R&D is not a total black box. Management sets a budget, "Scientists you are allocated $XXX this year for reagents + equipment." Sure there are plenty of unexpected gotchas (equipment X broke, unexpectedly announced equipment Y is a must-buy, etc), but more or less there should be some ability to forecast expenses. If the company is in the red, cut that number down.

All conjecture, but as a former bench scientist (spending money, not making budgets), it felt like our burn rate was reasonably constant.