Some limits to bank printing 'out of thin air'

In this post I want to highlight an interesting discussion I had over at MNE with other commenters about banks and their ability to print money for their acquisitions. An anonymous commented said:

Bank credit money is also created when banks make asset purchases or payments on their own account, not only when they make loans." to which Matt Franko added "imo they must be able to buy the property and construction of their branches by crediting bank accounts of the landowners and contractors.

I was at first resistant to Matt's assertion that banks can print every time they want to make an acquisition. But thinking it through, that's almost about right, but I wanted to add a clarification, based on my own understanding, that not all payments banks make, though they’re paid as credits to accounts, are new money. That credit will result in a debit, and sometimes the debit is to their equity.  For example, if they buy their office supplies, pay their workers, or settle their utility bills, it does not add new money to the economy. If the bank uses its earnings (part of its equity) which is money received from elsewhere, it is not money created at the point of credit. Anonymous commenter says: 

Whenever they spend or lend they do so with their own privately created credit. They only use reserves to settle with each other or with the central bank and treasury (or when they borrow/lend reserves from/to each other,)or cash when people withdraw it. Everything else is done with the 'inside money' they create themselves. ..

When a bank purchases something or pays someone they credit an account, as they do when they make a loan. The credit is typed into existence through a computer keyboard. This adds to the bank's overall liabilities. Banks have to maintain a *ratio* of capital and or reserves. That ratio has to "come from somewhere" in the case of said purchases/payments, but it's only a fraction of the overall credit/deposit money credited by the bank. If the created deposit is then transferred to another bank then of course settlement takes place with reserves, as always.

Not much I really disagree with. I just would add that loans and branch locations are two different animals. When a bank buys a branch property, it has to be funded by a corresponding liability - its own equity, a loan from somewhere else, or via deposits. But when a bank buys a loan, it expects a positive income from the spread between the loan and the cost of the liability that results from it. Then when a bank buys property for a branch, or a computer printer, there's usually no income assumption, as they’re an expense outlay for the bank.  While it is true that purchases or payments are created as an accounting entry by the bank, if they result in net equity outflow, that outflow cannot be compensated by the bank by printing reserves into existence. One has to distinguish whether that payment is for an expense or for funding a loan. Anything that is considered an expense outlay is not new money created by the bank. And any net cost that results from that purchase is not funded with new money, as is the case with equity capital that has to be raised to maintain the ratio. That’s why, as Tom Hickey said, "the bulk of horizontal money creation comes from credit extension, not asset purchases or payments on their own account". Tom also mentions in the discussion. 

When a bank creates bank money to purchase assets other than loans or fund expenses, it is in effect borrowing at the price of reserves, since it is creating deposits that have to clear (after netting). Even if a bak uses its own excess reserves, that's interest it is losing on lending in the interbank market. Banks can create money but not "for nothing." But it is true that banks borrow at a much lower rate than non-banks in the sense that they don't have the spread to deal with.

This is also why, I believe, a bank that has no earnings or no equity cannot continue spending just because it can create money out of thin air by crediting an account. If the spending is going to be funded by liabilities, then the bank incurs greater liquidity and insolvency risk.  As Dan Kervick also adds: 

But the money they create to buy stuff for themselves is just as much a liability as money they create in the process of making a loan. They create an account for the seller of the goods and credit that account. If the seller of the goods then writes a check on that account to someone whose account is at another bank, the first bank will have to make a payment to that other bank that is settled and cleared through the banks' reserve accounts at the central bank. And if the seller comes into the bank and decides to withdraw the total amount in the account, the bank will have to hand him a bunch of vault cash. Clonal also adds: 

I believe the fact that revenue is recognized immediately on making the loan is where all the problems are coming from. More money than the amount of the loan is created at the time of the inception of the loan. The excess money is then capitalized after accounting for things like provisions for loan losses etc.

To which I acknowledge, I could be referring to an outmoded banking model. As far as I know, if a bank keeps the loan in its books, revenue is recognized as and when it receives interest income, not when the loan is made. Of course, if a bank securitizes everything or most of the loans it makes, revenue will be recognized when the loan is sold, even if it's at the point the loan is made.

P.S. I add in comment discussion that Basel rules further limit this printing.