| plugins | |
|---|---|
|
Having examined anomalies in the current account (trade mispricing) in the previous chapter, we now turn to approaches to estimation of illicit financial flows (as some of the authors do, we also use the term illicit financial flows and capital flight interchangeably) that make use of anomalies in the capital account (unrecorded capital movements). We address three capital account-based approaches: of GFI; of Ndikumana and Boyce; and of Henry, the first two of which combine trade-based and capital account components.
While GFI and Ndikumana and Boyce estimate illicit financial flows, Henry’s estimates focus on the stock of wealth held offshore and to that aim he aggregates the estimates of flows. Indeed, how much wealth is held offshore and how much of it is illicit is another research question related to estimates of scale of illicit financial flows. Other than Henry’s estimates of offshore wealth (in 2012 in particular), Zucman, in 2013 and in his follow up estimates with co-authors, Alstadsaeter, Johannesen, & Zucman (2018), has produced influential estimates of wealth held in tax havens. Although methodologically different, we include these offshore wealth estimates in this chapter – also because Henry’s estimates methodologically overlap with the GFI and Ndikumana and Boyce.
Before Henry (2012) and Zucman (2013), some related research with the ambition to provide global estimates of offshore wealth or illicit financial flows was linked with development implications of tax havens and motivated by tax revenues not collected due to illicit activities that might be used to invest in social policy programmes in poor countries. A number of studies, mostly by non-governmental organizations and some academics, emerged around the year 2000 and provided some of the first estimates of assets held offshore and associated illicit financial flows and government tax revenue losses relevant for poor countries, using various methodologies. Oxfam (2000) estimated that poor countries suffered a yearly loss of around USD 50 billion due to tax havens, whereas Transparency International (2004) estimated that corrupt heads of states are responsible for billions of dollars in illicit financial flows out of their countries.
Quite a few other studies focus on what assets might be illegally already held abroad and could be recovered. Tax Justice Network (2005) estimated that the value of assets held offshore lies in the range of USD 11 - 12 trillion and suggested that the global revenue loss resulting from wealthy individuals holding their assets untaxed offshore may be as much as USD 255 billion annually. Cobham (2005), on the basis of shadow economy estimates of Schneider (2005) and Tax Justice Network (2005), derived a loss to poor countries of around USD 100 billion a year. Henry for Oxfam (2009) estimated that at least USD 6 trillion of poor country wealth is held offshore by individuals, depriving poor countries‘ governments of annual tax receipts of between USD 64 and 124 billion and, in a similar way, Henry (2012) estimated that a global super-rich elite had at least USD 21 trillion hidden in tax havens by the end of 2010 and that poor countries could be losing USD 189 billion in associated tax revenue every year. These studies were mostly first of their kind and put the related topics on the policy agenda. More recent research may offer greater rigor also. We discuss the studies of Henry (2012) and Zucman (2013) in this chapter after first considering the capital account-based IFF estimates of Ndikumana and Boyce and of GFI.
There have been a number of approaches to estimation of illicit financial flows on the basis of capital account data, with Erbe (1985) and World Bank (1985) being among the first to estimate the scale of capital flight, but most recent ones aim to estimate the difference between capital inflows and capital outflows. Capital inflows include net increases in external debt and net foreign direct investment. Capital outflows consist of the current account deficit and net additions to reserves.
The Hot Money ‘Narrow’ Method (HMN) uses only balance of payments data, usually from the International Monetary Fund, and is thus equivalent to the net errors and omissions reported there. This is the preferred method by the GFI.
Others, including the duo of Ndikumana and Boyce, prefer to use other data sources. For example, Ndikumana & Boyce (1998) argue that the World Bank’s data on debt provide more accurate estimates of the change in external debt.
Ndikumana & Boyce (2010) rely on data mostly from the IMF, specifically its International Financial Statistics, Balance of Payments Statistics, DOTS as well as IMF’s various country online information in selected issues and statistical appendix. Importantly, Ndikumana & Boyce (2010) also use the data from the World Bank’s Global Development Finance and World Development Indicators.
