Andres Knobel ■ Germany’s new statistics on exchange of banking information: a trove of useful data and clues


For some years now, countries have been sharing information about bank accounts held by each others’ residents, implementing “automatic exchange of bank account information” under the OECD’s Common Reporting Standard (CRS). (This is what we called for many years ago, and were scorned at the time for being naïve and utopian.)

As more countries join the system, however, lower-income developing countries are being excluded because without sophisticated information-collecting mechanisms they cannot easily reciprocate even though this is usually pointless (it’s unlikely that rich Swiss crooks will stash their loot in Nigeria, for instance). The other big problem is the United States, whose Foreign Account Tax Compliance Act (FATCA), which is similar to the CRS, is good at getting other countries to send banking information to the US, but shares very little information in return. (That is why the United States is one of the world’s biggest tax havens.)

Information is now being shared across borders, but journalists, civil society organisations and others cannot know much about automatic exchanges other than the fact that they are happening. Do we trust governments to use them wisely? No — and we especially mistrust many tax havens (in the EU, some countries didn’t even bother to open the CDs containing the information).

So we have been asking at least for statistics: we have even designed a template (see p37 of this report.) Global statistics would, among other things, show avoidance schemes and other red flags, say if rich Italians closed their accounts in Switzerland and moved them to Luxembourg, or if they are acquiring ‘golden visas’ to escape being reported altogether, or hiding behind lawyer-held accounts. This is the only way in which we can hold authorities to account for how they are using the information they are now receiving (and sending). In addition, we argue that countries should report about how they are using the received information, such as whether they were able to match it to the tax returns of local taxpayers, and whether sanctions were imposed.

However, these statistics, such as the ones recently published by Germany, don’t help developing countries that are excluded from the automatic system. That should change. But even if these countries are excluded from the direct exchanges of information, there is still plenty that can be done for developing countries. We argue that countries using the CRS must adopt the “wider-wider” or “widest” approach when collecting information from their local financial institutions (following the examples or Argentina, Australia, Estonia, and Ireland). Countries implementing the “widest” approach require local banks to collect and report to authorities information on all account holders, not only on those residents from countries participating in the automatic exchange scheme. This would cover residents from excluded lower-income countries.

Countries should then also publish statistics on what they have found. This would help lower-income developing countries, which would at least be able to know how much money in total their residents hold in each financial centre (but they still unfortunately wouldn’t be able to know who those accounts belong to because they are excluded from automatic exchanges).

In fact, Australia has already (thanks in no small measure to pressure from TJN-Australia) started publishing such statistics: they describe the accounts held at  Australian financial institutions by residents from 248 jurisdictions (of which only 100 participate in automatic exchange of information.)

Now, back to Germany.

Based on our research and proposals, Fabio de Masi (Die Linke) with support from Tax Justice Network-Germany made a parliamentary enquiry to obtain statistical information on automatic exchanges relating to Germany. They have received a reply which they shared with us, which raise a number of questions to be investigated further and addressed.

Germany’s statistics on automatic exchange of bank account information

FATCA information exchanges with the US:

  • Accounts held by Germans (separating between individuals and entities) in US banks: number of account + income (but no information on account balances)
  • Accounts held by Americans (separating between individuals and entities) in German banks: number of accounts + income + account balance

CRS information exchanges with other countries:

  • Accounts held by Germans (no distinction between individuals and entities) in banks from 88 countries: number of accounts + income + account balance
  • Accounts held by residents from 66 jurisdictions (no distinction between individuals and entities) in German banks: number of accounts + income + account balance

The statistics are not as comprehensive as we would like. For instance, while the CRS covers financial account information that may be held in depository banks, custodial banks, some investment entities and some insurance companies, the statistics refer to all accounts and income together, without making any distinction. Annoyingly, statistics were produced as a pdf rather than as machine-readable data, but after spending some time converting the data into an Excel document, here’s what we have found.

Preliminary observations on Germany’s exchanges under the OECD’s Common Reporting Standard (CRS)

  1. Secrecy jurisdictions opposing statistics

The UK, a self-proclaimed transparency champion, has refused to let Germany publish even statistical information on the information sent by the UK relating to accounts held by German taxpayers in UK financial institutions. (The same happened with Canada, Cayman Islands, Isle of Man and South Korea.) These refusals are strange since there cannot be a privacy problem. No personal information is published: only collective statistics.

Although it is nonsense that a country may prevent another from publishing statistics that involve no personal information, in any case this refusal could only refer to information sent by those opposing countries: for instance, if the UK told Germany that there are X Germans with bank accounts in UK banks. However, the UK should have no say at all on the information reported by German banks to German authorities about British account holders in Germany. That is pure German data.

