So customer A is doing stuff where you need to put up a collateral with some counterparty.
You can't use customer B's money (this is a key assumption that might be missing - it's all about the use of other customers money) for that collateral because, well, that collateral might not get returned in certain cases - that's kind of the point of having a collateral. You'd lose that collateral if the counterparty goes belly up (insolvency, fraud, whatever), and, most importantly, you'd lose that collateral if you become insolvent. That's not OK - this is regulated so that you are required to ensure separation of "your money" from "customers money that you're holding on their behalf", so that the customer's money is untouched and unclaimed even you go bankrupt. It's not your money, it's the customer's money that you're investing on their behalf, so you can't put it up as repayment or collateral for your liabilities; and you can't put one customer's money as repayment or collateral for another customer's liabilities.
But why can't customer B's money be used as collateral for customer B's trades? I guess because the collateral is all comingled when it is supplied to the clearinghouse?
That's my question as well. I think it's because the customer capital has to be transferred to the counterparty that issued the sell. I imagine that it makes things complicated if the DTCC has to send over 1-10% of this amount and then the brokerage that executed the buy has to send over the rest so, if I had to guess, they did this for simplicity's sake. Btw the requirement to not use customer collateral is enforced by the DTCC and not at a brokerage level discretion.
>Btw the requirement to not use customer collateral is enforced by the DTCC and not at a brokerage level discretion.
No, it's an SEC rule.
Broker dealers can fail, sometimes due to malfeasance but sometimes simply due to bad management or even bad luck.
In a system with a central clearing counterparty (in this case, the NSCC/DTCC), that organization mutualizes those risks by acting as "the seller for every buyer and the buyer for every seller": the process of novation splits each trade between buyer and seller into two. i.e. the seller sells to the NSCC and the buyer buys from the NSCC. Effectively the NSCC is guaranteeing the trades.
The collateral is effectively a security deposit that varies in size depending on how much risk a particular broker dealer is bringing into the system as a whole.
Does it make sense yet why this can't be done with client money?
I think the problematic scenario is when the customer wants to buy a security using the proceeds of a recent sale, which hasn't settled yet. Since the proceeds haven't made it to the broker's bank account yet, the broker would use their own capital in the interim.
You can't use customer B's money (this is a key assumption that might be missing - it's all about the use of other customers money) for that collateral because, well, that collateral might not get returned in certain cases - that's kind of the point of having a collateral. You'd lose that collateral if the counterparty goes belly up (insolvency, fraud, whatever), and, most importantly, you'd lose that collateral if you become insolvent. That's not OK - this is regulated so that you are required to ensure separation of "your money" from "customers money that you're holding on their behalf", so that the customer's money is untouched and unclaimed even you go bankrupt. It's not your money, it's the customer's money that you're investing on their behalf, so you can't put it up as repayment or collateral for your liabilities; and you can't put one customer's money as repayment or collateral for another customer's liabilities.