Ndikumana & Boyce (1998) thus measure the capital flight,$$KFit$$ , in a year t for a country i as, using a simplified version of their notation:
where
In a number of papers Ndikumana & Boyce (1998, 2003, 2010, 2011) have provided a number of estimates on the basis of this methodology, with a range of additional adjustments aimed at refining these estimates. For example, Ndikumana & Boyce (2010)� make four adjustments: for trade misinvoicing, exchange rate fluctuations, debt write-offs and underreporting of remittances. The trade misinvoicing estimate is discussed in detail in the previous chapter (Ndikumana & Boyce (2010) make the trade invoicing adjustment by comparing countries’ export and import data to those of its trading partners, assuming the high-income countries data to be relatively accurate and they interpret the difference as evidence of trade misinvoicing.) Ndikumana & Boyce (2010) make these adjustments, but do not highlight the scale of these individual adjustments, only the resulting overall estimates of capital flight.
In contrast with the other subchapters in this chapter, Ndikumana and Boyce do not aim to provide global results and focus on sub-Saharan African countries instead. Also, as mentioned above, their results do not disentangle the various adjustments they make. For example, Ndikumana & Boyce (2010) estimate that total capital flight from 33 sub-Saharan African countries between 1970 and 2004 amounted to 443 billion US dollars (and 640 billion US dollars when imputed interest earnings are included). Ndikumana & Boyce (2010) highlight that these estimates exceed these countries’ external debts (which in 2004 amounted to 193 billion US dollars).
Over the past two decades, Ndikumana and Boyce have provided likely the most prominent estimates of capital flight from sub-Saharan African countries. While these estimates have proved useful in raising awareness about the illicit financial flows, they do have methodological constraints. These are mostly shared with the other similar approaches discussed below and so we discuss them together in the following sub-chapter on the approach of GFI.
For capital account anomalies, the two most commonly used methods are the World Bank Residual Method (WBR) and the Hot Money ‘Narrow’ Method (HMN).
While most of these estimates in recent years have been prepared by GFI or Ndikumana and Boyce discussed above, UNDP commissioned a report by a lead author of the GFI estimates, �Kar (2011)�.
GFI uses the balance of payments published by the IMF as the only source for their estimates of balance of payments leakages (�Spanjers & Salomon, 2017).
The World Bank residual model subtracts the total of funds actually used by a country from the total of funds entering that country and, if there are more funds coming in than funds being used, the resulting shortfall is considered to be illicit flows. The hot money model considers all errors in a country’s external accounts as illicit flows. Both these methods rely on anomalies in the Balance of Payment (BoP) identity, as expressed in a notation by World Bank’s �Claessens & Naude (1993)� and followed by �Kar & Freitas (2012)� and others:
A + B + C + D + E + F + G + H = 0
Where:
A: current account balance
B: net equity flows (including net FDI and FPI)
C: other short-term capital of other sectors
D: FPI involving other bonds
E: change in deposit-moneybanks’ foreign assets
F: change in reserves of the central bank
G: net errors and omissions (NEO)
H: change in external debt
The WBR method captures the difference between recorded inflows and recorded uses, which is given by the (negative) sum of the current account balance, net equity flows, change in reserves of the central bank and change in external debt. By the BoP identity:
-(A + B + F + H) = C + D + E + G
Of the components on the right-hand side, however, C+D+E are licit: composed of other short-term capital of other sectors, FPI involving other bonds, and the change in deposit-money banks’ foreign assets. As such, the WBR method is likely to exhibit a substantial upwards bias as an estimator of IFF. Similarly, Fontana (2010) summarizes the World Bank residual method by the following equation: Illicit flows = (increase in foreign debt + increase in FDI) – (financing of the current account deficit + additions to the country’s reserves).
The main alternative, the HMN method, is given by the remaining right-hand side component, G: net errors and omissions. G is simply the balancing residual constructed to maintain the BoP identity, and so serves as an indicator of error – and possibly of illicitness – in the overall capital account. Again, Fontana (2010) summarizes the Hot Money model by the following equation: Illicit flow = all funds coming in (credit) – all funds going out (debt). Recently, this method, labelled as net errors and omissions (NEO) has been used by �Novokmet, Piketty, & Zucman (2018) to provide estimates of offshore wealth for Russia, which are three times higher than those estimated by another methodology by Alstadsaeter, Johannesen, & Zucman (2018). The longest-standing series of estimates, although published for African countries only, are those of Ndikumana & Boyce (e.g. 2008), discussed in the previous subchapter (they also contrast sources and uses of foreign exchange in the capital account and make a number of adjustments for exchange rate fluctuations on the value of external debt, for debt write offs and for under-reported remittances).
