Thursday, May 09, 2019

FATF and EU need to fundamentally rethink their approach to virtual assets/currencies...

Virtual currencies are on the radar of regulators for quite some time. Yet it is clear that they still struggle with definitions (which always happens when new technologies arise). The FATF is a key example now that they are seeking to harmonise international guidelines for applying FATF-rules to the crypto-world.

In this post I will look at some of the issues at stake and explain why the FATF-exercise requires a lot more time and thinking before the FATF (or EU) move forward. Do note that this is a longread, more geared to specialists in the field, than the general public.

For the public it boils down to this. The US is pushing all countries in the world to a situation where with each virtual or crypto transaction, your information needs to be distributed (by definition) to other players in the value chain.

But as the crypto definitions in countries diverge (and the FATF-definition is ill defined, potentially covering everything in the world), the only sensible thing to do is to stick with the local definitions of crypto-assets and to demand transaction information to be stored locally at the point of transaction. Any law enforcer wishing access to that information should thus approach the relevant local authority for that information.

Apart from this legal argument, we must acknowledge the recent regime changes in the world. It is by no means clear that countries that used to obey the law and follow the rule of law, will do so in the future. Thus, foreign law enforcers may become tools in the hands of local undemocratic rulers.

That is an additional argument that requires the EU (but also the FATF itself) to avoid the situation that a local law enforcer in an undemocratic country can get EU data by harvesting its home companies data for the EU-info, without having an appropriate legal warrant under EU-rules.

And now for the longread part of it...

Definitions: always tough
Back in 2012, the ECB had a hard time grasping the concept of cryptocurrencies. They used the fact whether or not virtual currencies were regulated as their guiding principle:
A virtual currency can be defined as a type of unregulated, digital money, which is issued and usually controlled by its developers, and used and accepted among the members of a specific virtual community.

The US regulator (FINCEN) chose the following approach in 2013:
In contrast to real currency, “virtual” currency is a medium of exchange that operates like a currency in some environments, but does not have all the attributes of real currency. In particular, virtual currency does not have legal tender status in any jurisdiction. This guidance addresses “convertible” virtual currency. This type of virtual currency either has an equivalent value in real currency, or acts as a substitute for real currency. 

FINCEN then applied the money transmitter laws in an extensive way to bring exchanges of virtual currencies into their supervisory remit.

Later on, the ECB changed its definition to:
For the purpose of this report, it is defined as a digital representation of value, not issued by a central bank, credit institution or e-money institution, which in some circumstances can be used as an alternative to money. 
The EU stance remained that cryptocurrencies did not conform with definitions of funds and such in the EU legislation, hence their exchange and use was not regulated as such. Of course the integrity and consumer risks were identified and warned for.

In the FATF-context (2015) we read:
Virtual currency is a digital representation of value that can be digitally traded and functions as (1) a medium of exchange; and/or (2) a unit of account; and/or (3) a store of value, but does not have legal tender status (i.e., when tendered to a creditor, is a valid and legal offer of payment)6 in any jurisdiction. It is not issued nor guaranteed by any jurisdiction, and fulfills the above functions only by agreement within the community of users of the virtual currency. 

While these definitions may seem to work at first sight, we still need some creativity to determine the boundaries of these virtual currencies. Essentially it is possible to bring any loyalty point scheme under these definitions, as they do not use a subject based qualification to determine what exactly virtual currencies are.

At that point in time, where the focus was mostly on payments and such, using the experience we had with e-money definitions, I suggested a framework based on objects of the digital values at hand:


User cannot buy tokens at all (loyalty-type)
User earns tokens and can buy additional (hybrid of loyalty/payment)
User buys and sells tokens
(payment-type)
Tokens used in digital issuer-domain only

World of Warcraft
World of Warcraft
Lynden Dollar
Tokens used in digital or physical issuer-domain only
Starbucks
Nintendo Points
-Digital Payment loyalty schemes for single retailers

Tokens used at other entities than the issuer
Frequent Flyer Programmes
Frequent Flyer Programmes
Bitcoin,
e-money on mobile phone's


I think it would be fair to say that, while we pretend to have solved the application of crypto-legislation to the payment-type currencies, we actually haven't truly done so. There are still classification issues pending, but they may have appeared to be too irrelevant to matter,

Enter: ICO's and token frameworks
The next stage however was the widening of the blockchain concept, the application of crypto to generic tokens and the use of tokens as a form of share, security or other representation of objects, value, cash flows. This leads to a big confusion all around the world whether or not to view some tokens as security tokens, utility tokens and such. So, while our first definition already had flaws, we chose a new wording to cover this brave new world: crypto-assets or virtual assets.

As ESMA noted in their warning on ICO's at the time:
Where ICOs qualify as financial instruments, it is likely that firms involved in ICOs conduct regulated investment activities, in which case they need to comply with the relevant legislation.
So the essential discussion of application of financial law was left to local supervisors interpretations and definition of financial instruments.

The definition-side remained quite weak, with crypto-assets being loosely described as:
Crypto-assets are a type of private asset that depends primarily on cryptography and Distributed Ledger Technology (DLT). There are a wide variety of crypto-assets. Examples of crypto-assets range from so-called cryptocurrencies or virtual currencies, like Bitcoin, to so-called digital tokens issued through Initial Coin Offerings (ICOs). Some crypto-assets have attached profit or governance rights while others provide some consumption value. Still others are meant to be used as a means of exchange. Many have hybrid features. 

ESMA noted then that there were many variations and that it was not necessary to regulate all forms of crypto-assets. In 2019 they published an updated analysis with still a very weak definition of crypto-assets:
Crypto-assets are a type of private asset that depend primarily on cryptography and distributed ledger technology as part of their perceived or inherent value. A wide range of crypto-assets exist, including payment/exchange-type tokens (for example, the so-called virtual currencies (VCs)), investment-type tokens, and tokens applied to access a good or service (so-called ‘utility’ tokens).

In their report they distinguish between payment, investment and utility token, to immediately outline that this distinction does not cover everything. So the definition issue remains as well as the question: which type of digital token falls under which type of regulation. Hence the EU is in need of more EU clarity on the subject.

On the other side of the ocean, the SEC has further fleshed out how to interpret generic financial sector rules to digital asset issuance/use. In a long awaited guidance note the answer ends up being: it depends on the way you structure the functionality of the token/asset and the use between investors and issuer. So depending on those features, it may well be a regular financial instrument and facilitating trading may constitute a regulated business of operating an exchange.

