What are the “whole shit ton of reasons” why shorter settlement periods are so difficult?
In particular, I’m curious why we didn’t jump straight from T+3 to T+1. Even if T+0 is especially challenging, what would make T+1 substantially harder than T+2?
As with all these changes, it's mostly about coordination and legacy systems.
Just as Y2K was a challenge and just as the 2038 date will be a challenge, it's all about pushing changes through legacy systems. ascii->unicode, ip4->ip6, python 2->3. We all know the drill. We've all lived through these things.
It's rarely a technical problem. It's about coordination across firms, domains, people, and systems that may not be known ... until they break.
Yes. I used to work at an investment bank. When the market moved from T+5 to T+3 it was a major project for all internal systems to be adjusted. It wasn't just a matter of changing a SETTLEMENT_DAYS macro in a header file. And even afterwards, it wasn't that, because there were different systems written in different eras and in different languages.
Multiply this by however many thousands of firms, all with their bespoke back-office systems. and it takes time.
Coordination problems can become technical problems - or be caused by technical decisions.
Moving a system/network of actors from one system to a new, incompatible system? You need to coordinate the switch so it all happens at once. Can’t coordinate? Then you need a compatibility layer between the two systems.
My understanding is that the delay is due to humans in the loop. Most of the clearing is automated, but if the numbers don't add up, humans still intervene to figure out how to reconcile the differences.
Other comments have claimed that the humans in the loop are because people insist on hand calculating everything. Not for when it goes wrong, but in the first place.
In particular, I’m curious why we didn’t jump straight from T+3 to T+1. Even if T+0 is especially challenging, what would make T+1 substantially harder than T+2?