This came up in a couple places but a comment by Unlearningecon offers an opportunity to focus in on it. He writes:
"Reserves are just another cost decision, along with staff, other inputs, etc. The availability of them does not enter into a bank's lending decisions, which is done purely on a bookkeeping basis. At the end of some time period, a bank will settle its reserve decisions, as with any other cost. The price of reserves can be altered to make banks lend less/increase the interest rates, but this is only in a similar way that some other 'supply shock' might."
I think some of this is good and some of this isn't. If by "endogenous money" you just mean "loans create deposits", that's certainly correct [supply and demand simultaneously determine equilibrium so I'd add "deposits create loans" too, but no need to get into that because I don't challenge the "loans create deposits" point].
But a lot of people mean a lot more than that. A lot of people mean that the level of reserves is endogenous to the rest of the banking system - that it's just a residual that pops out of the loan and deposit creation decisions, given an interest rate decision by the central bank.
This point - that the banking system generates a particular level of reserves - is what I challenge. The central bank does exogenously create and destroy reserves every day. This is the exogenous policy lever.
Another point of clarification: if you think "endogenous money" means that the central bank targets interst rates, that's fine but I don't think anyone disagrees with that point. My point is, they don't "set" an interest rate and then reserves become whatever level is consistent with that interst rate and the decisions of the banking system. No - the central bank has an interest rate target they are interested in, and then they "set" reserves by buying and selling bonds (or through other institutions in other countries) so that the interest rate they want falls out the other end.
It doesn't have to work this way, of course. This is all based on institutional set ups. You could have the causality running from interest rates, through the banking system, to reserves if you wanted. That would be a discount window approach. An interest rate is set and the Fed takes all comers. Base money is endogenous. We used to do that. But now we primarily work through open market operations and causality runs the other way. Reserves get shuffled and a desired interest rate pops out the other end.
When most Fed policy starts taking place at the discount window again rather than the open market desk, we can talk about the strong version of endogenous money again (i.e. - not the weaker "loans create deposits" version that I think is fairly non-controversial when we get down to it).