Likely the most well-known estimates are those produced by GFI. In 2012, GFI shifted from using the WBR method (e.g. Kar, Cartwright-Smith, & Hollingshead, 2010) to the HMN (e.g. Spanjers & Salomon, 2017, who label this method as balance of payments leakages). This change has naturally led to some inconsistencies in the results series over time (as well as the apparent increased role of trade misinvoicing in illicit financial flows), as discussed in some detail by Nitsch (2016). GFI combines these capital account estimates with trade estimates discussed in the previous chapter.
While illicit inflows could be considered to counteract detrimental effects of illicit outflows by increasing available capital resources, this position is questionable (see UNECA (2012) and AUC-UNECA (2015) for a more detailed discussion) because the damage of IFF to governance may be more important than the net resource effect. The benefits to the economy of illicit financial inflows to the economy may well be less than those of licit inflows, since the illicit inflows may themselves be going to fund the illicit economy (e.g. repatriation of profits by transnational organized criminal organizations may be used to fund expansion of activities in the country in question; the flows could also represent financing of terrorism); or be circumventing regulation or taxation designed to ensure fair competition. For our purposes, illicit financial inflows seem just as likely as illicit outflows to be distributed as or more unequally than funds in the licit economy, and so our primary interest is in estimates that do not 'net out' illicit financial inflows.
Having reviewed both capital account and trade approaches, it is also possible to combine these two types of models, capital-account and trade, and we discuss the results achieved by this combination. Most notably, the research by GFI uses the World Bank residual and hot money models and further makes adjustments for trade misinvoicing. Their hot money-based model estimates that the developing world lost USD 859 billion in illicit outflows in 2010 (significantly more than the USD 129 billion in aid by OECD countries in 2010). Their estimates, Kar & Freitas (2012), suggest that bribery, kickbacks, and the proceeds of corruption continued to be the primary driver of illicit financial flows from the Middle East and North Africa, while trade mispricing was the primary driver of illicit financial flows in the other regions. On the basis of this kind of estimates, Hollingshead (2010) uses national corporate income tax rates to estimate the tax revenue loss from trade mispricing in poor countries between USD 98 billion and USD 106 billion annually over the years 2002 to 2006.
In the most recent 2017 GFI analysis of illicit financial flows to and from developing countries between 2005 and 2014, Spanjers & Salomon (2017) estimate the illicit financial flows (or outlfows) from developing countries in 2014 at between $620 billion and $970 billion. In this report they publish such a range for the first time and they also put equal emphasis on inflows and estimate them in 2014 at between $1.4 and $2.5 trillion. In this most recent report, they still combine capital-account and trade approaches to estimating illicit financial flows. In their lower bound estimates of outflows, trade misinvoicing is responsible for two thirds of the total, while what they call unrecorded balance of payments flows (using net errors and omissions as a proxy for these) accounts for the remaining third. They estimate that Sub-Saharan Africa suffers most in terms of illicit outflows.
Cobham & Gibson (2016) show (in Figure 5) a comparison for estimates of total African IFF, between GFI methodology with WBR and HMN – Kar & Cartwright-Smith (2010), and Kar & Freitas (2011), respectively – and the Ndikumana & Boyce approach. Differences between the series frequently exceed the total value of the lowest estimate. Ndikumana & Boyce demonstrates greater volatility, as would be expected given in particular their use of net rather than gross trade mispricing. At the aggregate level, GFI’s updated (HMN) methodology tends to produce the more conservative estimates. These differences provide an important illustration of the sensitivity of estimates to assumptions. Note, too, that these are shown at the aggregate level; disaggregated, there are examples of quite different country patterns over time.
These capital account anomalies are of two types examined in this group of approaches. They either include changes in foreign portfolio investment, private and central banks’ foreign assets or they include only net errors and omissions. The limitations are obvious: the first is clearly not only anomalies and the latter is only anomalies, but not necessarily only illicit financial flows.
