2010/12 LEM Working Paper Series

Assessing elicitation task bias in time preference using experiments with artificial subjects

Oksana Tokarchuk, Roberto Gabriele
elicitation task bias, time preference, choice task, multiple price list, artificial agent simulations

  JEL Classifications
D03, D91, C63

Experimental results in research on time preference are often controversial. We propose a systematic investigation of choice task in multiple price list format (MPL) that is frequently implemented in experiments on time preference, through a computer simulation analysis. We conduct experiments with artificial subjects to demonstrate that elicited discount rates are highly dependent on the structure of elicitation task. We verify that implementation of choice task in MPL with nominal structure results in observation of hyperbolic discounting. Choice task in MPL with interest rates structure leads to elicitation of discount rates compatible with exponential discounting. Moreover, we show that the magnitude and intensity of corresponding pattern in data depends on the internal structure of elicitation task. Comparison between discount rates elicited with artificial and human subjects suggests that behavior of human subjects in experiments with MPL can be described by two simple rules: positive discounting and anchoring.

download pdf