I impose a spell failure chance, based on the proportional difference in value between the actual component and the required one. If your component is worth X percent of the required value, you have that same percent chance of having the spell work.
For instance, if the spell requires a 10k gp diamond and you try to use one worth 5k, you have a 50% chance that the spell will work. If you try a 100 gp diamond, you'd have only a 1% chance it would work (a 99% chance of failure). Failure on this roll still expends the spell slot and consumes the component.
A caster taking advantage of this will still pay just as much for components, over time; when using cheap components he'll have to make more attempts, and it'll all even out. I just think it's interesting for flavor purposes.