An experimental approach to extrapolative expectations.

Summary Several recent studies have shown that investors in asset markets form their expectations by extrapolating the most recent evolution of an asset's price. The aim of the experiment is to confront individuals assumed to be "rational" with "extrapolators" in order to characterize forecasting behavior and to see if the presence of "extrapolators" affects the behavior of "rational" individuals. To do this, we simulated the price of an asset over fifteen periods that fluctuates around its fundamental value. The experiment consists of five treatments. In order to differentiate between the two types of participants, we vary the information available to them to make their price prediction: the so-called "extrapolators" only have access to a window of five previous periods, while the so-called "rational" participants have access to the entire history up to the forecast period. Therefore, we posit two modes of remuneration: one based on the accuracy of the forecasts and the other based on the deviation from the average forecast. Thus, one group of each type of participant will have a different remuneration mode . corresponding to four treatments. Finally, the last group will be composed, in equal proportions, of "extrapolators" and "rationals" and will have an objective of deviation from the average forecast. This will allow us to see if the "rationals" correct their forecast when they are in the presence of "extrapolators". [source: abstract].