illicit financial flows – Page 2

The UK’s Financial Conduct Authority has revealed the basis on which it ranks jurisdictions as low or high risk for money laundering – and it seems inevitable that it will support debanking of poorer countries.

AML rules under pressure

First a little context. There has been growing pressure lately on anti-money laundering (AML) rules. In recent years, a string of major banks has faced large fines for apparently systematic sanctions-busting. This has been followed by a pattern of withdrawal – ‘debanking’ – from a range of countries where the risks of inadvertently channelling funds of sanctioned and/or terrorism-related entities and individuals have come to be seen as too high.

On the one hand, there are reasons to be rather cynical about this process. First, because supporting generally small-scale remittances to Somalia, for example, is a far cry from accepting and anonymising Iranian funds – and presumably much less profitable. And second, because it feels a little convenient for major banks to be making a case for reduced financial regulation, in which their interests align with those of some of the world’s poorest people.

On the other hand though, there are good reasons to take the issue seriously. (Disclosure – I’m on a CGD working group looking at just this question, so I would say that…) First, even if debanking is motivated by relative profitability of Somalian remittances compared to Iranian sanctions-busting, the potential development impact of remittance channels becoming more expensive is nonetheless substantial. (And we surely don’t expect banks not to respond to profitability.) Financial inclusion also seems to be associated with lower inequality.

And second, we should take the issue seriously because ultimately we want AML rules that work, for everyone, and demonstrably so – which is not the case now.

The question is not whether and how AML rules should be relaxed. It is this:

How can AML rules be designed so that the risks facing banks and other financial institutions are proportionate to the risks of carrying criminal flows, and not inadvertently supporting discriminatory outcomes against poorer countries (and people)?

An inexplicably bad approach

The UK’s Financial Conduct Authority (FCA) is accountable to HM Treasury and the UK parliament for regulating more than 50,000 firms to ensure integrity of financial markets. As Matt Collin points out in a great post, the FCA has just fined the (British branch of the) Bank of Beirut £2 million, and ordered it to sort out its AML procedures.

In the interim, the bank is barred from taking on new business in ‘high risk’ jurisdictions – which the FCA defines as anywhere scoring 60 or less out of 100 on Transparency International’s Corruption Perceptions Index (CPI).

Matt makes two important points about the weaknesses of this approach:

The CPI doesn’t reflect AML risks. Not a single one of the surveys which are aggregated into the CPI involves perceptions of money-laundering.
The threshold is arbitrary – and includes nearly 80% of the 175 countries for which ratings are produced. See Matt’s great figure.

Let’s add a couple of other points:

Even on its own terms, the CPI is a very bad measure of corruption. Sorry and all, and I think many TI chapters do really fantastic work; but the quicker the organisation drops the CPI, the better. Nor should anybody else be using it, as if it were some kind of objective indicator of corruption (never mind money-laundering) – it’s not.
And here’s the real kicker. The CPI is mainly telling you one thing: how poor a country is. Per capita income ‘explains’ more than half of the variation of the CPI (for 2012, which I happened to have to hand). The equivalent for the Basle Anti-Money Laundering Index, which includes the CPI among its components, is a little over a third.

CPI v lngdppc

So: the FCA is basing their AML risk measure on an arbitrary threshold, in a bad measure of corruption, which has nothing to do with money laundering, and mainly reflects income poverty.

An alternative approach

What could the FCA do instead? Well, they could use the Basle index. Or they could follow the lead of researchers at the Italian central bank, or a German rating agency among a good many others – and use TJN’s Financial Secrecy Index (FSI).

The FSI – which is also a component of the Basle index – brings together 48 variables, predominantly from assessments by international organisations, to create 15 indicators of financial secrecy – that is, of the risk factor for money-laundering, tax fraud and other financial crimes. These are then compiled into a single ‘secrecy score’.

For the FSI, this is combined with a measure of each jurisdictions’ global scale in order to produce a final ranking that reflects the relative potential to frustrate other countries’ regulation, taxation and anti-corruption efforts.

