Essays on consumer demand for Internet access. Evidence from the Internet demand experiment.

摘要

Neoclassical economics assumes that, when confronted with unfamiliar choices that are repeated, individuals learn to make optimal decisions. Behavioral economists, drawing on findings in economics experiments and psychology, have questioned this assumption.; I use a unique dataset from the Internet Demand Experiment (INDEX) to examine whether individuals in repeated choice situations make choices that are, or become more, optimal or become more optimal, and also whether any learning that occurs appears to transfer to new but similar choice situations.; I find that, learning from experience appears to affect the optimality of future choices in the same experiment. These effects appear to carry over to their decisions in the next experiment which had a very different pricing structure. In both experiments, however, there was a group of initially high-performing subjects for whom experience did not appear to result in more optimal choices over time.; I also use INDEX data to examine another aspect of consumer demand for Internet access: whether documented preferences for flat-rate pricing in telephone service and health club memberships extend to purchases of Internet access.; I find evidence that is consistent with the existence of preferences for flat-rate pricing for Internet access. Like a number of telephone demand studies, I find that after controlling for many explanatory variables, INDEX subjects remain significantly more likely to choose an Internet access plan with a flat-rate component.; I also examine ex post non-expenditure-minimizing choices---used in a number of studies as an indicator of preferences for flat-rate pricing. With demand uncertainty, however, there are alternative explanations for such choices, including mistakes in the prediction of usage, random demand shocks, inertia, and switching costs.; Using INDEX data, I find the evidence not consistent with subjects mispredicting their own usage nor with the explanation of random demand shocks. I find the observed behavior to be consistent with preferences for flat-rate pricing for Internet access, in particular with the purchase of pricing plans with flat-rate components as insurance. In addition, I find that the design of the INDEX experiment facilitates learning from experience, resulting in significantly lower percentages of ex post non-expenditure-minimizing choices compared with other studies.