pumpkinstrokemarrow

Have been continuing to ponder about banks and decided that yesterday I fell into error with my noddy model. So what follows is issued by way of an errata - but without going to the length of marking up the changes. Not as if I was still at work after all. You can always play spot the difference.

So let us suppose we have a bank the business of which is to take deposits from one bunch of people and to make loans to another bunch, while remembering that in the olden days one of the main functions of a bank was to provide a safe haven and depositors would pay for moorings in that haven. Nowadays depositors generally expect to get paid. So let us suppose that the total deposited is D and that the bank has to pay depositors DI each year by way of interest. Being a model for today, we assume that DI is positive. We also assume that some at least of the D can be withdrawn on demand. Contrariwise, leaving aside Christian and Muslim subterfuges to avoid being accused of usury, banks have always charged borrowers. Let us suppose that the total lent is L and that the borrowers have to pay the bank LI each year by way of interest. Let further suppose that the bank has cash assets of C and owners O owning shares to the value of the market capitalisation M, M being some function of of some combination of what the market thinks that the bank will pay in dividends, DIVI, and what the market thinks the bank should be worth. Asset strippers being one of the varieties of sharks that move into the market when the market gets it wrong and M gets too low. We forget about  other assets - buildings, systems, goodwill etc - and operating costs. Then, for a solvent bank we should have L+C>=D, so that the bank can pay off its creditors, should need arise. If C reaches zero the bank is said to be bankrupt. Such an eventuality might be staved off either by calling in a chunk of L, perhaps bankrupting lots of small businesses on the way, or by a rights issue, supposing the market to have any appetite for same.

We then poke the thing around in various ways.

If L and D do not change from one period to the next and we use '*' to denote a value for the next period, we have C*=C+LI-DI-DIVI. Which suggests the possibility that an unscrupulous bank might pay far too much dividend, eventually resulting in the shares in the bank being worthless, but with shareholders having done well enough in the meantime not to mind.

If we allow L and D to change, we get C*=C+L-L*-D+D*+LI-DI-DIVI. So making lots of new loans or lots of depositors taking their loans back will hit C.

If all that the bank was doing was providing safe haven then L & LI would be zero, C >= D and DI < 0. But apart from providing safe haven, most banks also act as a sort of unit trust. Depositors get a share of the returns from the pool of lenders, with it being unlikely that many of the lenders will go bad at the same time, with the bank taking most of the risk and with C being the buffer against problems with L. Because O are the ones taking the risk it is fair that while DI is now positive, LI is bigger than DI, with some at least of the difference going to O as DIVI.

If DI gets bigger than LI and stays that way for any length of time, the bank goes bankrupt. This might happen if, for example, a whole lot of L goes pear shaped and has to be written down. Or if L is small because the bank can't find enough proper people to lend money to and resists the temptation to lend money to improper  people. Resistance which was broken down in the first instance, I seem to recall, by President Clinton saying that he wanted houses to be built for the people.

If the depositors make a run on the bank and D falls rapidly, the bank may exhaust C before they can liquidate enough L to pay out the depositors. Again, the bank is bankrupt.

We get the same effect if a chunk of D matures and the bank is unable to roll it over. This is, in effect, what happened to Northern Rock.

Haldane makes the point that once upon at time O had unlimited liability to meet the D flavoured liabilities, a fact which made O pay serious attention to what the bank was doing, rather than just letting the hired help get on with it. Banks tended to be more conservative than they are now with L a much smaller multiple of C. Maybe as low as 1 or 2 rather than the 20 that we peaked at just before the crash. Which last means that L and D are pretty much the same large number. The up side is that, other things being equal, the profits go up in proportion: double the loan book, double the profits, all on the same initial capital. And the bankers themselves were not the only stakeholders who wanted a slice.

He also makes the point that not only is L a large multiple of C, it is also a large multiple of the GDP of the country in which the bank is registered. So a relatively small percentage loss on L might mean that digging the bank out of bankruptcy might easily engulf the host. Making banks smaller does not necessarily remove this risk: if all the small banks get into the same fix at the same time, fixing them all costs just as much as fixing one big bank.

All of which leads back to something that I have been suspecting for some time. Small countries should not be in the business of hosting big banks which run big risks if hosting is to include underwriting their losses. So, at some point the centre of gravity is going to move away from London to some place where the real money is.

Small countries should only host big banks which do not run big risks. Whose L is good and safe. Banks which exist more to provide serviced safe havens for the man & company in the street than to provide loans to grand & dodgy projects in far flung jungles. Or to grand & grossly overvalued housing estates nearer home. Perhaps in Ireland.

Against all this one might argue that C is dead money. Gold sovs. under the mattress. One wants to turn as much of C into L as possible so that it supports LI. That is where all the wealth comes from.

Perhaps now it is the time to leave all this to the experts. But at least I have dusted off some grist to the saloon bar mill.