There are two main reasons to consider additional approaches. First, anomaly-based estimates inevitably attract criticism over the possibility that they may confuse ‘innocent’ anomalies including data errors and mismatches due to timing and rounding errors with evidence of illicitness, and the sensitivity to some of the assumptions made – see for example the various views expressed in five chapters of the World Bank’s illicit flows volume (�Reuter, 2012�: chapters by Eden; Fuest & Riedel; Leite; Murphy; and Nitsch). As such, while the range of estimates have established the scale of the issue in terms of the broad order of magnitude, the degree of confidence in the estimates may be less suited to specific policy analysis at the level of countries and IFF types. The second concern relates to the bluntness of the leading estimates. While it is useful to compare the component attributable to trade with that attributable to the capital account, and to separate out some individual and corporate tax abuses, greater specificity of the channels of IFF would be valuable to support policy prioritisation. Underlying these issues is the simple fact that flows that are hidden by design do not lend themselves to measurement.
From the point of view of the creators, the advantages of using the estimates of illicit financial flows by GFI or of a similar type are obvious from their relative media and policy success – they provide clear figures that many people can relate to, and that the media as well as researchers and policy makers can reference.
The drawbacks might be less obvious, but may be more numerous and important. These estimates indicate the possible aggregate extent of flows, rather than necessarily providing an accurate guide to prioritise specific policy efforts. The models rely on official statistics that are sometimes of poor quality, especially in lower-income countries. They do not take into account flows resulting from illicit activities, such as smuggling or black market activity, because proceeds from such activities are not captured in national accounts; nor a range of multinational tax abuses that do not generate anomalies in the series in question. Due to data publication time lags, GFI results – in common with most we evaluate – have a near two-year delay in publication of its estimates. Additionally, GFI provide results for individual lower-income countries, but not for their higher-income counterparts; although these results could possibly be arranged with GFI or estimated independently.
Ndikumana and Boyce have generally focused more on the stock of capital held outside African countries, than on the annual outflows. Similarly, Henry (2012), in a report for Tax Justice Network (TJN), produces global estimates with a largely common methodology, scaling up from outflows to estimates stocks of capital held offshore. The alternative approach here is to use data on international asset and liability positions in order to establish anomalies in the position of particular jurisdictions.
Henry (2012) states his objective as measuring long-term unrecorded cross-border private financial capital flows and stocks or unrecorded capital flows and stocks, with a focus on developing countries in particular. He also identifies a relatively wide scope of estimates, although not all relevant economic activity is included and, for example, he omits some types of non-financial wealth. In addition to new empirical estimates, Henry (2012) provides an overview of some earlier estimates and discusses other relevant evidence such as on transfer mispricing.
Henry (2012) uses a number of various data sources, including the World Bank, the IMF, central banks and countries’ national accounts. Each method uses different set of data sources, for example, for the unrecorded capital flow method he uses in most cases data from the WB’s World Development Indicators. Henry’s (2012) analysis of private banking assets uses a wide variety data sources including banks’ annual reports and interviews with private banking industry experts. Henry (2012) uses data for the period 1970-2010 and presents the results for the year end of 2010, while his 2016 updated estimates are for the year end of 2014 (Henry, 2016).
To describe methodology, we base our description on Henry (2012) and the available documentation of his methodology (his 2016 updated estimates seem to be applying the same or similar methodological approach, Henry, 2016). He combines four methods. While the first method is comparable to the capital account-based methods used by the GFI and Ndikumana and Boyce (in his words, unrecorded capital flows: “sources-and-uses”), the other three are based on an accumulated offshore wealth model, an analysis of private banking assets and an offshore investor portfolio model. These multiple methods are used to explore consistency. We describe the methods one by one below.
With the sources-and-uses method Henry (2012) aims to model unrecorded capital flows. To that objective he uses an adjusted version of the World Bank Residual (WBR) method which we describe in detail in the previous two subchapters and which estimates unrecorded capital outflows as the difference between recorded sources of foreign capital and uses of foreign capital. Henry applied a similar methodology in a report for Oxfam (2009). Similarly to Ndikumana and Boyce, Henry makes several adjustments. He follows their practice of exchange rate adjustments. He also deducts exceptional financing from the debt stock series, uses adjusted debt stock rather than debt flows and incorporated debt reschedulings and change in arrears. He applies his method for period 1970-2010 for each of 139 countries (the selection of which might be discussed in more detail) that Henry considers key capital source countries and that are mostly low-middle income countries. For 2010 his sample of countries covered most of the world’s population and a half of global GDP.
