Steve Keen and Mish make a fallacy of composition error?

I agree with Steve Keen and Mike Shedlock (“Mish” for short) about 90% of the time. But they’ve slipped up on the subject of bank reserves, unless I’ve missed something. I’ll use the word “bank” to refer to commercial banks (as distinct from central banks)

Mish challenges the conventional wisdom on interest on reserves (IOR), which is that IOR is an incentive for banks not to lend. The conventional wisdom is that if a bank gets say 5% on its reserves, and given that lending $X means the bank loses about $X of reserves, it won’t lend unless it gets at least 5% after expenses from the potential borrower.

Mish challenges that by pointing out that the decision by an individual bank (or indeed the bank industry as a whole) to lend more has no effect on the industry’s stock of reserves. He concludes from that that interest on reserves does not provide banks with an incentive not to lend.

Well it’s true that the decision by a bank or the bank industry as a whole to lend more has no effect on the industry’s stock of reserves. However, the flaw in the latter “Mish” argument is that “incentives” as seen by an INDIVIDUAL bank are not the same as “incentives” as seen by the commercial bank system as a whole. That is, if an individual bank (to repeat and over-simplify a bit) gets say 5% on its reserves, it has no incentive to lend unless it gets MORE THAN 5% from the potential borrower after expenses.

The fact that IF IT DID LEND, there would be no effect on the TOTAL AMOUNT the bank industry gets by way of interest on reserves is irrelevant because banks just don’t collude when it comes to the decision by an INDIVIDUAL bank to lend – they couldn’t collude if they tried. To illustrate…

Suppose that bank A is contemplating a loan of $Y to a customer, and suppose that in the event of the loan taking place, relevant monies will be deposited at bank B. It is plain impossible for both banks to get together and work out the effect on their combined incomes as a result of making that loan: reason is that each bank will be making thousands of loans per day. Moreover, in the latter example, relevant monies will not necessarily be deposited at just one bank (i.e. bank B): chances are they’ll be deposited at several banks. And worse still: relevant monies will not stay at recipient banks for long. Those monies will be re-spent and deposited at yet another set of banks.

Thus the idea that banks can somehow collude with a view to working out the effect on their total income as a result of one bank making a loan is wholly, completely and totally unrealistic.