There are many such curves. Many of the testing books have this graphic somewhere in the book.
The numbers I seem to recall is that out of every development and maintenance dollar spent, 3 cents are spent on correcting a bug at the specification phase while 67 cents are spent correcting the same bug in the production phase.
A common quote is the 1:10:100 rule (dev/test/production). However, this might not be so important. Just remember that the actual cost of fixing a bug can be the tip of the iceburg. The impact of letting the bug through to production can become sky high. Lost customers, rectification of the damage (data etc), rectifying the outcome (perhaps it resulted in misbilling thousands/millions!), rollout of the fix, etc
The typical cost to fix curves are a good starting point. What needs to be factored in is when is the defect discovered and when is it resolved. The difference between requirements and release is a large cost (according to the curve, the gradient is large), but if it is found in design and fixed in development then the cost is minimal (the gradient of the curve is minimal).
What also has to be figured into the cost to fix is the amount of rework (which is the amount of time and effort to fix the issue and put it back through the cycle), this is the one a lot of companies miss and pay dearly for.