always reliably exclude transfers of short-term workers (non-bop) from those of immigrants. In the mid term it is planned to investigate the possibility to improve the coverage of the informal channel through a household survey, either setting up a specifically designed survey or adapting the existing ones. 10 In Albania, BoA is the responsible agency for compiling remittances data as part of BoP statistics. Given the economic profile of the country, where cash is used extensively as a means of payment, and its migration pattern, BoA uses different data sources in order to estimate these flows. Three basic sources are used: the inflow-outflow model, the Banking system & MTO data source and a Household survey. The model basically estimates cash remittances through informal channels along with the information provided by the formal ones (banks and MTOs). The Household survey is a new source of information on remittances. Through the survey, BoA provides information on the amount, origin, frequency and destinations of money transfers (remittances) from Albanian migrants abroad to their family back home. The survey sample is representative of the Albanian territory, which makes it a reliable and timely instrument about family remittances. This nationwide 9 The ITRS system was mainly hindered by the rapid growth of the MTO business, and the related multinational clearing systems, and by the introduction of statistical exemptions for transactions below the threshold of 12.500 euros. For further details on the Italian system for remittances, see Ortolani (2007). 10 As a pilot exercise, some questions on remittances have been added to the existing Bank of Italy s Survey on Household Income and Wealth; results will be available at the beginning of 2010. 10
survey of beneficiary households focuses on different aspects about the remittance peculiarities. It includes a range of dimensions in order to estimate the annual remittances flows and other characteristics of remittances as frequency, use of formal and informal channels and the trend of these flows in Albania. Although the combination of the data sources makes the estimation somehow reliable, the estimation presents many potential weaknesses. Given the large scale of Albania s informal economy, unaccounted inflows may be either money remitted by migrants abroad or any other type of flow originating from informal economic activities. Hence, the residual attributed to cash remittances could actually account for other unrecorded transactions. Moreover, there are several concerns related to the technical and the qualitative issues of the survey. Starting now a brief comparison of bilateral figures, Table 3 shows that asymmetries are very large. Albania s figures are consistently and significantly higher, as an average during the period 2005-2008 around three times, than those reported by Italy. In 2008, the Bank of Albania (BoA) reports inward remittances from Italy for 350 million, whereas the Bank of Italy (BoI) estimates the outward outflows to Albania as 143 million. This results in an absolute asymmetry of 207 million and 84% in relative terms. The main explanation is that Albania includes an estimate of informal transfers, whereas Italy is not. However, a more in-depth analysis should investigate also the possible effects of the mentioned methodological caveats of the two countries methodologies. Table 3 Workers remittances flow from Italy to Albania. Amounts in euro million. 2005 2006 2007 2008 Source: Bank of Albania 372 429 411 350 Source: Bank of Italy 119 139 144 143 Absolute asymmetry 253 290 267 207 Relative asymmetry 103% 102% 96% 84% 5. Conclusive remarks The exercise on the Italy-Albania corridor has shown how even a rather superficial analysis of bilateral figures and partner countries methodologies can lead to the identification of the main source of discrepancy in the two countries measures: the different way in which informal remittances where treated (no coverage at all in Italy, estimated through a model and a household survey in Albania). 11
A fully-fledged bilateral comparison, as said, requires a time and resource-intensive work of the two partners. The first phase of the work is the in-depth analysis of the respective countries concepts, definitions, data collection and compilation methods. The second phase consists in undertaking actions to harmonise countries theories and practices, in order to agree on more close estimates of bilateral flows. The effort can be capitalised, as problems tend to be solved on a permanent basis and asymmetries are consequently avoided also in the future. An additional advantage of the exercise, is the that it reinforces countries co-operative attitude, giving the opportunity to exchange views and best practices. We then conclude the paper providing some suggestions about possible future actions at the national and the international level to reduce remittance asymmetries. 