Hmm, by implementing stuff in-house we have eradicated an expensive external annual subscription product, reduced the cost of using the tool (halved the fees to an external agency that the product interacted with) and improved automation (over the poor annual subscription product) enough to cut a number of jobs. There isn't a value I can put on things really (I don't know the wages of the eradicated positions, etc, and costs are also lower because business is lower because the economy is crap).
Eclipse Metrics plugin tells me that this project has some 20k lines of code (excluding comments I assume). I'll assume it's saving £200k a year as a reasonable minimum. That makes some £10 per line of code saving, per year. Note that this year that code will be used on an additional new product that will bring in money, and hence should be a factor for income as well as a theoretical cost saving over using the old system for the new product.
It's a stupid metric because it doesn't count the hours taken to implement, the wages of the programmer and the business analysts, the meeting hours, etc. Should I factor in the lines of code in JSPs, ANT build scripts, etc? Lines of documentation and commenting?