The FATF-approach: hammering financial services law into hardly defined virtual assets
In essence, the idea of the FATF is now to make sure all crypto-related business is covered in a layer of regulation that at the least ensures proper KYC and AML/CTF rules. As such, this can be appreciated and understood as a recognition of the fact that cryptocurrencies and crypto-assets are here to stay. If we bring the sale of high-value items such as diamonds or gold watches under the FATF-KYC/AML remit, it makes sense to also do so for digital goods/assets/cryptocurrencies (whichever legal status they have).

We do have a problem however, which is that the definition used by FATF, since October 2018, is still shaky:
A virtual asset is a digital representation of value that can be digitally traded, or transferred, and can be used for payment or investment purposes. Virtual assets do not include digital representations of fiat currencies, securities and other financial assets that are already covered elsewhere in the FATF Recommendations. 

This definition is so wide, that the FATF needs to explain:
The FATF emphasises that virtual assets are distinct from fiat currency (a.k.a. “real currency,” “real money,” or “national currency”), which is the money of a country that is designated as its legal tender.

The further definitions of virtual asset service provider clarify the intent of the FATF-definition: they wish to cover both former virtual currencies and the ICO area and use a very broad definition to describe virtual asset service providers. These are companies that for a business conduct:
i. exchange between virtual assets and fiat currencies; 
ii. exchange between one or more forms of virtual assets; 
iii. transfer of virtual assets; 
iv. safekeeping and/or administration of virtual assets or instruments enabling control over virtual assets; 
v. participation in and provision of financial services related to an issuer’s offer and/or sale of a virtual asset

These definitions are very shaky grounds to use. One particular troublesome issue is that the virtual asset definition has a negative part: it does not cover currencies, securities and other financial assets that are already covered elsewhere in the FATF-recommendations. It is a catch all phrase that brings all loyalty points in the world under the FATF-remit. Now, the FATF will of course outline that that was not their intent, but as soon as you devise a crypto-based loyalty scheme, who is going to decide?

And taking it one step further: if I convert my multilevel marketing scheme into digitally represented agreements on a blockchain, do these new tokens qualify as a contract (not covered) or as their value and virtual assets? And how does this interpretation play out in the US vs the EU legislative context?

I am certain there is a host of applications/use cases where we will find the FATF definitions being not suitable for use. How about CO2-emission rights. World of Warcraft-tools. Shared ownership of my house or my bycicle. I would urge the FATF to do some more thinking in that respect. The negative catch-all in a definition (it is a virtual asset when all other definitions in our recommendations fail) is just not good enough.

I can only commend the FATF on one point however. The positive thing about the definition is that it speaks of representation of value. This implies a monetary or self-invented value/currency. It does not state that it is about the representation of physical assets or objects (such as real estate). Or that value can also be understood to consist of anything in the real world, to which value can be attributed (ie. everything).

Applying FATF-money transmission rules to crypto-assets: technicalities!
Right now the FATF has closed its public consultation on applying the money transmission rules to crypto-assets. They are hammering a payments-network idea onto cryptocurrencies and crypto-assets alike to not just demand identification and transaction monitoring. The idea is to also apply the addition of originator and beneficiary into crypto-transactions:
(b) R.16 – Countries should ensure that originating VASPs obtain and hold required and accurate originator information and required beneficiary information2 on virtual asset transfers, submit the above information to beneficiary VASPs and counterparts (if any), and make it available on request to appropriate authorities. It is not necessary for this information to be attached directly to virtual asset transfers. Countries should ensure that beneficiary VASPs obtain and hold required originator information and required and accurate beneficiary information on virtual asset transfers, and make it available on request to appropriate authorities. Other requirements of R.16 (including monitoring of the availability of information, and taking freezing action and prohibiting transactions with designated persons and entities) apply on the same basis as set out in R.16

Where the approach worked in 2001 in a world where a payment was a payment, funds are funds and wire transfers are wire transfers how can it work in a world where fundamentally the core definition of virtual asset or crypto-asset is as vague as it is in EU and the US?

The whole exercises strikes me as a hasty effort, given that the authors have not noticed that also the interpretative note for Recommendation 16 should be changed to include virtual assets (exempting intra-VASP payments and e-commerce virtual currency payments from the scope). And it is clear that the US is driving the FATF to adopt the above change hastily - and without solid analysis - by June 2019.

To me, there is only one logical conclusion: in the decentralised world of virtual assets, with jurisdictions each applying different boundaries to crypto-stuff, there is no sufficiently harmonised basis to enforce the attachment of data to each transaction. Requiring service providers to hold the info and make it available by request is not a problem, but sending it out as we did with the former FATF7-rules is impossible due to the patchwork of diverging definitions.

In my response to the FATF-consultation I have outlined this problem:

In addition I would like to note that the divergent legal status of virtual assets (considering its wide definition) in different countries may have the consequence that under some local laws the transfer is not financial in nature and will not be covered under the financial legislation and AML/TF frameworks. It is possible that a sufficient legal basis is lacking in some jurisdictions to apply the crossborder wire transfer regime to such non-financial transactions and that data protection regulations take prevalence. This could be solved by applying the domestic wire transfer regime to transfers of virtual assets, regardless of their potential cross-border nature. The further application of this regime on the domestic level can then be geared to the specific legal qualifications for virtual assets in that specific jurisdiction.

My proposal is to follow the most efficiĆ«nt way. Strike out the part that says: submit the above information to beneficiary VASPs and counterparts (if any).  It is simply not proportional and economically sensible to demand as the FATF to include privacy-sensitive information in crypto-transactions. Officers can can have access by asking and demonstrating lawfulness of the request via international channels. But the day and age of using local tricks and harvesting local companies for EU-data should be over.

The area of digital assets, virtual assets is so ill-defined that the FATF cannot claim a full competency, as the legal basis in a number of jurisdictions will not be there. We should also keep in mind that the catch all definition - not elsewhere regulated under these FATF-rules - is still written under from the FATF role of being Financial Action Task Force, focusing on financial industry and financial services as the main objective. So if my home country defines certain digital goods as digital goods and not in scope of crypto legislation, that to me would be the end of the remit for the FATF (and it would remain out of scope of the catch-all clause as well).

So much for the technicalities.

Applying FATF-money transmission rules to crypto-assets: geopolitics
We should recognize that we are in a different moment in time than in 2001, when the FATF-7 rules were introduced. At that point in time the US was a beacon for democracy and rule of law. But it isn't any more.

It's role became fuzzy when it turned out that US law enforcers had used US based servers of EU companies (Swift) to get hold of EU-data. And this made the EU sensitive to the protection of its citizens against unwarranted overly ambitious law enforcing in other countries.

