Sunday, 24 February 2013

The New Model Bank

In all my years of analysing systems there is one line you hear over and over again.

"That's how we've always done it"

The implication is always that there no possibility it could be done any other way. The assumptions are unchallengeable, the procedures unalterable and the reasoning unquestionable.

Then I come along and I question, challenge and where necessary alter things.

So let's do that with a bank - starting at the very beginning.

Banking is fundamentally a very simple spread operation. You charge more for your assets than you pay on your liabilities. Hence the old 3-6-3 line 'pay 3%, charge 6% and be on the golf course by 3pm'.

So we'll start with those principles for the initial model.

I'm going to build this model within the Sterling Monetary Framework - since that is the one that I'm familiar with and interested in. I want this model to be one that could run as an actual bank in the UK.

The Funding for Lending data page shows the lending size of regulated institutions in the UK. The newly formed Metro Bank is the smallest, holding £78m of lending. So operating at that sort of size it should therefore be possible to obtain regulatory authorisation, a Reserve Account and access to the Discount Window Facility at the Bank of England.

There is no Cash Ratio Deposit requirements for institutions under £500m of liabilities so the capital adequacy ratio is going to be, if I've read the Basel runes correctly, about 8-9%. So £100m of lending is going to require a minimum carrying equity of about £8m. (As a comparison Metro Bank has raised about £120m of equity to set up its operations).

Now as we know on the margins loans create deposits, but as the excellent JKH post shows within the current framework you still have to manage your liabilities mostly within the private sector.

However there is one point that post overlooks - which is that the liability always exists while the loan does. All that happens is that the person that the bank is liable to changes. An example will hopefully demonstrate what I mean.

So initially after you create the loan the balance sheet looks like:

Lending Bank Transactional Bank
Assets Liabilities Assets Liabilities
CB Reserves £0 Deposit £100

Loan £100
CB Reserves £0 Deposit £0

then the Lending Bank loan customer pays somebody at the Transactional Bank and you get:

Lending Bank Transactional Bank
Assets Liabilities Assets Liabilities

Deposit £0

Loan £100 CB Reserves £100 CB Reserves £100 Deposit £100

Then given that Transactional Bank gets just the bank rate of interest from the CB Reserves it will try and lend them out, and, importantly, the only taker will be Lending Bank (let's assume for simplicity this is the only loan in the clearing period we're analysing). So you then get.

Lending Bank Transactional Bank
Assets Liabilities Assets Liabilities
CB Reserves £0 Deposit £0 Inter Bank Reserves £100
Loan £100 Inter Bank Deposit £100 CB Reserves £0 Deposit £100

So the net result of the decision by the initial borrower to pay somebody at Transactional Bank is to force Transactional Bank to become the replacement depositor in Lending Bank. (That's 'force' in the sense of maximising income from the excess reserves).

That gives us the first cut of a new model for a Lending Bank - outsource the expense of managing retail deposits to Transactional Bank and just use the overnight market permanently to cover off the mortgage loans made. The interest rate hedge is that the loans made are simply charged at a percentage above the overnight rate.

Currently the best tracker on offer is priced at 1.88% above. Running that through the model with an average expected loss ratio (which is generally less than 0.5% of the outstanding book on these types of loans) seems to show a reasonable return on equity. The costs of running this loan model will be very low - particularly if you push the loan origination out via the brokers so that origination costs scale with turnover. You then get paid for your underwriting talent.

By creating the loan in an entity with a reserve account you pretty much force the system to loan you the matching deposit for as long as the loan exists. Therefore it doesn't seem to make any sense to pay a time premium for longer term money.

So would this work? Thoughts welcome.