''I build a dynamic discrete choice model to explore the effect of heterogeneities inliquidity constraints on R&D investment across value-maximising, innovative firms.Firms are offered special tax incentives for R&D, giving loss-makers a choice betweentaking cash from the government immediately or carrying the loss forward to set offagainst future profits. Claiming the cash immediately as incurred in the loss periodcomes at a cost; the rate at which the government pays cash to loss-making firms issignificantly lower than a future deduction. I combine this structural set up with anatural experiment from the UK, which changed the deduction and cash credit ratesfor different groups of firms at different times. The exogenous variation across firmsand over time in deduction rates and the availability of the cash option identifies theeffects of policy changes on cash flow, and consequently, outcomes of interest such asR&D investment and output''.