This is only true in the micro. When expanding one’s view to the macro it becomes obvious the exchange you described is not complete until the customer’s mortgage is retired. That cannot happen in NET without public money to replace the debt created and balanced by the mortgage. The bank simply trades its more accepted IOU for the IOU of the borrower. It never loans existing funds but creates them as a credit in the borrower’s account.

Reserves created when banks lend only facilitate interbank transfers and represent obligations of the banks that must be repaid with the high power money only Congress creates. As the bank earns interest it uses part of that to pay down those reserve obligations, always balancing obligations against the principal of the loan. Should the bank’s reserve balance fall short of its balance of extended credit it must obtain/borrow more from others who have excess.

The rate it pays for this borrowing is the “overnight” rate set by the Fed operations. This then becomes the “base rate” from which banks compute their profit and the rate the borrower will pay for credit. Both the reserves borrowed and the interest paid by banks in this function must come from its owned public money reserves at the Fed, usually represented in Treasury bonds. Moving public money from bonds to reserves is a cost/loss to banks, so they avoid doing so whenever possible.

This is only confusing to people who assume “money” is always positive, but ask any banker about his/her reserve balance and you’ll get a different perspective. This system greatly advantages large banks that can maintain a better balance of reserves via more internal transfers that net to zero in the aggregate. This is why smaller banks offer incentives for new checking accounts that mean fewer external transfers from their area.