> This too smells fishy and rigged to me. Modern databases don't need T+2 to achieve consistency.
The problem is, you have lots and lots of databases-- there's not one central ledger of every position. Settlement is effectively an atomic commitment protocol, so that if you are selling a stock you expect to receive from one firm to another, and someone else fails to deliver, you need to be able to unwind the downstream trades-- but also need all transactions to eventually become final.
The T+1, T+2, T+3 periods are convenient in human units, because sometimes it's desirable to have a human in the loop when unexpected things happen.
If only a public shared ledger was available for these institutions to settle trades on.
People love knocking Bitcoin here, but this is why the technology exists. No reason to continue letting these institutions run smoke-and-mirrors with everyone's money
Maybe eventually. Right now, there's a whole lot of problem with scaling for distributed ledgers. Big markets do 20M trades per day-- and that's just across the market itself.. This is like 50k TPS average (not peak) across a trading day; some distributed ledgers do better but compare to several transactions per second across Bitcoin and a few thousand TPS per second for most performance-oriented ledgers.
You might also not want to have a public ledger of all the ownership interests.
It also doesn't exactly solve the problem. You still have the prospect of double spend, etc, until commitment/ledger settlement. And for high stakes transaction you probably want humans in the loop talking about what to do when these things happen instead of some automated policy.
Don't get confused - other markets have end of day settlement, stocks are a different animal but its not because humans in NYC aren't familiar with databases and transaction processing.
That's the thing. If we were to start from scratch building a stock market or banking system today, we might take different approaches.
But there's a market that exists at substantial scale that was originally built around T+14 settlement (and then T+7, T+5, T+3, all the way down to T+2 which is really still brand new). And while there's substantial risks and costs that come from delayed settlement, there's also substantial benefits... and an entire global financial system that relies upon the properties we get from these settlement delays. Another equilibrium being perhaps slightly better overall doesn't necessarily justify reinventing the wheel.
I suspect we'll get to T+1 settlement for stocks and many banking transactions in a few more years (we're already there for options)... and then it may be time to discuss the benefits of various types of "instant" settlement for some subset of financial activity.
The problem is, you have lots and lots of databases-- there's not one central ledger of every position. Settlement is effectively an atomic commitment protocol, so that if you are selling a stock you expect to receive from one firm to another, and someone else fails to deliver, you need to be able to unwind the downstream trades-- but also need all transactions to eventually become final.
The T+1, T+2, T+3 periods are convenient in human units, because sometimes it's desirable to have a human in the loop when unexpected things happen.