Masterclass

Climate and Economics (Tropical Forests)

Speaker

Jose A. Scheinkman (Columbia University)

Date & Time

From: 29 April 2024, 09:00
Until: 30 April 2024, 17:00

Type

Masterclass

Estimates of total carbon stored in tropical forests exceed 200 billion tons. If released by deforestation, this stock would produce CO2 emissions that would exceed the estimates of total US emissions since pre-industrial times. On the other hand, major tropical forests systems have been substantially deforested. For instance, an area the size of the US State of Texas has been deforested in the Brazilian Amazon mostly for use in agriculture. This ecological disaster has also created an enormous opportunity for carbon capture through reforestation.


The goal of these lectures is to acquaint researchers with data and methods that are used to evaluate the potential of these natural systems to capture carbon and the associated costs, as well as other important externalities generated by reforestation. We will use extensive physical and economic data from the Brazilian Amazon, the largest tropical forest in the world. These data show marked cross-sectional variability in the potential for carbon absorption and in productivity of the current alternative uses. We will assess the consequences of imposing alternative social costs of carbon emissions, including subsidies from abroad, on the optimal patial/dynamic allocation of land-use in the Brazilian Amazon. While we consider a single central planner, the heterogeneity in productivities and natural state-constraints preclude the use of standard recursive dynamic methods. To treat risk, we will introduce Modified Predictive Control, a method
originally developed in control theory and engineering to study multi-plant production in real time. To deal with the unavoidable uncertainty on the value of crucial parameters, we will consider model-determined robustness adjustments to the subjective probabilities of these parameters. We will once again discuss the necessary numerical methods; in this case Hamiltonian Monte Carlo, a method that has substantial advantage over the familiar Metropolis-Hastings approach in treating high-dimensional problems.