Unfortunately, Germany got it wrong and failed to publish statistics on the accounts held in German banks by residents from these opposing countries (for example, how much money Brits have in Germany). In contrast, Australia did publish statistical information on accounts held by residents of the UK, the Isle of Man, Cayman and more than 200 other jurisdictions. Germany can easily do it.

In any case, we have taken the UK’s refusal as an encouragement to submit a Parliamentary Enquiry in the UK for them to publish the data. Let’s see what comes back.

  1. Voluntary secrecy affecting statistics

Another factor limiting German statistics even further is the “voluntary secrecy” chosen by some countries under the CRS. This is where a secrecy jurisdiction chooses not to receive any bank account information from foreign countries. According to the OECD’s Global Forum which assesses countries compliance with exchange of information, this option was chosen at least by Anguilla, The Bahamas, Bahrain, Bermuda, British Virgin Islands, Cayman Islands, Kuwait, Marshall Islands, Nauru, Qatar, Turks and Caicos Islands and United Arab Emirates.

As we had warned, this voluntary secrecy by those countries means that German statistics report less data regarding non-residents holding German bank accounts. Germany published information about Germans’ accounts in 88 countries, but when it comes to accounts held by foreigners (non-residents) in German financial institutions, Germany published information for only 66 countries. So 22 jurisdictions are missing: all the voluntary-secrecy jurisdictions (plus Aruba, Belize, British Virgin Islands, Costa Rica, Curacao, Grenada, Ghana, Macao, Montserrat, Lebanon, Samoa, St. Kitts and Nevis, St. Lucia, St. Vincent and the Grenadines and Vanuatu. (For aficionados of our Financial Secrecy Index, this list correlates to Indicator 18.)[1]  

The data

The data produces a number of anomalies and questions.

a) Germans’ foreign accounts in 2018 (info that Germany received from other countries)

As already reported by Business Insider and der Spiegel, in 2018 Germans held most of their overseas bank account money in tax havens: Jersey (180 billion Euro), followed by Switzerland and Luxembourg. Other than EU countries and India, two other secrecy jurisdictions stand out: in 9th place is Guernsey (10 billion Euro) and 11th Singapore (3 billion Euro).

If we mine the same data to calculate average account balances, another strange picture emerges.

The top 15 countries by average account balance are all typical tax havens and secrecy jurisdictions: Jersey is top, with 15 million Euro per account. But it is odd that the average account balance of German account holders is 100 times larger in Jersey than in Switzerland. Does this relate to the fact that German individuals have more personal money in Switzerland, while German entities or investment funds have more in Jersey?. We don’t know, because German statistics fail to disaggregate this, but it should prompt further investigation.

The strangest thing, however, is the average income per account. In this case, Germans reported an average of 2 million Euros of income per account in Colombia, followed by Argentina and Curacao (ca. 150,000 Euro of income per account). In Switzerland, the average was merely 50,000 Euro of income per account. Was this a mistake, or are Germans making vastly more money in Colombia from financial assets? Again, further investigation is needed. Similar outliers are found when comparing income and account balance (how much income was reported for every Euro of account balance). For Colombia, the relationship is 19,000 per cent – 190 times more income than account balance – which is ridiculous. In contrast, in Switzerland or Luxembourg the relationship was closer to 30 per cent. This, too, merits further investigation.

More questions arise when we consider the accounts in Germany held by non-residents.

b) Accounts held by non-residents in Germany in 2018 (info sent by Germany to other countries)

These show that:

Residents of Switzerland, France and Austria, and mostly of EU countries, have the highest number of accounts and account balances in German financial institutions.

However, when looking at the income from financial assets held in German financial institutions, Dutch account holders are spectacularly successful compared to the rest. 210 billion Euro, compared to incomes between 4.3 and 2.4 billion Euro for account holders from Switzerland, Austria and Luxembourg. Again, this merits further explanation.

When comparing the relationship between income and account balance, the Netherlands also stands out, with a value of 2,524 per cent (25 times more income than account balance, on average), which again makes no sense. Next are Saudi Arabia and Russia with 50 per cent each. Another thing to look at. Is it a mistake in the numbers, are Dutch account holders removing their money from the banks right before 31 December when the account balance is calculated, or is there some other explanation?

Again, when looking at the highest average account balances in German banks the list is a gallery of tax havens and secrecy jurisdictions. The only jurisdictions whose residents have more than 100,000 Euros on average are Jersey (726,000 Euro), Monaco (500,000 Euro), then Guernsey, Liechtenstein, Cook Islands, Panama, Luxembourg, San Marino, Cyprus and Malta.

