An economic perspective on risk and vulnerability
This dissertation is comprised of four essays investigating the notions of risk and vulnerability, with a particular emphasis on climate risk and climate vulnerability. Because of differences in data, methodologies, and the independence of research objectives across the various chapters, each chapter stands alone as an independent essay. ^ The first essay focuses on rural-urban migration and land rentals as household-level responses to risk. Using a longitudinal data set comprised of households in nine provinces in China from 1991 through 2006, we test the extent to which households have utilized migration and land rental markets in response to various sources of risk, including idiosyncratic and covariate income risk, labor supply risks, expenditure risks, and weather risks. We find that these risks affect household behavior, but not in a manner consistent with expectations. The results suggest that households respond to changes in the values of the marginal productivity of land and labor in a manner consistent with profit maximization, though not necessarily consistent with risk aversion. ^ The second essay uses spatial data and estimates a cereal yield response function conditional upon climatological and topographical features. This exercise uses a recently-developed estimator for spatial process models when sample selection is of concern. We control for localized spatial correlation in unobserved disturbances affecting both the selection to plant cereals as well as in the resulting conditional yield response. Based on the estimated sensitivities of yields to temperature and precipitation, we are able to simulate the yield impacts of climatic changes and test the extent to which increased irrigation infrastructure can mitigate the climate change-induced yield declines. We furthermore demonstrate that failing to control for spatial correlation in sample selection models can lead to biased estimation of the marginal effects. ^ The third essay examines the extent to which economic development reduces both a country's disaster risk and its social vulnerability to climate-related disasters. Using data on climate-related disasters and their resulting societal outcomes drawn from an international disaster event database (EM-DAT: The OFDA/CRED International Disaster Database - www.emdat.be - Université Catholique de Louvain - Brussels - Belgium) regressions indicate that the ability of economic development to reduce disaster risk depends on a country's income level; additional income becomes less effective in reducing disaster risk as countries become wealthier. Conditional on a disaster occurring, higher incomes generally reduce a country's social vulnerability to such disasters. The results additionally suggest that underlying social and political structures may have a significant effect on the human costs of disasters. ^ The final essay tests an underlying hypothesis implicit in Kyoto-type emissions-reductions policies, namely that low-income countries should be able to continue increasing their energy consumption as a means of increasing income and resilience to impending climate change, thereby reducing their vulnerability. Cross-country data on energy consumption, per capita GDP and a social vulnerability index are used to measure changes in vulnerability associated with changes in per capita GDP and per capita energy consumption. Through its nonlinear effect on per capita income, energy consumption has the effect of reducing a country's overall vulnerability, though these reductions in vulnerability are greater at moderate incomes than at low or high incomes. An implication is that country-specific climate change policies which emphasize carbon reductions through reductions in energy use, especially in developing regions, are unlikely to significantly affect vulnerability to the risks arising from climate change, especially at very low incomes.^
Gerald E. Shively, Purdue University.
Climate Change|Economics, General|Economics, Agricultural
Off-Campus Purdue Users:
To access this dissertation, please log in to our