1. The first, obvious, prerequisite to tackle asymmetries is that remittances data geographically disaggregated are available to compilers. Even more obviously, this requires that countries endeavour to collect and publish geographically broken down remittance figures to the largest possible extent. 2. These data should then be efficiently disseminated. Outside the European Union, which has recently started collecting and circulating remittances broken down geographically by main destination country, this does not appear the general rule. Hence, it seems desirable that a worldwide international agency (e.g. the IMF or the World Bank) acts as a data-hub for remittance statistics, systematically collecting them from countries and disseminating to compilers, building and maintaining a full data matrix of bilateral flows. This would be an important step forward, as it will save countries the burden to periodically exchange figures with partners. For the sake of transparency, with the consent of compiling countries, the matrix could also be made available to the general public, to allow analysts to evaluate the quality of data. 3. Despite the matrix should be as exhaustive as possible, the bilateral work of country pairs should begin addressing main corridors (Table 4, which lists the main world corridors according to a OECD 2006 study, may provide hints in this respect) or country pairs with largest bilateral asymmetries, if bilateral data are already available. 12
Table 4 - Remittances from OECD donor countries to the developing world 2002 2005. Top ten remittance corridors. Sending country Receiving country Amount (billion USD) United States Mexico 61.70 United Kingdom India 6.68 Canada China 6.36 Japan Brazil 6.27 Spain Colombia 5.78 France Morocco 5.73 Germany Turkey 5.40 Italy China 3.75 Australia Lebanon 3.16 Netherlands Turkey 1.76 Source: Authors elaboration from OECD (2006). 4. International agencies, possibly the same that manages the data-hub, and developed countries should encourage and help the relevant developing countries undertaking the bilateral comparisons, considering also the provision of technical and financial support when needed. References Bank of Albania (2006): Remittances: Albanian Experience. Presented at the first meeting of the Luxembourg Group on Remittances, June 2006 Brown Stuart (2004), Statistics on Remittances, Seventeenth Meeting of the IMF Committee on Balance of Payments Statistics, Pretoria, 26 29 October, 2004, BOPCOM-04/39. Eurostat (2002): Evolution of the Intra-EU Asymmetries in the Current Account, Fifteenth Meeting of the IMF Committee on Balance of Payments Statistics, Canberra, Australia, October 21 25, 2002, BOPCOM-02/18. Eurostat (2006): Asymmetries in the EU current account data, Working Papers and Studies, Eurostat, 2006. Eurostat (2007): Remittance flows to and from the EU, Methodologies and working papers, Eurostat, 2007. International Monetary Fund (1993): Balance of Payments Manual, fifth edition, International Monetary Fund, 1993. International Monetary Fund (2008a): Balance of Payments Manual, sixth edition, International Monetary Fund, 2008. International Monetary Fund (2008b): Balance of Payments Statistics Yearbook 2008, 2008. 13
International Monetary Fund (2008c): International Transactions in Remittances: Guide for Compilers and Users, December 2008. International Monetary Fund and Committee on Balance of Payments Statistics (1997): Comparison of Bilateral Balance of Payments Between Portugal and Germany, prepared by Beatrice Timmermann, Balance of Payments Statistics Division, Deutsche Bundesbank, 1997. Lianos T., Glytsos N. (2004): Remittances in Europe, in Wilson S., Beyond Small Change: Making Migrants Remittances Count, Inter-American Development Bank, Washington DC, 2004. OECD (2006): 2005 Development Co-operation Report, Vol. 7, No. 1, Paris, 2006. Ortolani Giovanni Giuseppe (2007), Measuring Remittances: The Experience Of Italy, Meeting of the Eurostat Balance of Payments Working Group, 22 23 March 2007. Ratha Dilip, Shaw William (2007): South-South Migration and Remittances, World Bank Working Paper No. 102, 2007. Reinke, J. (2006): Remittances in the Balance of Payments Framework: Current Problems and Forthcoming Improvements, Conference of the International Association for Official Statistics, 6-8 September 2006, Ottawa. The Suitland Working Group (2009): Using Household Surveys to Measure Migration and the Size, Distribution and Characteristics of Migrant Population, March 16-17, 2009. Final Conference Report. World Bank (2006): The Italy Albania Remittance Corridor. Shifting from the Physical Transfer of Cash to a Formal Money Transfer System. Presented at the Conference on Remittances: An Opportunity for Growth, Bari, Italy, March 3-4, 2006. World Bank (2007): International Working Group on Improving Data on Remittances, Final Report, June 2007. 14