We should again be sensitive. The EU, but also the FATF, also have an obligation to protect their citizens from unduly harassment and intrusion by law enforcement authorities. And creating tons of data outside the consent-scope of the citizen does not sound like a good protection at all.

Right now, we can witness around the world, an increase in countries with all kinds of 'strong leaders' that violate human rights agreements, do not obey the rule of law, that are involved in money laundering schemes, do not listen to lawful requests of their constituents and ignore climate agreements.

I think the EU has a duty to not cooperate with implementation of so-called FATF-requirements when it is clear they are increasingly unable to protect the privacy and guarantuee the lawfulness of the data exchange. Requesting other states to go get the data (and ensure that it is proportional) is a better way forward.

In sum: improve definitions and reconsider the worldwide distribution of transaction data for virtual assets/currencies
While I think that FATF should fully reconsider its definitions and redo its homework, this virtual-asset momentum and this train that is being pushed by the US may be rolling too fast to stop it. So as a stop-gap one could propose to eliminat 7b or at least strike out the distribution line:
(b) R.16 – Countries should ensure that originating VASPs obtain and hold required and accurate originator information and required beneficiary information2 on virtual asset transfers, submit the above information to beneficiary VASPs and counterparts (if any), and make it available on request to appropriate authorities. It is not necessary for this information to be attached directly to virtual asset transfers. Countries should ensure that beneficiary VASPs obtain and hold required originator information and required and accurate beneficiary information on virtual asset transfers, and make it available on request to appropriate authorities. Other requirements of R.16 (including monitoring of the availability of information, and taking freezing action and prohibiting transactions with designated persons and entities) apply on the same basis as set out in R.16
The FATF-proposal is disproportional, technically unsound and uneconomic. We'd better store the citizens data locally and ensure distribution on piecemeal basis, based on solid legal grounds, only when there is a true virtual asset under local definitions.

To the EU I ask to protect my reasonable concerns as a private citizen and not implement the proposal that comes out, until it ensures that my data stay local where they are and are not distributed at large to possibly evil states, dubious countries and their law enforcers.

The latter holds particularly true when we can observe that the chair of the FATF, the US Treasury Secretary, is not living up to his national constitutional obligations to comply with the US law himself.


PS. I noted that the interpretative note to recommendation actually also holds an additional new definition, apart from the main text:
1. For the purposes of applying the FATF Recommendations, countries should consider virtual assets as “property,” “proceeds,” “funds”, “funds or other assets,” or other “corresponding value”. Countries should apply the relevant measures under the FATF Recommendations to virtual assets and virtual asset service providers (VASPs).



Tuesday, January 08, 2019

So what's up with the Google-licenses

Last weeks, we hear all kinds of stories on the Google-license, so let's have a closer look.

Google already has a license since 2007
Most people forget this, but the earliest register entry for Google dates back to 2007 for E-money, and was handed out to Google Payment Limited in London. I blogged about it then, and since then we could see a Google Wallet in the works, Google bucks. The register of the FCA/FSA still has the entry here, demonstrating that it was effective until 19/5/2011. The brand name in the register is for Google-checkout.

Passporting
Then from 19/5/2011 onwards there is the next register entry (with the register later being handed over to FCA by 31st of March 2013). The register entry is still for e-money with passports to other countries. These passports date from 18-5-2011 as our Dutch e-money register shows and the firm is also licensed to perform payments under the PSD1 definition. Offering additional PSD2 services is not part of this license.

As for trading names the FCA entry shows Google Wallet is used from 4-1-2013 to 23-1-2018 and since 20-2-2018 it has the brand names: Google Pay and Google Pay balance in the register as well. So the sum entry of all brand names in the UK register is:

  • Google Pay,
  • Google Pay balance,
  • Google Checkout,
  • Google Payment Limited,
  • Google Wallet

Brexit coming: seek refuge
Now, with the Brexit coming up, there is of course the question how to manage future uncertainty. Many players have been trying to solve the puzzle and my assumption is that the recent moves towards Lithuania and Ireland are Brexit-related. Lithuania is quickly becoming a hot spot for e-money licenses and taking over the dominant role of London in this respect. The license there will allow Google to continue operating in the e-money and payments domain and also offer Payment Initiation and Payment account services.  This makes Google Brexit-proof and PSD2-proof.

Also, we should note that Lithuania does a nice job in offering a digital form of license as well. Have a look at it over here.

When checking the Dutch register, I noted that there is no passport for the Lithuanian entry, but still the UK one. I expect however that the new passporting will become effective in a couple of months, so Google can continue its operations in the EU, now under the Lithuanian passport rather then the UK passport.

As for Ireland, the license is limited to issuing payment instruments and accepting payment transactions, which would point to the fact that Ireland may be the corporate base that also has a role in shaping the future payments infrastructure for Google. It also suits the concept of PSD2 that one has to get the license in the country where it is also used.

Conclusion
Google is since many years in the payments domain and treading carefully, applying different concepts and such. They made themselves Brexit and PSD2-proof by moving to Lithuania (where they are still wating for the passport procedures to finalise) and created additional future business flexibility by applying for payments issuing/transaction acquiring in Ireland.

Friday, March 23, 2018

EC gives open banking to bigtech via PSD-2 and Apple closes its doors to banks in return ?

Just the other day I attended a session of the Dutch Foreign Bankers Association, all about Fintech disruption and innovation. Guest speaker Jesse McWaters, who is the project lead for the Fintech programme of the World Economic Forum, shared his insights into the tech-revolution and how this impacts the business models in financial industry.

Banks, big tech and big data: the uneven battlefield- thanks to PSD2 
One very important observation that he made had to do with the place of banks in the future value chain. They can choose whether to be a product provider or whether to engage in battling for the end-consumer experience by providing multi-party platforms. In this latter approach, it is a big data game. Both banks and big tech will be battling in the same arena where banks need bigtech data and bigtechs need bank data to complete their 360 views of their customers.

In this respect Mc Waters had an interesting question to us, Europeans. He asked if anyone at the European Commission would have understood the huge impact that PSD2 and obliged open banking will have on the competition balance between banks and big techs in the market. Doesn't this skew the balance in favour of the bigtechs without anything in return for the banks?

My response was that in essence the whole open-banking idea in the PSD2 originated from an EC-monoline bureaucratic approach to solving a competition case between one fintech and the European Payments Council (see newsbulletin).

I also sketched that the implicit rule of the PSD2 appears to be that such access without prior commercial contract would be free, even though an analysis from our Dutch competition authority outlines why there is a good case for putting in place a compensation for banks for the access to the customer data. And no, the access is not reciprocal. Big data companies would not have to open up their accounts full of customer information for banks.

