Unless I've dropped a clanger, which is more than possible. Anyway, what I mean is this.
The initial effect of negative rates is simply to impose costs on commercial banks. For eg., commercial bank X, which has $1 billion in reserves and which previously paid / received no interest on those reserves, and which then has to pay 1%, would face a bill of 1% times $1 billion, i.e., $10 million pa.
Commercial bank X now has a bigger incentive to lend more, because for every $100 loaned, about $90 ends up being deposited with OTHER banks. That means that bank X owes the latter banks $90 in reserves, which, of course, bank X is happy to dispose of.
However, in order to persuade its customers to borrow more, bank X has to reduce the interest charged. Assuming it can discriminate between new borrowers and existing ones, the latter "additional loans" strategy will work for bank X. Of course, other banks will be attempting the same ploy. Nevertheless, the net result ought to be additional loans for the economy as a whole, as far as I can see. Reserves become a sort of hot potato which every bank tries to pass on to other banks.
However, banks' existing customer / borrowers are not going to be too pleased when they find out that new borrowers who are identical in every respect to themselves are being offered loans at a lower rate. But if the small print in loan agreements bars existing borrowers from renewing their loans, there's not much they can do about it. So in that case, negative interest rates on reserves will increase lending.
On the other hand if, or to the extent that, existing borrowers can renew their loans at the new lower rate, negative interest on reserves won't work. The reason is that (to repeat) the initial effect is to impose a cost on banks, and assuming banks were earning a standard return on capital prior to the negative rates, and assuming they have to continue to earn that standard return, the banks just have to pass on the cost of negative interest on reserves to customers. That is, the interest charged to borrowers will rise. Ergo, the total amount loaned will fall.
Or have I missed something?