Optimal mixed taxation, credit constraints, and the timing of income tax reporting

• Date: 01 August 2019
• DOI: http://doi.org/10.1111/jpet.12382
• Published date: 01 August 2019

J Public Econ Theory. 2019;21:708–737.wileyonlinelibrary.com/journal/jpet708

|

© 2019 Wiley Periodicals, Inc.

Received: 14 August 2018

|

Revised: 4 March 2019

|

Accepted: 15 April 2019

DOI: 10.1111/jpet.12382

ORIGINAL ARTICLE

Optimal mixed taxation, credit constraints,

and the timing of income tax reporting

Robin Boadway

1

|

Jean‐Denis Garon

2

|

Louis Perrault

3

1

Queen’s University and CESifo,

Kingston, Canada

2

Université du Québec à Montréal and

CESifo, Montreal, Canada

3

Georgia State University, Atlanta,

Georgia

Correspondence

Jean‐Denis Garon, Université du Québec

à Montréal and CESifo, Montreal, H2X

1L7, Canada.

Email: garon.jean-denis@uqam.ca

Abstract

We study optimal income and commodity tax policy with

credit‐constrained low‐income households. Workers re-

ceive an even flow of income during the tax year, but

report their incomes and make tax payments (receive

transfers) at the end of the year. They spend their

disposable income on multiple commodities over the

year. We show that differentiated subsidies on commod-

ities can be optimal even if the Atkinson–Stiglitz

Theorem conditions apply. When the optimal policy

leaves low‐income households with binding credit con-

straints, it may be optimal to subsidize differentially the

good that they consume in higher proportion. Uniform

subsidies would also relax the credit constraint, but

would be more costly to the government since they would

equally benefit unconstrained households. Numerical

examples suggest that commodity tax differentiation

increases with basic needs and with the interest rate at

which government borrows.

KEYWORDS

basic needs, commodity taxation, credit constraints, redistribution

1

|

INTRODUCTION

Whether redistribution should be achieved through direct transfer payments or through

differentiation of commodity tax rates is an old debate among economists. While there is

widespread consensus as to the efficacy of targeting redistribution through income‐tested

transfers, maintaining certain VAT exemptions for redistributive purposes is still an open

debate. For instance, Crawford, Keen, and Smith (2010) propose to broaden VAT tax bases by

abolishing zero‐rating and exemptions, and to use the proceeds to increase all income‐tested

transfers by 15%. However, they acknowledge the technical difficulties related to applying VAT

to goods such as durable and financial services.

1

Moreover, enacting comprehensive tax base

reforms can be politically contentious.

While the appropriateness of replacing all VAT exemptions by income‐tested transfers is still

debated, we offer an argument in favor of maintaining some of them on redistributive grounds.

We justify differentiated commodity taxation and subsidies in an environment where credit

constraints are a feature of the economic environment of households who must fulfill their

basic needs. The foundation of our results lies in the fact that many transfer programs are

income‐tested and delivered through the tax system based on incomes reported at the end of the

tax year.

2

This induces both lumpiness and delays in delivering transfers to low‐income

recipients. Those with low enough income to be eligible for a transfer from the government will

have low income during some parts of the tax year and large end‐of‐period transfers. The fact

that such transfers are often spread over the subsequent year only delays their payment further.

Even more crucially, governments most often condition transfers on past years’information

since keeping track of individuals’changing circumstances is costly.

The implication is that in many cases the poorest working individuals will only receive the

appropriate amount of transfer when the fiscal year is over. With perfect credit markets, individuals

who expect to receive a transfer would prefer smoothing their consumption stream over the year by

borrowing. However, they may be precluded from doing so by a credit constraint. Financial

institutions may be unwilling to lend to them except at exorbitant interest rates, especially if they do

not have a credit rating or if the financial institution cannot verify the expected transfer.

We address this issue in a theoretical framework that features an optimal mixed taxation

system consisting of nonlinear income and linear commodity taxes. The economy contains

several skill types of households who supply labor and consume two commodities. The economy

lasts for one discrete period representing a tax year, but households transact throughout the

period. To simplify matters, we divide the period into two subperiods and assume that

transactions can occur at two discrete points: in the middle of the period and at the end.

Preferences are weakly separable so in the absence of credit constraints, the Atkinson–Stiglitz

Theorem will apply: Optimal commodity taxes will be uniform at indeterminate rates given that

proportionalcommodity taxation is equivalentto proportional income taxation.In the presence of

credit constraints, the two commodities are not consumed in the same proportions by different

skill types, and this will leadto differential commodity subsidization. We show that differentiated

commodity taxes and subsidies are optimal when it is costly for the government to reallocate

transfers from the end of the tax year to within the year.

3

Central to our argument is the claim that the redistributive efficacy of the tax‐and‐credits system

is compromised when the poorest individuals cannot use the transfers when they need them for

consumption. There is convincing empirical evidence that credit constraints may be a serious issue

for redistribution even in presence of direct transfers. Agarwal, Liu, and Souleles (2007) find that

liquidity‐constrained individuals are the biggest tax rebate spenders and exhibit hand‐to‐mouth

behavior. Bertrand and Morse (2009) also study what people do with their tax rebates. They have

1

See also Cremer and Gahvari (1995).

2

Examples include refundable tax credits that decline in income, such as the Earned Income Tax Credit (EITC), the Additional Child Tax Credit, and the Health

Coverage Tax Credit in the United States; the Canada Workers Benefit, the Canada Child Benefit, and the Goods and Services Tax Credit in Canada; and the

Working Tax Credit, the Child Tax Credit, and the Universal Credit in the United Kingdom.

3

The cost is due to an interest rate spread between the borrowing cost and the interest on saving. This causes the interest rate that the government faces when

borrowing to transfer funds to earlier in the year to exceed the interest return that individuals receive when they same some of the funds transferred. The cost of

transferring funds could be generalized to compliance and administrative costs.

BOADWAY ET AL.

|

709