12 Comments
User's avatar
Martin Chorzempa's avatar

If it's centralized like this, then why is any blockchain required? Why other than PR is a coin necessary? Couldn't they just offer intraday repo as a service, with the treasuries in custody and bank deposits instead of JPM coin on the other end? If this is the future of crypto, then I'm wracking my head trying to figure out why it should exist...

Expand full comment
Daniel H. Neilson's avatar

Agreed - I will try to clarify this in a future post. I don't think blockchain is the point. In marketing terms maybe it helps them make a pitch. Customer: "crypto?" JPM: "blockchain repo."

But the possibility of programmable contracts would be something genuinely new. The whole triparty system is set up to keep the money leg and the collateral leg connected. If you could make them simultaneous in a trustless way, I could see how that would be attractive to repo market participants.

Expand full comment
Tycho's avatar

Not sure how this blockchain based repo is different from a repo using a centralised database. The blockchain is built by the JP Morgan owned Onyx team and I assume JP Morgan is the only one running nodes. I don't understand why anyone would use a blockchain when it adds no value of decentralisation. This seems like a nice PoC but nothing more

Expand full comment
Daniel H. Neilson's avatar

Agreed. I'm not certain that JPM runs all the nodes, though. I haven't seen a lot of details yet about the system.

But see my note to Martin Chorzempa's comment above - programmable contracts might be sufficient rationale. Also we could just think of the whole intraday repo system as a public prototype: it could be worth it to JPM to have the expertise in-house by the time the rest of the world reaches consensus on what form and role these technologies will take.

Expand full comment
Tycho's avatar

Yeah that makes sense! Programmable contracts are definitely the killer application here - they can be moved to a sufficiently decentralised base-layer later which is what other crypto app developers are also doing to some extent

Expand full comment
AncientLisper's avatar

I'm a retired software engineer, and have yet to study blockchain in sufficient detail. But I tentatively agree www.stephendiehl.com/blog.html, a software engineer who argues that centralized software operated by a trusted agent is unambigously better than decentralized solutions for every use case except crime. Specifically wrt instant clearing and smart contracts: smart contracts are essentially like what traditional databases call stored procedures. You can certainly make an Oracle database run such a procedure at a specified time to close a repo, and the database doesn't need to wait for 5 o'clock to update.

So I'm interested in any arguments that derive business benefits concretely and specifically from the use of blockchain. Diehl's point is that the flood of "decentralized woo woo" around blockchain never really gets down to explaining that. A good driver-question is: what trust relationships are supposed to be obviated by the use of blockchain? It seems clear to me that you'd better trust JPM anyway if you use their repo exchange, so why would it be better or more worthy of your trust if they use blockchain as against more conventional (and reliable) I.T.? Nor is blockchain needed to indemnify you against your counterparty; exchanges do that now.

My own naive theory is that there is indeed no technical benefit, but all that decentralized woo woo attracts users to your shiny new instant-clearing smart-contract repo exchange. And JPM is doing this because they get to handle your assets in flight and hold them at rest, and if, as first-movers, they end up with the biggest repo exchange, that's a *lot* of assets.

If you've discussed this already, please point me at it; I haven't read all your old posts yet. Good blog though!

Jim Goodwin

Expand full comment
Daniel H. Neilson's avatar

I think I basically agree with you - programmable contracts are the benefit, not blockchain per se. Recently I wrote https://neilson.substack.com/p/stablecoin-report and https://neilson.substack.com/p/regulating-defi which get into some of these issues. Thanks for reading!

Expand full comment
AncientLisper's avatar

Dan Nielson's most recent post and the video by Christine Moy to which he refers have helped me understand this a lot better (thanks Dan!). What is happening is indeed a wholesale replacement of the entire infrastructure for interaction between banks, to make things faster, cheaper, and more verifiable, particularly wrt regulatory compliance. As a software guy, I understood intuitively why the legacy protocols are too hard to upgrade incrementally, but why blockchain? Christine Moy's example of one bank asking another to verify an account makes this a lot clearer to me. Start watching at about 5:00.

She doesn't point it out, but smart (i.e., programmable) contracts fit into that example too. When the destination bank, 12 time zones away, wakes up, finds the request, and confirms the destination account is legit (a manual action), then the transaction can execute automatically and immediately, even if NYC is asleep.

Expand full comment
Daniel H. Neilson's avatar

Thanks for reading and commenting James, your questions helped me see that there was more to figure out after this post.

I've written about programmable contracts before (e.g. https://neilson.substack.com/p/e-cny), but I am feeling the need for another post soon that looks at JPM's programmable contracts, as you mention here. Stay tuned!

Expand full comment
rsm's avatar

Is it just me, or are they doing this so their rates are invisible to regulators?

Expand full comment
Daniel H. Neilson's avatar

I think JPM is a market-maker here, not a lender, and I think it's too small to make it worthwhile to hide anything. I would say they're testing the tech and building proof of concept for a real use for programmable contracts.

Expand full comment
rsm's avatar

Can the Treasury tokens trade on secondary markets? Will JPM have more control than the Fed?

Edit: can programmable contracts facilitate t+0 settlement on retail financial trades (an issue in the GME story)?

Expand full comment