Bunq opening up Apple Pay for Dutch customers but then being foreclosed by Apple 
The interesting thing is that we were having the above exchange of thoughts in a week where Bunq had announced to move its systems fully into the could of Amazon (bigtech). And Bunq had also opened up Applepay for its customers. By tweaking the geography settings, Dutch users could start using their phone for Apple-pay.

The fun for bunq-ers didn't last too long though. Apple used its powerful bigtech position to shut out the Dutch bunqers from using Applepay. And my guess is, that its arguments for doing so would be pretty much the same arguments that Sofort heard when they connected to German banks. It would not be safe, there would be no required commercial contract allowing this access and so on.

Time for reciprocity?
It seems that already some time ago the EC course on Bigtech has been changing. We are beginning to realize that we may need to protect our citizen's data somewhat better and that we should not help them avoid taxation. Hence the announcement this week of a 3% tax for bigtech, to make sure they do not get a free ride here in Europe.

It would be very much in line with this new vision towards bigtech if the European Commission mandates open acces to customers big-tech information for banks or any other licensed entity that have the customers permission to request it.

If the Commission truly seeks to achieve a balanced market with proper competition, it should redress the design errors in the PSD-2 and allow banks to ask fees for access and/or allow them reciprocal access to the customer data.

Monday, February 12, 2018

Retailer Albert Heijn demonstrating innovation and making payments the afterthought it is

Three years ago, when we were all pondering the impact of PSD2 on the payments market, I was challenged to outline where I believed the innovation in payments would occur. My response then was that we needn't forget the area that we forget to look when speaking at banks. I was in particular referring to retailers, using technical innovations to provide shopping solutions with an almost invisible payment experience afterwards (based on old school direct debits).

Fast forward to today, where it turns out that by summer (the time we may indeed implement the PSD2 in the Netherlands) Albert Heijn will open its first checkout-free AH to go store. The store, the first of its kind in the Netherlands, will be located at a high-traffic location in Amsterdam and will be deployed with “tap to go” technology that allows customers to pay for groceries with a card or a smartphone without going through the register. 


As the video shows, the card itself will be an NFC-card version of the well-known loyaltycard (Bonuskaart). It uses the identification features of the card to register the purchases, with payments occuring later via the regular direct debit mechanisms of banks.

So there we are: we find a non-bank, identity-based commercial app that uses new technology to make payments the afterthought that they really are. And we will see many more to come.

Sunday, January 14, 2018

PSD2, interchange fee and surcharging: lessons from history

Now that we have the second Payment Services Directive going live in a couple of jurisdictions, the news media are reporting on one of its prominent features: the retailers do not have the right to pass on payment fees to consumers for all cards that are subject to the EU regulation on multilateral interchange fees (MIFs).

As can be seen in the tweet below, the UK chose to extend this to 3 party schemes as well.
As I expect quite some further discussions on this topic (and retailers circumventing the rules by introducing other types of charges to their customers) I figured it would be good to provide some background on the reasoning that is occuring in the area of MIF-regulations. A further look at history would even suggest that we need to take one step further back in order to rethink our analysis so far.

Some background on MIF-modelling (and subsequent regulation)
If we try to assess the arguments pro and con interchange fees, some important thinking in this area is done by Nobel prize winner Jean Tirole, who worked together with Rochet on competition in two-sided markets, such as payment-cards markets. At the heart of their studies is the question what the pricing structures are in two-sided markets and under which conditions and to which extend compensation payments between different sides of the market (interchange fees) are justified.

This work has been at the core of many regulatory strategies but it is subject to a lot of discussion. All market players, whether 3 or 4 party card schemes, issuing banks, acquirers, processors, retailers, consumers and regulators, heavily debate pro's and con's of interchange fees. And in doing so they must use the reasoning and models of Rochet and Tirol. In essence the model tries to determine which cost and interchange fee levels are relevant for competing products/platforms such as cards vs cash.

Do note however that already more than 10 years ago, Brookings Institution released a very good paper on interchange fees, concluding that it is impossible to prove one or the other side of the arguments on interchange fees, let alone determine which level of interchange fee is correct. I think their analysis still stands which means that in the end, interchange fee regulation is more about lobbying and power politics than actual econometric calculations.

Europe's reality under PSD2 and MIF regulation
What is happening under the current payments regulation in Europe, is that:
- a maximum cap is defined for the interchange fee of credit-card and debit-card transactions (reducing the cost to the retailer),
- retailers are forbidden to surcharge for payment costs, when the customer uses a card which has a capped interchange fee (reason being that they may not gain from reduced costs on the one hand while keeping surcharging intact on the other hand).

In essence this means that the EU has bought into the argument that multilaterial interchange fees were being set to high and require regulatory intervention (thus emulating the behaviour of other regulators such as the FED and Australian Reserve Bank). In doing so they accept and embrace academic models which mostly focus on the topic of optimal price regulation in a stationary market with alternative platorms/products.

Historic approach: where did the interchange fee come from in practice?
It strikes me that all the economists at play use an empirical description and mathematic approach to start their reflections on interchange fees. One very obvious element is missing in a lot of papers (except Baxter, who goes at length to discuss history): what were the market players thinking when they wanted to introduce these fees? What is the industry trade-off they are facing?

In Dutch payment history, the banks have been very keen not to disclose their cost/benefit considerations and finances. It was only in 2005 that they allowed McKinsey - as a notary - to have a full look at all internal costs and benefits, in order to draw up a report on costs/benefits based on a full insiders view. I have been personally committed to this effort and helped making it become a reality based on the belief that a lot of misconceptions can be eliminated by being open on ones business model.

However, while this report shows the situation in 2005, it doesn't tell us where the interbank fees came from. The archives of banks do however and to me it is stunning that very few academics in the interchange fee domain have tried to uncover these sources to calibrate their reasoning. Because if they would, they might be able to enrich their analytical approach.

Core question: can we avoid double charging for the retail customer ?
Reading through the old reports of collective groups of players in the Dutch payments domain, we can see an interesting game-theoretic approach to the multilateral fee discussion (which I will be disclosing more in detail in an upcoming publication on the history of Dutch retail  payments).

In essence, in the 1980s the emergence of new players, that were going to be piggy backing on an existing infrastructure without interbank compensation, fueled an existing debate on the distortion of the cost/benefits of banks. Banks with a lot of private consumers would usually bear many processing costs, while banks with corporate customers would reap the benefits. And the winner in the game would be new players, for example investment funds, that would hold pooled accounts for customers funds at a large bank, without having to pay anything for the incoming deposits that came from all the Dutch banks.