The accumulated offshore wealth model builds on the sources-and-uses method and aims to estimate how much the capital outflows are worth over time. Henry (2012) assumes that most (he writes “50 to 75 percent, on average”) of the flows offshore are reinvested there and the resulting offshore earnings are neither repatriated nor subject to any taxes. He assumes that the capital outflows are invested offshore at “a modest CD [certificate of deposit] rate”. Henry (2012) shows that China (and round-tripping via Hong Kong) is a counter example to the assumptions of this methodology and he argues that an adjustment for round-tripping is in place - 17 percent in the case of China, but perhaps in the case of other countries as an upper bound as well. In his analysis of private banking assets, Henry (2012) focused on cross-border private banking assets under management at the top 50 international private banks for the period 2005-2010.
Henry’s (2012) headline estimates are based on an offshore investor portfolio model, a version of which was in 2005 developed and applied by the TJN (Tax Justice Network, 2005). Henry (2012) builds upon and at the same time critically reviews the work of Tax Justice Network (2005), which he argues tended to underestimate the offshore financial wealth, which Tax Justice Network (2005) put at 9.5 trillion USD (and an additional 2 trillion USD in non-residential offshore real estate). The model takes data from the Bank for International Settlements (BIS) on cross-border deposits and other asset holdings by non-bank investors and scales them up (a multiplier of 3, which he considers conservative and provides references for values around 4) to arrive at total financial assets.
While the logical reasoning behind the four methodological approaches is clear, the relative lack of details (such as a detailed presentation of results or individual data sources) makes it challenging to evaluate the methods comprehensively. We tend to consider the accumulated offshore wealth model as an important reminder that outflows can be invested and multiplied over time (but in this specific method the assumptions play a crucial role and the estimates are by definition based on estimates from a different method and this might multiply some of the inaccuracies), while the analysis of private banking assets serves as a reality double check and indeed serves as a triangulation point (but it is, for example, not clear from the available data how much of the assets are held offshore or onshore).
The other two methods are the core of Henry (2012). While a critical evaluation of the sources-and-uses method (a version of the WBR method) is already included in the previous two subchapters, we briefly discuss the offshore investor portfolio model here. There is the crucial assumption of the multiplier, which directly influences the estimated offshore wealth and while they are estimates of this multiplier, it does not seem possible to judge its accuracy beyond interviews with experts and the like. Additionally, there are now newer estimates of the multiplier as well as much broader range of the BIS data (including bilateral information on deposits) available and it might be interesting to revisit the estimates of Henry (2012) and to evaluate the method’s estimates empirically with this improved data.
Henry (2012) estimates that the offshore financial assets of high net worth individuals are in a range from 21.02 trillion USD to 31.53 trillion USD in 2010. Henry (2016) published updated estimates and the range has increased to 24-34 trillion USD and from 9 trillion USD to 12 trillion USD for developing countries. Henry's (2012) 9 trillion USD estimate for developing countries derives from the application of the sources-and-uses method (an adjusted World Bank Residual method) that he applied to 139 countries and results into a range of 7.3 to 9.3 trillion USD. Equivalent estimates by Henry for Oxfam (2009) are a lower bound of 6.2 trillion USD by 2007 and they imply 150-200 billion USD annual outflows out of developing countries and we might expect similar figures in later years, although Henry (2012) does not provide this (outflow) form of presentation of the estimates. The accumulated offshore wealth model of Henry (2012) may add as much as 3.7 trillion USD to global total unrecorded capital outflows.
Henry's (2012) headline range of 21-32 trillion USD comes from the offshore investor portfolio model and includes financial offshore wealth only (the range is a result of assuming the multiplier is either 3 or 4.5). Henry's (2012) analysis of private banking assets finds that at the end of 2010 the top 50 international private banks managed more than 12.06 trillion USD in cross-border invested assets from private clients (including through trusts and foundations) and he considers this to be consistent with the results of the offshore investor portfolio model. He also observes that the top ten banks in his group are stable and grew even faster than the top 50 as a whole (20% in comparison with 16% per year on average between 2005 and 2010). Henry (2012) argues that the multiple estimates are consistent with each other (for example the results of sources-and-uses model for developing countries seem consistent with the combination of the offshore investor portfolio model and of the assumption of 25-30% share of developing countries in offshore wealth). Furthermore, Henry (2012) compares his overall estimates with the 2011 Credit Suisse Global Wealth Report, which puts global wealth at 231 trillion US dollars, and argues that it makes his estimates seem reasonable and conservative.
