Mega thread by @DanNeidle on the Financial Secrecy Index & from @alexcobham in response (respect to both for engaging, and to anyone ploughing through it)
I want to add something which isn't picked up in the point-by-point discussion, but is important to assess the Index. The stated aim of the index is identify the extent to which jurisdictions "use secrecy to attract illicit and illegitimate or abusive financial flows" (1/)
Its about identifying 'crimeogenic hotspots for multiple evils'. This is how the FSI describes itself. And this is how most users in the media, public and other organisations see it. So this seems like clear common ground to evaluate it from. (2/) 
The only score not considered secretive is '0' (i.e. perfect compliance w the criteria). Everything else runs on a scoring scheme from "moderate secrecy" to "extreme secrecy". (NB: UK mainly gets moderate scores - so this is the end of the scale that Dan & Alex are discussing)
This is not just a presentation issue. Because of the scale weighting, "moderate secrecy" scores in a large jurisdictions generate worse ratings than more severe scores in smaller jurisdictions. This is an issue picked up in the statistical audit financialsecrecyindex.com/PDF/FSI-Method… (4/)
So what kinds of thing can get you into the 'moderate secrecy leagues'?. Here is a list (my list, from the methodology financialsecrecyindex.com/PDF/FSI-Method…) . A lot of them I think are quite surprising (and not many of them are about secrecy)
For example (1) and (17) are about secrecy and based on FATF compliance. But a country which is 'largely compliant' across the FATF scores is rated as moderately secretive. Similarly (3) on beneficial ownership. A jurisdiction that is FATF compliant can be 'moderately secretive'
Then there are some that are not about secrecy. e.g. UK's non dom regime marks it down as moderately secret under(12). & exemption for tax on dividends from foreign subsidiaries under (13) - These may be policy preferences, but not really 'secrecy'
Then there are some that are more transparency than secrecy. e.g. is there Open Data on property ownership, Ltd. partnerships, beneficial ownership & company accounts? (4,5,6,7) opendatahandbook.org/guide/en/what-… - Can argue for Open Data, but odd to view full mandatory disclsure as secrecy
Then there are two on country-by-country reporting 8 & 9 - which sit strangely in an index which is about money laundering, lack of international cooperation etc.. 9 is rates a country as moderately secretive if it carries out exchange of CBCR according to OECD agreement
Anyway, I think these design questions are behind some of the gap between Dan & Alex here -- e.g. Dan is saying 'X doesn't make sense as a measure of secrecy/crimeogenic environment' and Alex is saying 'you don't understand how the index is constructed.' Both are true.
Ultimately for Index to be reliable for users it should assess what it says on the tin, objectively & recognisably to practitioners. I hope this conversation continues. Tax people & civil society folks shld talk more. Probably reached limit of Twitter as a medium on this one tho.