Setting an interchange fee in this context leads to:
- a more appropriate allocation and compensation of payment costs along the payments value chain: the entity that benefits from a payment will also bear part of the originating cost,
- the elimination of free-riders in the system,
- interbank understanding that consumers would not be levied fees for instruments where an interchange fee arrangement existed.

The explicit reasoning in a well documented cost/benefit study here in the Netherlands (never officially published as banks made sure to not disclose their thinking) was that it was of course also possible to not introduce an interchange fee system. The involved payment experts noted however that this would lead to bank fees on both the consumer and corporate side of the market. Absent coordination and agreement of reasonable interchange fees, the expectation was that those individual fees of banks to their respective customers would be higher than necessary.

In addition, the bankers expected the corporate side of the market to also add on surcharges and administrative charges to the consumer, which meant that effectively the consumer would then pay twice for payments processing costs: one time at the issuing bank and the other time at the retailer/corporation side, where the payments costs were incorporated into the price of services. Thus, for the Dutch society as a whole, this situation with mark ups on payment processing costs starting at the bank individual level, would undoubtedly be more costly, than the situation where the bank layer coordinates its cost/fee level and thereby avoids double charging of the customer.

The EU-choice revisited: higher payment costs as unintended side-effects of the MIF-regulation
It is clear that the EU perspective on interchange fees is:
- we don't trust the diverging interests of banks and competing card schemes in combination with competition to lead to an appropriate level of interchange fees, so we set an interchange level by ourselves,
- we also don't trust merchants and want to avoid them pocketing the benefits of lower interchange fees at the costs of consumers.

In practice we may now see in Europe that:
- issuing banks have shifted their income base and costs to consumers are increasing,
- acquiring banks will use other fee mechanisms to charge the retailer for the relevant payment costs in the card chain (as long as it exists as such, because a move to instant sepa credit transfers looks pretty enticing of course),
- retailers will use other fee mechanisms or labels (service charges) to charge the consumers for payments,
- consumers are paying higher fees to their issuing bank,
- consumers will effectively be paying twice for the relevant payment costs, in spite of the EU goal that they don't.

Where Europe has chose to intervene in interbanc dynamics, in order to achieve the best result for society, I am not sure if this will indeed work out as such. Yet, I must confess I am not an academic scholar in MIF-models and I get dizzy when reading all the equasions. However, I sense that some of the MIF-modelling doesn't match the actual game-theoretic constellation that occurs in practice.

Personally, I would rather place my trust in the diverging conflicts of interests between closely collaborating banks in an industry (leading to an interchange fee level that is scrutinised among quarreling bank experts) to keeping my payment costs appropriately low, than the good intentions of regulators that expand their interventions towards those mechanics themselves (thereby unleashing the possibility of double charging to me).

Friday, January 12, 2018

Why Bitcoin Core and Bitcoin cash both make sense !

The other day I was reading an excellent blog by Vinny Lingham on the differences between Bitcoin Core and Bitcoin Cash. What I appreciated very much is his attempt to take the heat out of the intense discussions between 'followers' or 'supporters' of these two bitcoin types. And his overview inspired me to reflect a bit more on the reason why actually both Bitcoin Core and Bitcoin Cash make a lot of sense and why there is no need for petty bickering given the open source philosophy on which both protocols are based.

Essential philosophy of Bitcoin: everyone may participate in each role
If we look at the design philosophy of bitcoin, the essential idea is that all participants may act as a user or a producer (miner) at the same time. This is very much in line with all other P2P business models that we see emerging, based on networked technology. With AirBnB, owners of houses also become producers and numerous other sharing platforms seek to achieve the same.

The combined producer/consumer role in bitcoin is very much in line with the design principle to outsmart the centralized institutions of a central bank that issues fiat-currency or with similar players as banks which have their own specific their role in the money-value chain. 

However, the obsession with the dominant role of central banks and banks is so strong in the bitcoin world, that there is too little attention for standard logistics/economics, when looking at the basic proposition of provision of payment services by all to all. So let's have a look at those logistics.

Logistics of any peer-to-peer value exchange system
Anyone participating in payment systems will recognise that a full peer-to-peer system is intrinsically inefficient. Imagine a market fair where all participants are both buyers and sellers. They could pay each other after each sale and hand over cash or other payment instruments. This provides everyone with instant security and payment, but also with cash in hand and time spent to make and accept payments.

Now, if the fair happens regularly, some smart participants will come up with a more efficient idea, which is to not pay after each sale, but to keep track of all payments and net out the payments via clearing cycles. Effectively this is what already happened during the payment days of late medieval fairs and can be witnessed in all kinds of informal or formal constellations (clearing houses etc) that have come into existence since then. 

The conclusion is therefore that, wherever payments are made, at some point in time a new underlying structure will evolve in which specialisation occurs. Those that are very active participants will have different roles and interests than those that only use it sometimes. And it's the active community that will seek some efficient alternatives, while the less-active part will go with the flow. Eventually you will end up with a specific constellation that works for that payment community. 

In these constellations, you will see distinctions in roles such as:
- small users
- big users
- big providers of payment mechanisms
- clearing agents for providers
- settlement agents for providers
- super settlement agents (mostly, but not necessarily, central banks).

The inevitable specialisations in the bitcoin community
If you start a payment system, such as bitcoin, from scratch by allowing everyone to fulfill every role in the system, economics dictate that in the end specialized roles will occur. We've seen this happening early on with the difference between light-wallets and full nodes. Similarly different players have chosen to serve a different part of the value chain.

In the context of an open source protocol, it is just as inevitable that opinions differ on the most efficient way forward. And that's where both Bitcoin Core and Bitcoin Cash have a valid proposition. 

As the blog by Vinnie Lingham outlines, the Bitcoin Core community sticks with the idea that as a design principle all users must be able to run a full node. Scaling and performance issues for the payments mechanism can be solved by adding an outside transaction layer, meaning that Bitcoin Core chooses to be the settlement layer (in classic payment terms). 

Bitcoin Cash on the other hand takes the economics of payment systems as a starting points and accepts the economic reality that in practice, the majority of bitcoin users will not (and cannot) be producers at the same time. Their philosophy is to not tamper with security choices in the area of digital signatures but simply increase the block size, in order to cater for the end-user need to have easy, cheap, simple and fast value transfers. In doing so, they remain IMHO closer to the original idea in the Bitcoin paper (provide an alternative easy fast and cheap transfer mechanism) than Bitcoin Core. 