Overall, the Tax Justice Network report of Henry (2012) made a significant contribution to the study of wealth held offshore, and it provided one of the first and most elaborate empirical estimates of this phenomenon, helping to bring public and researchers’ attention to it. The contribution is especially valuable if one considers it as an “open challenge to the IMF and the World Bank – to all comers, in fact - to see if they can come up with better estimates” (Henry, 2012, p. 4). The international organisations have not responded yet with their estimates, although they do now devote themselves more to some related research and policy than in the past. The following subchapter presents a formidable response to this challenge, coming instead from academic research.
In an original contribution, Zucman (2013) estimated how much undeclared wealth held might be hidden in tax havens using detailed data on financial wealth managed by Swiss banks on behalf of foreigners that he further updated and presented (Zucman, 2014, 2015). In a recent follow-up study with co-authors, Alstadsaeter, Johannesen, & Zucman (2018) disaggregate the earlier estimates of offshore wealth by Zucman (2013) by country. Alstadsaeter, Johannesen, & Zucman (2018) still uses the detailed Swiss data but enrich it with other international sources, including the recently disseminated bilateral data from the Bank for International Settlements (BIS) on deposits in a number of tax havens by foreigners. This enables them to present how much wealth various countries’ citizens hold in tax havens (or offshore, two terms that we use here interchangeably, but more detailed specification is provided by Alstadsaeter, Johannesen, & Zucman, 2018, p. 7).
In this subchapter, we present research by both Zucman (2013) and Alstadsaeter, Johannesen, & Zucman (2018), but focus on the latter that include country-by-country estimates, which are in line with this book’s objectives of having country-level indicators that enable tracking over time. Zucman (2013) and Alstadsaeter, Johannesen, & Zucman (2018) provide methods and estimates that likely provide the currently most reliable estimates of wealth in tax havens, although better data in the future should enable further research to improve on them in a number of areas that we discuss below (for example, they do not capture non-financial wealth and they provide estimates of only financial wealth).
Alstadsaeter, Johannesen, & Zucman (2018) use three main data sources. First, they use detailed statistics from the central bank of Switzerland on the bank deposits, portfolios of equities, bonds, and mutual fund shares managed by Swiss banks on behalf of foreigners. This is data described in detail and first exploited for estimates of offshore wealth by Zucman (2013). In addition to having this data by definition of its source only for the Swiss banks’ operations in Switzerland, the crucial limitation of this data, acknowledged and dealt with by the authors, is that a large share of wealth owned by foreigners in Switzerland belong on paper to shell companies and other legal entities such as trusts and foundations that disguise the country of the beneficial owner.
Second, they use bilateral data on deposits in a number of tax havens by foreigners disseminated by the BIS since 2016. Until then the BIS published only the country-level data and it was thus not available when Zucman (2013) did his original analysis. The BIS data allows to exclude interbank deposits (deposits between banks that do not involve households), which Alstadsaeter, Johannesen, & Zucman (2018) do. The main limitation of this data is that, in contrast with the Swiss data, they do not include information on portfolio securities, which is the largest form of offshore wealth in the Swiss data.
Third, they use the IMF’s balance of payments and international investment position data to quantify the discrepancy in international investment positions. The equities, bonds and mutual fund shares owned by households on foreign accounts are recorded on the liability side, but not on their asset side (due to tax havens not reporting assets owned by foreigners).
They further use data from central banks, discussed by Johannesen & Zucman (2014), to exclude cross-border bank deposits by corporations and to keep only those by households.
Alstadsaeter, Johannesen, & Zucman (2018) first estimate the global offshore wealth using the discrepancy in international investment positions, in this following the approach of Zucman (2013).
Zucman (2013) observes differences in the securities assets and securities liabilities of all countries in the world: at the end of 2008 there are more liabilities (40 trillion USD) than assets (35.5 trillion USD), because, as he argues, tax havens are responsible for this difference and they usually do not report about assets owned by foreigners. An exception is Switzerland, on which data Zucman (2013) draws. He makes a number of assumptions (such as assuming that 25% of household offshore wealth worldwide takes the form of deposits and 75% of securities, as is the case in Switzerland), which are described in detail in his paper and are mostly needed due to data gaps, to arrive at an estimated 8% of total wealth worldwide held in tax havens. Alstadsaeter, Johannesen, & Zucman (2018) make similar assumptions as Zucman (2013), for example, they also assume on the basis of data from central banks that a given share of cross-border bank deposits belong to corporations and keep only those that belong to households.
