Having worked in a GSIB Treasury I can answer this: In short banks do not co-operate, they are completely independent of each other. To manage liquidity they project all their expected outflows and inflows over a variety of time horizons under both normal and stressed conditions (capturing everything - cc purchases, drawdowns on overdrafts or other credit facilities, deposit movements and transfers, mortgages and other loans, refinancing of funding, debt issuance, staff and building costs etc etc etc, there's a huge amount of modelling behind this). Under US or UK regs Banks have to hold large liquidity buffers, in the UK this is enough to finance 2 months stressed outgoings assuming no incoming. It's pretty punitive, and runs into the 100s of billions for big banks.
Thank you :) Sounds like a hell of an optimization problem, since there are not only many variables to model, but also many ((risk) allocation) decisions to make!
When a bank considers giving a loan, can it model where the money would/might be eventually transferred to - whether it would stay on the bank's books and thus require less liquidity to cover net outflows?
I have the impression that central banks try to model national economies via "flows between large groups of actors", do commercial banks analyze/simulate this too on a (smaller) scale, to find out how not only money flows between them and other banks, but within itself too? Since accurate models would result in more profitability, is it safe to assume that banks have the best economic "world model"?
When I buy groceries, I have no idea where the money ends up even in the next "step", maybe banks attempt full "network simulations"?