Other anomalies and questions emerge from odd changes in account balances and income. For account holders from Gibraltar, the number of accounts in German financial institutions remained stable at around 800 accounts between 2017 and 2018, but the reported account balance totals went from 1 billion to 73 million Euros, and the income from 730 million to 8 million Euros! Meanwhile, accounts from Jersey residents also stayed at 400 between 2017 and 2018, while their income went from 87 million to 13 million Euro on average. Something similar happened with the Seychelles, where there were also ca. 400 accounts in 2017 and 2018, while the reported income went from 20 million to 4.7 million Euro. Were these account holders terribly unlucky between 2017 and 2018? Did they change investments while keeping their accounts open? Did they keep part of their income unreported? Again, we need to know more.

Preliminary observations on exchanges with the US under FATCA agreements

Once again, the statistics show that the USA is a gigantic tax haven. No information on German beneficial owners of US accounts is provided, because the US does not exchange information at the beneficial ownership level. (Put simply, if a German individual holds an account in a US bank using a non-German company or trust, Germany won’t get any information about it.)

In principle, however, if the US bank account is held by a foreign company, the US may at least provide some information to the country where that company is resident. The figure below illustrates what can and can’t happen.

The first case in the diagram, “Anna” and Company A, involves countries that signed a “Model 2” agreement with the US where they send information to the US but  receive nothing back from the US.  These countries include Austria, Bermuda, Chile, Hong Kong, Japan, Macao and Switzerland, among others. Here, if a German individual holds an account in a US bank through an Austrian company (Company A in the figure below), no country will receive information about that because Austria decided not to receive information from the US.

Panama is different: with a German (“Paul”) owning a Panamanian company, Panama will receive information because the account holder is a Panamanian company.  Germany will receive no information, however. Germany will receive information from the US if a German entity (Company C) or a German individual (Markus) directly hold an account in a US bank: Germany signed a Model 1A agreement so it will receive some basic information from the US.

In contrast, Germany exchanges information with all participating countries, including the US, both at the level of the account holder (entity or individual) or the beneficial owner. Here it is, in another diagram, showing the scandalous imbalance between how much, and what type of information Germany provides the US, including at the beneficial ownership level, compared to what it gets in return.

The US sends Germany information on the income received by German accounts holders but not on their account balances. In contrast, Germany shares information with the US both on the income and the account balance of US account holders (and also for US beneficial owners). This imbalance becomes abundantly clear by comparing the following two tables.

Account balance reported by Germany

(“Konten naürlicher Personen” means “accounts of natural persons”, the next column is “accounts of non-natural persons,” and the third column is “all accounts.”)

Account balance reported by the US:

As regards the publication of statistics, given that Germany collects and exchanges information at the beneficial ownership level (eg the cases of Mary and John), it could easily publish statistics disaggregating what information refers to beneficial owners (who aren’t the direct account holders). However, German statistics fail to disaggregate information at the beneficial owner both for American accounts held in Germany, as well as for bank account information sent and received within automatic exchanges with other countries under the OECD’s CRS.


Statistics on banking data can be very useful to get indications of illicit financial flows, and to look for avoidance schemes or potential mistakes. Statistics may not determine wrongdoing by themselves but signal many outliers that should prompt an investigation, and allow journalists and civil society groups to ask difficult and challenging questions to prevent cover-ups. If we can find all these anomalies based on totals for only three years, provided by one country, imagine all the checks and investigations that would be possible with more details.

It would be still more useful to have disaggregated information based on:

  • Account holders who are natural persons or entities,
  • The country of residence of the beneficial owners of the account holders which are entities.
  • Disaggregated information above based on the type of financial institution that holds the account (eg depository bank or investment entity). For example, Germans had $X million in US depository institutions and $Y million in US investment entities and $Z million in US insurance companies.
  • Germany should follow Australia’s example and publish information not only on those account holders who are resident in a jurisdiction participating in automatic exchange of information but also on residents of lower-income countries which are still excluded from the automatic exchange system. In other words, Germany published information on the accounts held in German financial institutions for 66 jurisdictions. Australia did the same for 248 jurisdictions, including all African, Latin American and Asian countries.

All the points above refer to statistical information on the information exchanged. However, authorities have access to the actual identity of account holders. We would thus need statistics on what authorities have done with the information, including on:

[1] These countries may include also those that failed to meet confidentiality requirements and are thus unable to receive information (EU countries are still mentioned by Germany because they do have to report information under the EU Directive of Administrative Cooperation (DAC 2)). It’s not clear why Dominica isn’t mentioned in Germany’s statistics, though.

Related articles

Comments • 1

  • Longtail
    April 18, 2021 - 12:15 pm

    Sadly, there is a serious flaw in this data.

    This data makes the unfortunate assumption that that all credits made to an account is “income”.

    However, movement of capital between accounts (which could be significant) also gets swept up into this “income bracket.”

    So the conclusions outlined above could be partly based on “meaningless noise”.

  • Leave a Reply

    Your email address will not be published. Required fields are marked *

    This site uses Akismet to reduce spam. Learn how your comment data is processed.