Stop bickering please !
I am amazed at the amount of energy and emotions that are wasted in the crypto community in what comes down to petty bickering. Whether this is between Bitcoin Core and Bitcoin Cash or between whichever of the other cryptocurrencies, blockchains or DLT-designs. 

It is and was inevitable for the bitcoin protocol to run into the standard efficiency problem for evolving payment and settlement mechanisms. It is just as inevitable that different solutions exists and both Bitcoin Core and Bitcoin Cash are equally valid. The one just wants to end up becoming a settlement layer provider while the other wants to be stick to the user layer of payment systems. Where is the harm in that?

The open source philosophy means that we will see in practice which solutions play out best for the ecosystem(s) and rather than bickering about advantages or disadvantages we should cherish the variation and choice that the open source philosophy brings. 

Monday, December 04, 2017

Cryptocurrencies, initial coin offerings (ICO's) and tokens: we remain puppets on someone else's strings !

Screenshot www.flippening.watch 
Now that the bitcoin price is heading towards $10K and beyond, we see another wave of interest in the cryptocurrency domain. Regulators warn about the risk of Initial Coin Offerings and news bulletins keep on re-discussing what is happening in this space.

It's a matter of private currencies and tokens
What we are looking at in the cryptocurrency, blockchain and distributed ledger space is that seemingly public and democratic technologies are being put to work for specific groups of individuals/companies. While the technology may be sold as serving a public purpose, in the end it's just a variation of any medieval local Duke issuing his private currencies to the population: profits to the Duke and potential losses to the public.

What is happening now is a convolution of low interest rate regime, overhyped media and greed, leading to private individuals scrambling for profits in the area of cryptocurrencies. They can do this either by investing in the cryptocurrencies themselves or as an entrepreneur by sharing a mining pool, setting up exchange's or trade functions or developing new token types or blockchains. And as with the gold rush, it will be the sellers of shovels and buckets that will in the end really make the profit.

Beware: we remain puppets on a string
While in the initial stages of the bitcoin blockchain, we could see a whole lot of alt-coin scams, we are now seeing a range of ICO-scams, as people are exploiting the ethereum capabilities to venture out in the cryptocurrency world. Only a very few of those will survive and the rest will disappear just as the nonsense alt-coins did.

So, whoever participates in cryptocurrencies or ICOs deliberately hands in governance to an unknown constellation of companies and individuals. Or as I put it in this 2013-blog:
The redistribution of value that can occurs with these new currencies may look democratic, but that is a wolfe in sheep' s clothes. Effectively the new currencies are and will be the domain of private individuals trying to seek private gain rather than anything else. And there is no guarantee whatsoever that this constellation will have the interests at heart of all people in society. 
It will be Darwins' survival of the fittest all over again, which will exclude certain groups of citizens from participating fully in society. As democratic as a crowdbased currency looks: you will still be a puppet but on a different string, with unknown gains being made by unknown players in the value chain of this collective currency. 

To bubble or not to bubble? 
The question that is now on the public table: is the value of bitcoin a bubble as the tulipomania? Despite the tendency to say yes, I would argue that the answer could effectively be no.

First of all, the whole western money system is a bubble right now, as central banks have inflated our financial systems to an enormous degree with the Quantitative Easing. We should realize that neither bitcoin nor any other good in this society has its proper value right now. Thanks to these central banks, the amount of money that I get in my savings account has been too low for almost a decade now.

Second, within this skewed monetary world, market-forces do still apply. They will also apply to bitcoin-core so that with each fork (bcash) or new blockchain (ethereum, eos) a potential new competitor may turn out to become the winner. The word used for a shift in this momentum is: the flippening (as sentiment may flip to different assets or infrastructures in a very brief moment of time).

The interesting thing is of course, that in this digital world, anyone can monitor this real-time on sites such as Flippening Watch. It will be like watching the strings of the puppets move, without knowing who the real players are.

Sunday, November 12, 2017

Why some countries started using cheques and others chose giro

These days I am busy writing a book on the history of Dutch retail payments. The focus of the book is on the dynamics of the Netherlands, without the aim to compare with other countries or to explain country differences. Still, the process leads to some observations that I would like to share as they may be useful for other researchers in the field.

What explains the origins of giro and cheque countries?
One of the basic facts in retail payments is that there is a structural difference between so-called giro-countries (Austria, Switzerland, Japan, Germany, Netherlands, Belgium etc) and cheque-countries (France, US, Canada, Australia).

In his excellent dissertation on payments and network effects, Gottfried Leibbrandt sets out to answer this question. After a thorough investigation of literature, he concludes that network effects are an important factor that helps explain the typical development path per country. At the same time, he finds it hard to trace the origin of the difference between cheque- and giro-countries
1. There is no satisfying explanation for the country differences. Empirical studies find that country idiosyncrasies rather than variables like GDP and crime explain the differences in instrument usage.

In a similar vein, the first BIS working group on Retail Payments observes:
Use of different retail payment instruments in the so-called cheque countries and giro countries can be explained by the differences in: 
• concentration of market supply among traditional providers of retail payment services; 
• financial incentives for providers with respect to debit and credit transfers; 
• nature of the risks in the value transfer processes for the two types of payments; and 
• legal framework and regulatory environment.
Of course the above list is long enough to be right and the report also mentions numerous French regulations that influenced its use, in particular also the rule that payment with cheques should be free to the people. And indeed this legal factor must not be forgotten.

Where is the cheque in the Netherlands?
In the Netherlands, a well functioning system of so-called cashiers notes existed, alongside regular cash for quite some time. Then in 1814, the establishment of the central bank and the introduction of bank notes introduced competition for the cashiers. In the early 1830s one of the cashiers fought a heavy legal battle to preserve the use of its cashiers notes, but eventually gave in to the reality that the bank notes were becoming the standard.

Ever since, the central bank was careful not to obstruct the cashiers and bankers too much in their business operations. So as a practical measure, the central bank took care to set its fees for deposits and discounted bills of exchange at a less competitive rate than the market. But fact of the matter remains that a possible candidate for a privately issued Dutch payment cheque, was no longer available.

Some parliamentary proceedings suggest that the legal rules with respect to bills of exchange 'wissels' were insufficient to really create a sound basis for the use of cheques as a payment instrument. As a result the market used bank notes in combination with clearing arrangements.

This intrigued me so when looking for more information on the topic I encountered a college book (by Mr. W. Molengraaf) that suddenly shed more light on the situation in other countries.


