The experiment was designed to simulate an aggregate expenditure model of the economy. Participants had to spend a minimum autonomous component of consumption. Participants had the choice of saving all other income and savings, or spending more, up to the limit of their total income and accumulated savings. Unemployment with zero income was randomly assigned, and the likelihood of unemployment increased if previous round aggregate consumption decreased. This scenario is intended to test if the likelihood of unemployment, without public assistance for the unemployed, would cause still employed consumers to reduce consumption. Such is a behavior leading an economy into a downturn, consistent with the paradox of thrift. Results found that in later rounds most participants chose to spend less than they had during earlier rounds. Such behavior increased the likelihood of the average participant becoming unemployed, but reduced a participant's personal risk of having no savings if he himself became randomly unemployed. The experiment carried severe penalties for becoming unemployed without savings. Such behavior is consistent with the "buffer stock" hypothesis of consumer behavior. At the beginning of an economic downturn working consumers reduce their spending, causing pro-cyclical effects on business cycles. The competing hypothesis would expect a working consumer's marginal propensity to consume to lead to counter cyclical consumer spending behavior.