A regime switching model based on real options shows that optimal land use switching boundaries are contingent upon the stochastic behavior (trend and volatility) of timberland and cropland returns, conversion costs, and return correlation. Using this modeling framework applied to timberland and cropland in the US South, tree planting subsidies are shown to increase forest area over baseline projections. However, timber supply feedbacks and partial eligibility for subsidies are shown to generate substantial benefit leakage. Given the context of market feedbacks it's not clear that tree planting subsidies alone are an effective policy instrument for expanding forest area.