摘要: A menu of paired lottery choices is structured so that the crossover point to high-risk can be used infer degree risk aversion. With normal laboratory payoffs several dollars, most subjects are averse and few loving. Scaling up all by factors twenty, fifty, ninety makes little difference when high hypothetical. In contrast, become sharply more actually paid in cash. hybrid "power/expo" utility function with increasing relative decreasing absolute aversion nicely replicates data patterns over this range from dollars hundred dollars.