The Impact of Global Warming on Agriculture: Reply
回应Cline对MNS论文的批评，辩护李嘉图方法在评估全球变暖对农业影响中的有效性，并指出其估计的损害小于传统生产函数方法。
In our paper with Daigee Shaw (Mendelsohn et al., 1994), henceforth MNS, we developed a new approach to measuring the impact of global warming on agriculture. We call this approach because it relies upon standard rent theory as a way of identifying the impact of changes on net economic welfare. We compared the Ricardian approach with the traditional production-function approach, which uses agricultural production functions but has great difficulty identifying all the other adjustments that farmers make in response to changing external conditions. The Ricardian approach is particularly well-suited to a tremendously heterogeneous sector because it can rely upon reduced-form estimation and does not require the impossible task of constructing structural models of hundreds of crops in thousands of locations. Estimating the model using cross-sectional data on climate, farm-land prices, and other economic and geophysical data for almost 3,000 counties in the United States, the Ricardian approach shows a significantly lower estimated impact of global warming than the traditional productionfunction approach. Indeed, our preferred statistical approach showed modest benefits of climate change. In his comment on MNS in this issue of the Review, William R. Cline (1996) raises three concerns. First, he notes that the analysis assumes that output prices remain constant. Second, he asserts that the analysis assumes that the supply of water for irrigation is perfectly elastic. Third, he calls for testing the results of the model on alternative climate scenarios from general circulation models (GCM's). His first and third points are useful additions, while the second is incorrect. Our Ricardian approach assumes that the existing pattern of agricultural land rents and land prices reflects the long-run equilibrium economic effect of climate and other geophysical and economic variables. Standard economic reasoning shows that, by calculating the estimated effects of perturbing the climatic variables, we can project the impact of climate change on economic welfare. We do so in a partial-equilibrium approach which assumes that output prices are invariant to the climate, and Cline is correct to point out that this might lead to biased estimates because the partialequilibrium estimates tend to underestimate damages and overestimate benefits. In fact, the bias is small given standard parameters for agricultural demand and supply functions. We illustrate this point using linear supply and demand functions for agricultural crops. Let demand be given by Qd = ao - a1P while supply is Q. = Po + PI P, where Q and P are output and price and ai and [,i are parameters. In equilibrium at the old climate, market outcomes are Qo and PO. Now suppose that global warming contracts supply so that