Once �Alstadsaeter, Johannesen, & Zucman (2018) estimate global offshore wealth using the same approach of Zucman (2013) and more recent data (they put it at around 10% of world GDP and 5.6 trillion in 2007), they proceed in three steps to allocate it according to who owns the wealth: they start with who owns wealth in Switzerland, then proceed who owns wealth in the other tax havens, and, lastly, they combine the estimates from the two previous steps. The main obstacle they need to overcome in the first step is that most owners of Swiss offshore wealth is hidden behind anonymity, which has markedly increased after an EU regulation known as the Saving Tax Directive was introduced in 2005. This thus enables them to use data from 2003-2004 about the countries of owners as likely proxy for the owners of the shell companies in later years. Specifically, they assume that a country’s share of wealth not owned via shell companies in 2003-2004, it also owns the same percentage of the wealth owned via such shells. They support this assumption with consistent evidence from the leaked data of the Swiss subsidiary of the banking giant HSBC discussed in a related paper by the same authors (Alstadsaeter, Johannesen, & Zucman, 2017).
Similar to Swiss central bank, most other tax havens’ authorities also collect data on who owns wealth in their banks, but they publish them through the BIS only since 2016 (but the data are retrospective until 2000s or earlier) and in a less detailed form. At the time of their research (August 2017), Guernsey, Hong Kong, the Isle of Man, Jersey, Luxembourg, Macao, and Switzerland reported the data, while other important tax havens did not – the Bahamas, Singapore, and the Cayman Islands. However, for these latter tax havens they estimate the deposits owned in them by other countries as a residual. Crucially, the BIS data include information about deposits only, not about other offshore wealth. They thus make an important assumption that the distribution of deposits is the same as that of offshore wealth. In essence, to estimate the amount of offshore wealth in each tax haven using the BIS data, Alstadsaeter, Johannesen, & Zucman (2018) thus assume that the ratio of deposits to portfolio securities is the same in every tax haven (as in Switzerland), an assumption already made by Zucman (2013). They acknowledge that this might lead to potential biases (e.g. US corporations may own most of the bank deposits in Cayman Islands, but US households might own only a small share of the total offshore wealth in the Cayman Islands), but they argue, and we agree, that with the current data it is difficult to control for or quantify the size of the potential biases. What further research should investigate in detail even with the currently available data is the sensitivity of the estimated offshore wealth to this and other important assumptions.
Then, in the third step, Alstadsaeter, Johannesen, & Zucman (2018) sum the estimates of wealth held by other countries in Switzerland and in the other tax havens to arrive at their final country-by-country estimates of offshore wealth. Overall, the methodological approach to country-by-country estimates of wealth in tax havens of Alstadsaeter, Johannesen, & Zucman (2018) can be roughly summarised as their estimated offshore wealth for country
$$\times \frac{WealthinSwitzerland(Swiss central bank){it} + Wealth in the other tax havens (BIS){it}}{ Wealth in Switzerland(Swiss central bank){t} + Wealth in the other tax havens (BIS){t}}$$
Alstadsaeter, Johannesen, & Zucman (2018) then proceed to show implications of offshore wealth for distribution of wealth for ten countries, but we do not discuss these here in detail because they are not sufficiently relevant to our main objective of illicit financial flows.
We consider the methodologies and estimates of Zucman (2013) and Alstadsaeter, Johannesen, & Zucman (2018)� as the most reliable estimates of offshore wealth available, in terms of the country breakdown in particular. Still, these research papers naturally have their limitations. Both of these inevitably focus exclusively on financial wealth and their estimates are thus underestimates because they ignore non-financial wealth such as real estate, gold, works of art etc. Also, Alstadsaeter, Johannesen, & Zucman (2018)� present main results for 2007 and argue that the more recent period is contaminated by the use of shell companies. So there is obviously a scope for detailing the development in recent years, which might be important, as they acknowledge, for example in the case of China.
