Consumption adaptation, anticipation‐bias, and optimal income taxation

• Date: 01 June 2017
• Author: Ronnie Schöb,Thomas Aronsson
• Published date: 01 June 2017
• DOI: http://doi.org/10.1111/jpet.12227

Received: 26 May2016 Accepted: 4 August 2016

DOI: 10.1111/jpet.12227

ARTICLE

Consumption adaptation, anticipation-bias,

and optimal income taxation

Thomas Aronsson1Ronnie Schöb2

1UmeåUniversity

2FreieUniversität Berlin

ThomasAronsson, Umeå University,

SE-90187Umeå, Sweden

(thomas.aronsson@econ.umu.se).Ronnie

Schöb,Freie Universität Berlin, D–14195 Berlin,

Germany(ronnie.schoeb@fu-berlin.de).

Theauthors would like to thank an anony-

mousreferee and an associate editor, as well

asClemens Hetschko and Magnus Wikström

forhelpful comments and suggestions. Aron-

ssonwould like to thank the Bank of Sweden

TercentenaryFoundation,the Swedish Council

forWorking Life and Social Research, and the

SwedishTaxAgency (all of them through project

numberRS10-1319:1) for generous research

grants,and Freie Universität Berlin for gener-

oushospitality. Schöb would like to thank Umeå

Universityand the Department of Economics in

particular,for great hospitality.

Adaptation is omnipresent but people systematically fail to correctly

anticipate the degree to which theyadapt, leading them to make irra-

tional intertemporal decisions. This paper concerns optimal income

taxation to correct for such anticipation-biases in a framework

where consumers adapt to earlier consumption levels. The analy-

sis is based on a general equilibrium OLG model with endogenous

labor supply and savings where each consumer lives for three peri-

ods. The results show how a paternalistic government may correct

for the effects of anticipation-bias through a combination of time-

variant marginal labor income taxesand savings subsidies/taxes. The

optimal policy mix remains the same, irrespective of whether con-

sumers commit to their original life time plan for work hours and

consumption or reoptimize later on when realizing that they have

already adapted more than expected.

1INTRODUCTION

Adaptation is omnipresent as people adapt to almost all life circumstances. For instance, when income and thus the

consumption possibilities rise, adaptation implies that the increase in utility may only be transitoryand potential utility

gains evaporate over time. Indeed, empirical evidence suggests that adaptation eliminates 60% of the initial positive

effect of an increase in the individual income level on happiness within two years (cf.Clark, Frijters, & Shields, 2008).

These psychological processes of adaptation have not played anyimportant role in economic analysis. Becker and

Murphy (1988) concede that adaptation is important but causes no problem for normative economic theory.Adapta-

tion can easily be incorporated into the rational choice framework as long as people can foresee the mechanisms of

adaptation and addiction. With perfect anticipation, agents will, irrespective of the degree of adaptation, always act in

their own best interest.

However,recent psychological research provides strong evidence that people do not anticipate correctly how they

adapt to changes in life circumstances. Wilson and Gilbert (2003) summarize the evidence and argue that while people

can correctly anticipate the valence of future feelings and specific emotions (such as joy or sadness) they makesystem-

atic errors in predicting the intensity and durability of future feelings. If the person today cannot predict the feeling

of the person tomorrow, the consequences are somewhat similar to the consequences of externalities. The cost one’s

Journal of Public Economic Theory.2017;19:713–731. wileyonlinelibrary.com/journal/jpet c

2016 Wiley Periodicals,Inc. 713

714 ARONSSONAND SCHÖB

decision imposes on others—in this case on one’s own future selves—is not adequately accounted for in one’s decision.