What Molengraaff states is that cheques came into existence in the United Kingdom as 'sight-bills' on a banker. They were brought into circulation as an alternative payment instrument to circumvent the stamp duty that would apply to a regular bill of exchange. He continues to state that also in France the cheques owe their existence solely to tax considerations.

Up next I figured I would take a look at a US stamp duty register from 1866. And we can indeed see the difference there. A bank check would cost 2 cents, regardless of the sum involved, while a bill of exchange would cost 5 cents and possibly more when higher amounts were involved.



To me it's clear that stamp duty may well be an important part of the puzzle, explaining the difference between giro and cheque-countries. But we should add an analysis of market structure, competing instruments to complete the picture.

The particular Dutch situation.... 
In the Dutch situation we will see that the discussion on setting up postal giro services started around 1902, already with references to the situation in other countries. It took us more than 10 years to decide on the establishment of the postal giro services and indeed market structure and some regulatory capture may have influenced the discussion. Also it became pretty clear that the incumbent bankers and cashiers did not want to move forward with their own version of bankers giro. But that's stuff for a next blog.

In the mean time, if you enjoyed this blog, you may want to let me know that you're interested to be informed when my book on hundred years of Dutch giro payments will be published. This will help you understand more of the intertwined dynamics that are at work when retail payment systems and instruments are developed.

Just send an e-mail here and I will notify you when the book is coming up.

Tuesday, October 17, 2017

DAO-fork at odds with Ethereum terms and conditions

Last month, Antonio Madeire nicely summarized the discussion on the DAO-hack and the fork which brought Ethereum classic into being. I remember that my contribution to the discussion at that time was that the Ethereum developer community should not revert to a hard fork but to the judge and/or arbitration.

Governance and terms and conditions
The other day, I was discussing with Ian Grigg, a long time mutual topic of interest: making technology work by adding proper arbitration to smart contracts and agreements. This can even be done in code, as he had demonstrated way back in the 1990s in his ricardo system.

This prompted me to actually take a look at the Ethereum terms and references to see what it said about disputes. Well, have a look yourselves:

All disputes or claims arising out of, relating to, or in connection with the Terms, the breach thereof, or use of the Ethereum Platform shall be finally settled under the Rules of Arbitration of the International Chamber of Commerce by one or more arbitrators appointed in accordance with said Rules. All claims between the parties relating to these Terms that are capable of being resolved by arbitration, whether sounding in contract, tort, or otherwise, shall be submitted to ICC arbitration...... 

.... And so on.

What does this mean for Ethereum governance?
While I hugely appreciate the development of Ethereum and all the efforts that have gone into it. it does strike me that when push came to shove, the developers brushed aside their own terms and conditions. The use of Ethereum was instrumental to setting up the DAO, so why not revert to the ICC Arbitration?

My guess would be that, not being lawyers or into governance, the developers used the tools that came in handy and quickly. Alternatively, it might be the case that they might have invested in the DAO themselves quite considerably.

Regardless of the exact reasons behind not using the dispute resolution mechanism, the paradox is that, while there is a formal basis for dispute resolution under Ethereum, the likelihood exists that in future instances of trouble, the developers will again fork their way out of trouble.

Create an additional dispute resolution layer.
Any practical use and implementation of Ethereum should therefore come accompanied with additional agreements on dispute resolution, so that organisations that cooperate on the basis of the ethereum blockchain create their own governance basis.

Thursday, September 21, 2017

Ceci n'est pas une 'payment instrument': a reflection on fuel cards and the PSD2

In september this year, the FCA published its policy statement and approach document on the PSD2. I've been eagerly watching this document to find out what their final take would be on the understanding of the limited network exemption in relation to the specific nature of fuel cards. Because there is more than meets the eye here.

In essence, some fuel cards effectively function as a purchase button on a website. They don't initiate payment orders at all. They would thus fall outside of the PSD2-scope, as any other shopping-button on websites. Due to a twist of faith however, the bank supervisors seem to be keen to ignore this reality for fuel cards and bring those under the PSD2.

In this post I will highlight the flaws in this approach and conclude that the result is that if all regulators start re-writing the definitions themselves, we better label the PSD2 the Purchase Service Directive (see also the full  and more elaborate analysis on the subject here).

Background
In the original payment services directive article 3k provided for a proportional application of the PSD1. Instruments with a limited geographical reach and scope, such as store cards and fuel cards were not subject to its provisions. The exemption 3k) was thus called the limited network exemption.
3 (k) services based on instruments that can be used to acquiregoods or services only in the premises used by the issuer orunder a commercial agreement with the issuer either withina limited network of service providers or for a limited rangeof goods or services;
In its proposal for the new version of the PSD, the Commission claimed the existence of payments systems, waivered as “limited networks” with massive volumes, which imply greater risk and no legal protection for payment users as “feedback from the market”. However, this feedback was not really a result of the external impact analysis onthe economic impact of the PSD1

What could be seen though is that the interpretations of local supervisors ranged from strict to very lenient, which distorted the playing field in Europe. In addition, some observers noted that there was a strong desire by supervisors to have stricter rules for in particular the fuel cards market (see the mystery of unregulated massive payment volumes, as discussed in the Paysysreport of March 2014).

In the end, the net result was a very strict version of article 3k in the PSD2, to ensure that its future application would be for truly limited networks only. In addition, any organisation that uses this specific exemption has to notify the supervisor. But let's take a good look at the pre-amble and the exemption text.

The definitions
The pre-amble states that payment instruments covered by the limited network exclusion could include store cards and fuel cards, but it isn't conclusive. They could, but they could also not.
(14) Payment instruments covered by the limited network exclusion could include store cards, fuel cards, membership cards, public transport cards, parking ticketing, meal vouchers or vouchers for specific services, which are sometimes subject to a specific tax or labour legal framework designed to promote the use of such instruments to meet the objectives laid down in social legislation.
Where such a specific-purpose instrument develops into a general purpose instrument, the exclusion from the scope of this Directive should no longer apply. Instruments which can be used for purchases in stores of listed merchants should not be excluded from the scope of this Directive as such instruments are typically designed for a network of service providers which is continuously growing. The limited network exclusion should apply in combination with the obligation of potential payment service providers to notify activities falling within its scope.
Article 3k is actually more clear, certainly in comparison to the previous version. It now refers explicitly to services based on specific payment instruments
(k) services based on specific payment instruments that can be used only in a limitedway, that meet one of the following conditions:

  • (i) instruments allowing the holder to acquire goods or services only in the premises of the issuer or within a limited network of service providers under direct commercial agreement with a professional issue;
  • (ii) instruments which can be used only to acquire a very limited range of goods or services;
  • (iii) instruments valid only in a single Member State provided at the request of an undertaking or a public sector entity and regulated by a national or regional public authority for specific social or tax purposes to acquire specific goods or services from suppliers having a commercial agreement with the issuer; 

The legal conclusion is thus: first you need to have something that is a payment instrument and then it may fall under a limited network exemption. 