Using the discrepancy in international investment positions similarly to Zucman (2013), Alstadsaeter, Johannesen, & Zucman (2018) observe the scale of offshore wealth over time since 2001. They find that it remained equal to about 10% of world GDP (5.6 trillion USD in their benchmark year 2007). This is slightly lower and similar in trends to interview-based estimates by Boston Consulting Group (2017)�. Other companies and researchers also usually arrive at higher estimates, including Henry (2012) above, and they thus consider their estimates conservative. According to the detailed Swiss data, a large share of this estimated offshore wealth is held in Switzerland, but it is declining in recent years (30% in the recent years, compared to 40-50% in the 2000s). In contrast, Asian tax havens are on the rise; Hong Kong in particular experienced a steep increase in recent years and is now the second most important tax haven after Switzerland according to their estimates.
From the Swiss data �Alstadsaeter, Johannesen, & Zucman (2018) estimate that some countries own more wealth in Switzerland relative to their GDP (Saudi Arabia, United Arab Emirates, Spain, France, Argentina, Egypt), while other countries less (Denmark, Norway, Sweden, Japan, India, China). Overall, they do not observe clear patterns. From the other tax havens’ BIS data, they find that in 2007 Singapore, Luxembourg, Jersey and the Cayman Islands were most important. Asian countries seem less represented in Switzerland and more in other countries (Singapore in particular, and we argue, likely, in later years in Honk Kong after its increase in importance). Russia seems to have a lot of wealth in Switzerland and elsewhere, especially in Cyprus. While many European countries favour both Switzerland and other tax havens (including Luxembourg and Jersey) almost equally, Middle-Eastern countries favour Switzerland.
From the sum of these two estimates, they learn that while an equivalent of 10% of world GDP is held in tax havens globally, there are important differences across countries: a few per cent in Scandinavian countries, around 15% in Europe and up to 60% in Russia, Gulf countries and some countries in Latin America. They do not find a relationship between the scale and tax, financial or institutional characteristics, but do find that geography (such as proximity to Switzerland and reliance on natural resources) and history (such as instability since the Second World War), which is consistent with recent findings of Andersen, Johannesen, Lassen, & Paltseva (2017) that flows to tax havens are related with oil prices and political shocks. As a robustness check, Alstadsaeter, Johannesen, & Zucman (2018) show that their estimated offshore wealth is mostly very well correlated with how many shell companies a country created as indicated by the so called Panama Papers (the main exception being China, for which they do not estimate much offshore wealth, but for which many shell companies were created). In the remainder of the paper that we do not discuss here, Alstadsaeter, Johannesen, & Zucman (2018) show that taking into account the estimated offshore wealth increases the top 0.01% wealth share substantially in Europe and in Russia.
Despite recent progress in research on the scale of offshore wealth, there is still a long way to go to have reliable estimates of offshore wealth. The main limitation seems to be the available data, despite improvements in recent years, including the publication of the BIS bilateral data. Still, Switzerland is the only tax haven that publishes comprehensive statistics on the amount of foreign wealth managed by its banks. We thus join Alstadsaeter, Johannesen, & Zucman (2018) in calling for improving statistics.
Recent research has shown that a focus by tax authorities on wealth held by individuals offshore might bring the wealth onshore. For example, Johannesen, Langetieg, Reck, Risch, & Slemrod (2018) find that the US Internal Revenue Service’s enforcement efforts initiated in 2008 caused approximately 60 thousand people to disclose offshore accounts with a combined value of around $120 billion.
It is possible to estimate the income streams (perhaps comparable to the notion of illicit financial flows) that may accrue on offshore assets and Johannesen & Pirttilä (2016) provide a comparison of these: both Henry (2012) and Zucman (2013) estimate an offshore income stream of around $190 billion annually (Henry assumes a much more cautious rate of his return, on his much higher estimated stock). When country-level estimates are available, this may provide an alternative source of hidden income data to include in national distribution analysis. However, the additional extrapolations (from outflows to stocks, and then to potential income streams) inevitably add a higher degree of uncertainty.
Going ahead, it should be interesting to have a detailed comparison of various offshore wealth estimates - comparison of three methods and results: Zucman 2013, net errors and omissions from BoP by, for example, the GFI or Henry (2012) and at least of the other approaches applied by Henry (2012).