In warm climates, trees could provide natural air-conditioning by shading homes. Yet, there is little rigorous empirical evidence on the shade benefits of green infrastructure. This paper uses data on tree removal permits from a tree protection policy in Gainesville, Florida, between 2000 and 2016, and applies a difference-in-difference estimation model to examine whether tree shade reduces electricity demand. I find that a typical tree removal leads to 4-6% more electricity consumption annually, and 8-10% more during summer months of peak air-conditioning demand. Effects increase with the amount of shade loss: from zero effect (no shade) to 8-12% (large shade loss) and 10-20% (very large shade loss). I then use these estimates to calculate the private benefits of shade trees ($450 to $1,900) and find that they are similar to other energy efficiency investments (retrofits, building codes) assessed using comparable evaluation methods. I also calculate the social benefits (shade benefits to neighbors, avoided emissions, and avoided generation costs) of the tree protection policy and find that they are equal to the city expenditures on this policy. Overall, this study presents a road map for utilities and policy makers to assess returns from investments in green infrastructure.