Do People Bundle Sequences of Choices? An Experimental Investigation

Economists and psychologists have sought to model and explain both impulsive behavior and the costly but often successful mechanisms by which people control it. Ainslie [1975][1992][2001] suggests that self-control is often achieved on account of a phenomenon he calls "choice bundling." This refers to re-framing of series of discrete choices as single choices over whole series. Whereas other core elements of Ainslie's account of self-regulation, such as hyperbolic discounting and intrapersonal bargaining among temporally distinguished selves have been subject to extensive modeling by economists, choice bundling has been absent from the economic literature because it has never been empirically isolated in a controlled setting that meets the methodological requirements of the discipline. We report a laboratory experiment that fills this gap. Subjects made choices between smaller, sooner and larger, later real monetary rewards under experimental treatments that allowed us to discriminate between choice bundling, reliance on pre-commitment, and possible magnitude effects on intertemporal discounting. Risk preference measures were used to obtain accurate discounting estimates, based on estimation of mixture models that incorporate exponential, hyperbolic and quasi-hyperbolic discounting functions. We use structural econometric procedures which are well established in the literature on binary choice and find strong support for the hypothesis that subjects bundled choices when conditions allowed them to do so, and consequently exhibited different discounting behavior in these conditions.