For a risk measure, you’d only want to use the secrecy score (or perhaps a subset of indicators that are most tightly relevant to money laundering). Relationships with per capita income are much weaker and of mixed direction, reflecting the basis in objectively assessed secrecy and scale criteria rather than perceptions of corruption.

Conclusion

To recap: If a financial regulator were to design a simple risk measure that would be most likely to lead to debanking of poor countries, while at the same time having no impact on the most risky jurisdictions, it’s hard to see how they could have done better than the FCA.

The broader lesson for the necessary rethinking of AML rules seems fairly clear. What are needed are context-sensitive measures that encourage responses proportionate to the actual financial crime risks – rather than encouraging the blanket withdrawal of services to poorer countries and/or people.

Intuition for a risk-based approach

The alternative, or complementary approach, is to pursue a risk-based analysis. Because of the behaviours involved, whether IFF are strictly legal or not, they contain some element of social unacceptability that means the actors involved will prefer to hide the process. For that reason, the risk of IFF will be higher – all else being equal – in transactions and relationships that are more financially opaque.

That will mean, for example, that the chances of uncovering IFF will be higher in anonymous shell companies than in companies with complete transparency of accounts and beneficial owners. Not all anonymous shell companies will be used for IFF, but the risk is higher. Similarly, at a macroeconomic level (at which level much data tends to only be available, unfortunately), trading with a relatively financially secretive jurisdiction such as Switzerland will be characterised by a higher IFF risk than trading with a relatively financially transparent jurisdiction such as Denmark.

Scoring financial secrecy

At present, the most common measure of financial secrecy is the Financial Secrecy Index (FSI), published every two years by the Tax Justice Network, and now used widely—for example, as a component of the Basle Anti-Money Laundering Index and of CGD’s Commitment to Development Index, and as a risk assessment tool recommended in the OECD Bribery and Corruption Awareness Handbook for Tax Examiners and Tax Auditors.

The secrecy score on which the FSI is based reflects 49 measures, grouped to form 15 indicators, which capture a range of aspects of financial secrecy from transparency of beneficial ownership and accounts, through international juridical cooperation. The secrecy score ranges in theory from zero (perfect financial transparency) to 100 per cent (perfect financial secrecy); in practice no jurisdiction has scored less than 30 per cent.

Calculating IFF risk measures

Consider an illustration, involving one country’s exports – say Ghana. For each trading partner, we combine its share of Ghana’s exports with its secrecy score (which ranges from zero to 100). The results can be summed to give an overall level of secrecy for all of Ghana’s exports, and this score reflects Ghana’s vulnerability to IFFs in its exports (the flow-weighted average financial secrecy of all partners). If we multiply this vulnerability score by the ratio of Ghana’s exports to GDP, we obtain a measure of the country’s exposure to IFF risk, which can then be compared across other stocks or flows.

A vulnerability of 50, for exports equal to 10 per cent of GDP, would give an exposure of 5 per cent. This is equivalent to Ghana carrying out 5 per cent of its exports with a pure secrecy jurisdiction (that is, one scoring 100 out of 100), while all other exports go to completely transparent trading partners. The exposure can then be thought of as Ghana’s pure secrecy-equivalent economic activity, as a ratio to its GDP. (Note: Where no secrecy score is available we apply the lowest observed score of 33. This will bias scores downward, though much less so than assuming a zero score.)

This measure of intensity of exposure to IFF risk can then be compared (given data), across time, countries and stock or flow types (with some important caveats). Table AIV.4 from the Mbeki panel report provides an indication of the overall intensity of exposure across African countries (excluding the major conduit jurisdictions).

Further detail can be found in Annex IV of the Mbeki panel report, while Alice Lépissier and I are working on a full paper to follow. Comments on the approach are very welcome indeed.

Tag: illicit financial flows

Poverty – a bad money-laundering risk factor