The effects of agent heterogeneity in environmental preferences on social welfare estimates derived by the contingent valuation method

Human society is facing a variety of broad and unique environmental challenges as we enter the twenty-first century. The economic consequences of environmental changes, such as biodiversity loss, depletion of the ozone layer and global climate change, could have very serious impacts on human societies and economies around the globe. Because environmental goods and services are not traded in established markets, economists have developed the contingent valuation method (CVM) as a tool to place economic values on public environmental goods. These values can then be used to compare the relative economic advantages and costs of development projects using cost-benefit analyses or for environmental litigation purposes. One technical issue of particular interest is related to the effects of agent heterogeneity on the accuracy of the social welfare benefits derived using CVM. This study uses an artificial market approach to examine whether valid estimates of aggregate social welfare are obtained using CVM when artificial agent preferences are heterogeneous with regards to environmental goods and whether there are systematic biases away from true welfare as measured by compensating surplus. I found that estimated values of willingness to pay for three levels of environmental change were significantly less than true willingness to pay for the artificial agents. In addition, an analysis of residual errors showed that the use of the log-linear utility function in the empirical utility difference CVM model provided significantly improved, albeit inaccurate, estimates of true willingness to pay compared to the use of the linear utility function. This study has shown that welfare measures estimated suing CVM are unreliable when there is agent heterogeneity, as is the case in the real world, even under simple conditions when agent preferences are specified in accordance with economic theory and there are no technical complications relating to survey design and delivery.Land and Food Systems, Faculty ofGraduat