Education Policies and Taxation without Commitment

• Published date: 01 October 2018
• Author: Dominik Sachs,Sebastian Findeisen
• DOI: http://doi.org/10.1111/sjoe.12246
• Date: 01 October 2018

Scand. J. of Economics 120(4), 1075–1099, 2018

DOI: 10.1111/sjoe.12246

Education Policies and Taxation without

Commitment*

Sebastian Findeisen

University of Mannheim, DE-68131 Mannheim, Germany

findeisen@uni-mannheim.de

Dominik Sachs

University of Munich, DE-80539 Munich, Germany

dominik.sachs@econ.lmu.de

Abstract

Westudy the implications of limited commitment on education and tax policies set by benevolent

governments.Consistent with real-world practices, a government can decide to subsidize different

levelsof education at different rates. A lack of commitment, however,affects the optimal structure

of education subsidies. The direction of the effectdepends on how labor taxes are designed. With

linear labor tax rates and a transfer for redistribution, subsidies become more progressive. By

contrast, if the government is only constrained by informational asymmetries when designing

taxes, subsidies become more regressive.

Keywords: Inequality; time inconsistency

JEL classification:H21; H23; I21

I. Introduction

Public finance economists have long recognized that the challenges involved

in designing optimal education policies and income tax systems are closely

related. Income taxation influences the incentive to invest in education.1

Education subsidies and policies, in turn, influence the choice of an

optimal income tax system as they directly affect both the level and

the distribution of wages. Many papers have studied the designs of

education and tax policies jointly from a normative perspective – see,

for example, Bovenberg and Jacobs (2005) for a state-of-the-art treatment

*We thankFriedrich Breyer, Leo Kaas, Fabrizio Zilibotti, Josef Zweimueller, and the two

anonymous referees. D. Sachs’ research was partly funded by a post-doctorate fellowship of

the Fritz ThyssenFoundation and the Cologne Graduate School in Management, Economics and

Social Sciences.

1See, for example, Abramitzky and Lavy (2014) for quasi-experimental evidence or Trostel

(1993) for a structural approach.

C

The editors of The Scandinavian Journal of Economics 2017.

1076 Education policies and taxation without commitment

in a heterogeneous agent model.2This strand of literature assumes that

individuals make human capital investment decisions rationally, reacting to

incentives set by the tax code and by education subsidies. Importantly, it

is assumed that the government fully commits to the income tax schedule

that it has announced before decisions on education are made.

In their work, Boadway et al. (1996) draw attention to the issue of

time consistency, in the spirit of Kydland and Prescott (1977), which is

key to the design of optimal tax and education policies. If the government

fails to credibly commit to tax policies at a time when individuals are

making education decisions, this can dramatically reduce the incentives

of young individuals to invest in human capital. In their framework,

Boadway et al. (1996) show that this results in underinvestment, and make

a case for mandatory education as a second-best policy in the presence of

commitment problems.

This paper looks at the implications of limited commitment and policy

credibility on education and tax policies from a new perspective. Consistent

with real-world practices, a government can decide to subsidize different

levels of education at different rates. The idea here is that governments

typically intervene at primary, secondary and tertiary education levels.

However, the rates at which these three education levels are subsidized

differ greatly.

We formalize this by allowing the government to set a non-linear

schedule of education subsidies. We derive our results with a transparent and

simple heterogeneous agent model with a two-type Stiglitz (1982) model.

Consistent with empirical evidence, individual wages are determined by

both innate abilities and education levels. We show that the effect of a lack

of commitment is dependent on the structure of the labor tax.

First, we analyze a linear labor income tax schedule with a lump-sum

transfer as in Sheshinski (1972). Second, we study income taxation in

the tradition of Mirrlees (1971), where the planner is only constrained by

informational asymmetries – we often refer to the latter case as non-linear

taxation.

Linear Labor Tax Rates

We start with the benchmark of full commitment. The optimal linear

income tax rate takes into account education incentives and is lower than in

Sheshinski (1972). Education subsidies for the high type are set such that

2See Richter (2009) for a recent treatment in a Ramsey setting with a representative agent.

See Da Costa and Maestri (2007)orAnderberg (2009) for a Mirrlees treatment with ex ante

homogeneous agents and uncertainty.

C

The editors of The Scandinavian Journal of Economics 2017.