August 12, 2007

The fundamental _barrier to entry_ in the business of payment systems

Dave Birch reads Leo van Hove's new article "Central Banks and Payment Instruments: a Serious Case of Schizophrenia":

This article analyses the competition between cash and payment cards against the backdrop of the dual role of central banks - as issuers of cash and as institutions with a mandate to foster the efficiency of payment systems in general. It is argued that this dual role results in a number of policy dilemmas, namely concerning pricing, traceability of banknotes and the choice of denominations of coins and banknotes. On a general level, the article argues that central banks should place greater emphasis on improving the efficiency of retail payments and less on protecting their self-interest. More concretely, the article repeats the suggestion - originally put forward in VAN HOVE & VUCHELEN (1996) - that the ECB should place the upper limit of its banknote series at EUR 50 instead of EUR 500. It is also argued that policy makers should explicitly foster the use of cost-based pricing and in particular create a legal environment that makes it possible for commercial banks to start using it.

Key words: payment instruments, central banks, cash, banknotes, payment cards, public policy, efficiency.

And Dave does the summary, which is probably very good if you want a shorter read. (Indeed, I haven't read the paper myself!) He raises one point that I take issue with:

Meanwhile on the supply side, the barriers to entry are significant. It is simply not the case that anyone can start offering new payment instruments, as distinct from payments processing services.

I don't think that's true ... although I'm not at first blush sure why. Let's work it through.

Certainly the barriers to entry are significant. I once counted 8, then lost the napkin. By way of a few seconds thought, there is:

  • regulation, the subject of Leo's article.
  • robust technology, client-side and server-side.
  • network effects, or as described by Dave, the chicken & egg problem.
  • the application choice.
  • sustainable system of value backing (e.g., reserves).
  • governance.
  • border transactions / primary sales.
  • theft.
  • ...

But, even though it might be granted that those barriers are present, we now have more information on how to do this. Substantially more information, indeed, I claim, solutions abound for each of the problems. We can now count around 10-15 years experience with successful open payment systems, more if you are old enough to remember the various DigiCash and other experiments.

I claim, if you give a competent FCer any of the above barriers, they'll find you a solution.

I concluded in or around 1995 that it takes around 1m of capital to put in place a payment system built primarily in software. (I also said, 100 times that amount, if doing it with smart cards or the like.) Since then, I have seen little to change that view. But, we have seen so many systems that have failed in ways that prior systems also failed, so obviously something is wrong.

Here's what I think is wrong: If there is a barrier in this special field of financial cryptography, beyond the 1m of pocket change needed to build an open payments system, it is in the FC7 thesis: there are too many disciplines to cope with.

Is knowledge the insurmountable barrier?

I like to think it simply requires an open mind, and willingness to read the experiences of others. The evidence against insurmountability barrier is over on the right hand side of this blog, where you can see a personal list of those people who've seen enough to know the layers and work through all the barriers.

But, as time goes on I lose the confidence that more people becoming financial cryptographers is an answer of utility. The evidence for that position is that only a few of those FCers are ever consulted by operators of payment systems. Nor are they often employed by banks, who classically operate payment systems, and occasionally build new ones. Pretty regularly, both repeat the errors of the past, and lose bucket loads of money, but that's no barrier, it seems.

So I guess Dave is right. There are insurmountable barriers to payments systems, and people can't just enter the business. I still don't know why, but it seems as if this institutional knowledge barrier exists, and is insurmountable, even though we as individuals know how to do this.

Posted by iang at August 12, 2007 02:02 PM | TrackBack

and now for something completely different ;-)

Regardless which e-mail client I use (ms outlookup, apple mail, squirell webmail), the HTML-markup is not interpreted and clutters the text in your e-mailed newsletters. That's a pity.

Posted by: kr, Twan at August 13, 2007 10:58 AM

On behalf of the entire service department I apologise a thousand times for our substandard performance.

:) that's because it is text. You see, the crappy blog I use allows me to format in text/html at the same time. But, the mailout, only sends out the text in raw, not the HTML. So, I have a real time-waster problem: write in pure text, publish, then mailout, then *add* all the HTML.

The end result is I have to take an extra 20-30% of time for a reasonably short post ... and a short post takes an hour or so ... I am always trying to speed up the process by inserting the HTML where it doesn't interfere with readability beforehand.

I intend to change blog software, but switching costs, switching costs..... :)

Posted by: Iang at August 13, 2007 11:00 AM

one of the big infrastructure issues in the 70s was interchange and the associations. before that both the merchant and consumer had to be with the same institution. this was not just a technology interconnect problem but also contractual and legal issuef. the value-added networks to address the interconnect problem have somewhat been obsoleted with the growth of the global internet. however, the legal and contractual issues still remain.

for instance, in some countries, at least in the late 90s, and possibly still true today, required bilaterial, contractual agreements between every accepting merchant and every issuing consumer institution.

the associations allowed merchants to have (contractual) agreements with their financial institutions, consumers have (contractual) agreements with their financial institutions. then all financial institutions have contractual agreements with the associations (as opposed to every individual financial institution required to have bilaterial contract with every other financial institution) This reduced N*M problem to a N+M ... aka N are number of merchant financial institutions and M are number of consumer financial institutions (with each on the order of tens of thousands)

old posts discussing the N*M issue: New Method for Authenticated Public Key Exchange without Digital Certificates The Worth of Verisign's Brand The Worth of Verisign's Brand

the issue of interchange/association (or lack there-of) also reared its head in the digicash trials ... being limited to a single, common institution that served both the merchants and the consumers. disclaimer ... in the digicash liquidation ... we were called in to evaluate the patent portfolio.

Posted by: Lynn Wheeler at August 15, 2007 11:39 AM

The problem is both technologically complex and organizationally complex, so requires someone who is both a pretty good geek and a pretty good suit.

Trouble is, the problem also requires critical mass, hence widespread consensus, hence a committee, and a committee is dumb as a rock.

Posted by: James A. Donald at August 15, 2007 04:08 PM
Post a comment

Remember personal info?

Hit preview to see your comment as it would be displayed.