This behavioral failure might be called adaptation internality. Insofar as a person todaydoes not adequately project the

consequences of her decisions for the person she will be tomorrow, it follows that the person today makesa decision

that may harm her future self.We may end up wanting things that do not make us happier or we may not want things

that would makeus happier (Dolan & Kahneman, 2008). So far, such anticipation-biases have hardly been incorporated

into economic modeling. A noteworthy exception is the study by Loewenstein,O’Donoghue, and Rabin (2003), being

the first to analyze how systematic errors in anticipating adaptation processes mayaffect the c hoicespeople make and

lead to economic outcomes that are not in their own best interest. We will return to their study below.

If individuals’ decisions lead to suboptimal outcomes for themselves, market corrections by the government may

improve the individuals’ welfare. Behavioral economics has identified manysituations where people do not appear to

do what is best for them. Therefore, policy interventions may actually lead people to do things that improve their own

well-being. The normative concept of methodological individualism does not justify such governmentintervention but

many economists have nowadays become supportive of paternalistic interventionsthat help people to avoid system-

atic decision errors without curtailing individual autonomy (cf.Loewenstein & Ubel, 2008). We understand our paper

as a positive analysis of the consequences of adopting the normative concept of optimal paternalism. In particular,the

purpose of our study is to examine how the governmentcould use income tax policy to correct for behavioral mistakes

caused by anticipation-bias that we label adaptation internalities. Thereby,we will integrate this paternalistic motive

to correct for internalities that people impose on themselves into a frameworkfor optimal redistributive taxation with

heterogeneous agents. Taxesare particularly interesting as a means of correction because they leave people with the

freedom of choice, while at the same time altering individual incentives to eliminate the behavioral failure. Hence, dis-

cussing the policy options for a paternalistic government as we do in this paper is important for both supporters and

opponents of the normative concept of paternalism, because one should be aware of the consequences for a govern-

ment that becomes paternalistic and these consequences can only be analyzed in a rigorous theoretical frameworklike

the one we have developed.

Our model is based on an extension of the framework used by Loewenstein, O’Donoghue, and Rabin (2003), who

developa simple overlapping-generations “eat-the-cake”model with habit formation to illustrate how anticipation-bias

leads to misallocation of the consumption of a given resource stock overtime. We extend their model by incorporating

production and labor supply decisions. Each consumer is assumed to live for three periods, which is the minimum num-

ber of periods required to distinguish between a commitment solution in which the consumer follows the original life

time plan for consumption and work hours, and a solution where the consumer reoptimizes in the second period of life

when realizing that he/she has already adapted more than originally anticipated. We therebymodel anticipation-bias

as an internal process, where individuals systematically underestimate the extent to which they adapt to their cur-

rent consumption. This makes our paper distinctly different from the recent contribution by Tuomalaand Tenhunen

(2013), where anticipation-bias is modeled by combining habit formation with the assumption that the consumers and

the government use different utility discount factors. Wewill return to their study below.

The tax instruments faced by the government are nonlinear taxes on labor income and capital income. A major

advantage with a nonlinear tax system is that tax wedges have a natural interpretation in terms of an optimal choice

made by the government instead of reflecting functional form assumptions or the necessity to raise revenue, allow-

ing us to focus on the policy incentives created by anticipation-bias in the simplest possible way.The optimal tax pol-

icy is derived by comparing the choices individuals make with the choices they would have made in the absence of

anticipation-bias—which we consider as the benchmark of an intervening paternalistic government.

We show that a paternalistic government has an incentive to use both tax instruments to correct for the

anticipation-bias, because there are two channels by which the utility from future consumption is affected. Adapta-

tion implies that if people work more and save less when they are young, they get used tohigher consumption levels,

which means that the utility derived from any given level of future consumption will fall. If people correctly foresee

this adaptation to higher consumption, they will take into account this fall in future utility and makeoptimal intertem-

poral labor supply and savings decisions. In the presence of an anticipation-bias, however,they are likely to work too

much and savetoo little, at least when middle-aged; for young consumers, anticipation-bias may either induce the con-

sumer to save too much or too little. Our results show how the simultaneous use of marginal labor income taxes and