The FCA's approach: let's not be clear about the payment adjective
The above may not be how the FCA are looking at it. Both in their consultation and further guidance they seem to that 3k is written as pertaining to all instruments, not just payment instruments. I pointed this out in a response to the consultation (see this separate blog) and asked for further clarification.

Yet, their feedback document, doesn't mention anything on this definition question at all, which is a bit disappointing given the timely and good job efforts that the FCA usually put in with all their consultation work. So the ambiquity stays: while the notification forms clearly outline that applications must clarify the nature of the involved payment services and payment instruments, all the guidance does is steer towards an understanding of the 3k article as pertaining to all instrument (as under PSD1). 

This still leaves us with the question: when would a fuel card qualify as a a payment instrument. Or is it just as exempt from the PSD as a purchase button in an Internet-webstore? 

Fuel card as a payment instrument or purchase device?
Let’s have a closer look at the workings of a fuel card and what it does in terms of business processes. Generally speaking, fuel cards are delivered by oil companies to corporate fleet owners, sometimes distributed via resellers or co-branding arrangements. They effectively are a tool that validates the legal competency of its holder, to take out goods/services from service delivery stations.

The company to which the cards are provided takes full responsibility for all services/goods delivered to the users of the cards and receives a monthly overview of all purchases done with the cards. It can set usage levels per card, ensuring that no more than a certain amount of goods and services are to be delivered to the cardholder. It can also set the range of goods to be delivered from narrow (fuel only) to wide (fuel and shop goods).

Every month, the fleet owning company receives an invoice with an overview of all purchases made and the rebate applied (mostly volume based). This specifies the purchases made in the network of the oil company itself as well as those in other networks and by other service providers. These other networks of service stations may also deliver goods/services to the card holder. What happens in practice is that prior to the actual delivery, the cardholders’ oil-company buys the whole service/goods package that the card-holder wishes to take out at the selected other networks with whom the oil-company has struck delivery and service agreements.

This results in a chain sale of goods/services from:
-              the service station dealer to its country organisation,
-              the service station country organisation to the oil company national organisation
-              the oil company in a country to the corporate client that distributed its cards to the employees.
While technically there may be many variations to this flow, it does serve to achieve an important effect in VAT-terms. It allows the involved oil companies and networks to reclaim the relevant VAT from local authorities and thus lower the end-fee to the corporate fleet-owners.

What's the difference between purchase and payment?
In the table below, I've summarized the functional difference between the use of a payment card or a purchase card at an oil station.

Using a purchase tool
Using a payment instrument
Used to instruct the retailer or service station to deliver goods/services
Used to instruct the bank to make a payment to a third party bank account
The amount to be paid is unknown. At the end of the month, rebates are applied and the reconstruction of what the actual equivalent price at this moment of sale would have been, is always a mathematic reconstruction.
The amount to be paid is clear.
Authentication of the card holder equals the right to receive goods/services up to a certain threshold
Authentication of the card holder equals the digital signature of the payment transfer
Positive response by oil company equals the formal sale of the services/goods from service station to oil company and the mandate to provide the services/goods to the card holder
Positive response equals the proper processing of the payment instruction
Holder of purchase instrument is not (by definition) authorised to give payment orders that relate to the billing account of the fleet owner
Holder of instrument is by design authorised to give payment orders from that account to the payees account
Holder only receives proof of purchase / delivery but not proof of payment
Holder receives proof of payment and possibly also proof of purchase
No cash-back possible
Cash back might be possible under the rules of the cards-account
Oil company may design its own purchase, control and billing procedures, use its own set of purchase tools and may set its own acceptance and risk parameters. Intercompany delivery agreements will apply.
Card is a payment instrument and payment transactions with it fall under legislation (PSD) and payment brand regulation, with bank specific acceptance and risk parameters
VAT-recovered and rebate applied to purchases by all cardholders of the fleet-owner
VAT-recovery not included
Aggregated invoice for goods and services delivered, sent to the corporate treasurer of the fleet owner, and paid for using the direct debit instrument
Periodic account statement for payments made or (as in the case of credit cards): aggregated invoice for total value of payments made, followed by direct debit.

While the bank card ticks all the boxes, the fuel cards as outlined above, do not qualify as payment instruments under the current Payment Service Directive. There is no request being made to place, withdraw or transfer funds, hence there is no payment transaction, no payment order and no payment instrument. Hence, article 3k is nowhere close to being relevant. 

What will happen to the fuel cards niche?
As the editors of the Paysys report outlined earlier in March 2014, there may exist a hidden regulatory agenda in Europe to capture fuel cards under the 3k article of the PSD2.

This seems to be exactly the discussion right now for the relevant stakeholders around this subject in Europe. While technically the legal argument is straightforward, leading to fuel cards being out of scope, some lawyers point to the spirit of the article or the statements of regulators (in whichever respect being made) to claim that fuel cards do fall under article 3k and require notification.

As in many situations, it's not the final legal analysis that is relevant but the legal uncertainty. Arguing the above case with a regulator may take too much time and is not the preferred option for risk-averse large issuers of fuel cards. So we may well see some players in the oil industry ending up not arguing their legal case and abiding with a flawed regulator view that takes fuel cards into the exemption scope of PSD2.

The wider implications: Payment Service Directive becomes Purchase Services Directive
If the fuel card case is not being challenged in courts, it sets an interesting precedent. Because in essence, there is no analytical difference between the fuel card and PIN mentioned above and the user-id / password combination that is in use by retail customers that are shopping at websites, using purchase buttons. Both tools are and should be out of scope for a payment service regulator. Just arguing that the instrument looks to much like a payment instrument is just not enough. Ceci n'est pas une pipe 'payment instrument'.

Forgetting the adjective 'payments' in article 3k means that the second Payments Services Directive may well turn into a full swing Purchase Services Directive. And by the looks of it, this is what the FCA may be doing in the fuel cards niche right now. This leaves the rest of the market wondering if their niches may follow at some point in time. 

Supervisors should however not cross that Rubicon and avoid transforming the PSD2 into a Purchase Services Directive. They should stick to the legal definition and if they don't like the consequence should not take justice in their own hands by forgetting adjectives that stand in the way of their (hidden) agenda's.  

Let commerce be commerce and payments be payments.