How banks price their loans and products

When I googled this title, guess how many results I got? Zero.zilch.nada. That in itself spurred me on to write something on which there is little organized information. I am not(yet!) an expert on pricing but from what I have seen of the structuring, trading and risk sides of banks, I think I can venture some informed views on this manner. More knowledgeable readers are welcome to comment on this primer in FAQ form.

1. Every business needs to price its products. What is so exotic about banks? Feedback is given to a real world business via market share, inventory etc. But inventory is non existent in banks(OK maybe cash is that if you stretch the definition). And thanks to poor business intelligence, even the market share data is hard to come by.


2. How is risk priced in? We do not have(yet) that one universal calculator which will spit out an integrated figure for all risks. So often the systems are fragmented. Credit risk calculations are often bifurcated into counterparty(interbank/other FIs) and client risk(normal corporate transactions). And there are many ways to splice other transactions. So finally, an excel/other manual systems are needed to make sense of this mess of figures


3. But why does not some one automate it? It is a control issue. Given the high sensitivity of the pricing 'secret sauce', it was historically not brought on the platform except for record keeping purposes! Thankfully, the success of platform driven retail banking has persuaded banks to adopt the IT company mantra of automation. So in the years to come, we should see better practices in this front. 


4. We learn this in marketing 101. Do those principles apply here? Yes and no. Yes because the exact pricing decision will depend on how well the pricer knows the client, which is again related to marketing. But no because the performance management system of banks ensure that the cost of funds charged to the trader, is often determined post facto, and so there need to be relatively accurate and robust models available for trade negotiation and evaluation. And while geeks/quants can design it, the end users need to know the ins and outs of the model, including its limitations, so that for extreme cases they can use modifications. For example, while executing