How much should we fund the IRS?

• Published date: 01 February 2022
• Author: Hanjoon M. Jung,Meng‐Yu Liang,C.C. Yang
• Date: 01 February 2022
• DOI: http://doi.org/10.1111/jpet.12532

J Public Econ Theory. 2022;24:120–139.wileyonlinelibrary.com/journal/jpet120

|

© 2021 Wiley Periodicals LLC

Received: 3 November 2020

|

Revised: 17 April 2021

|

Accepted: 20 May 2021

DOI: 10.1111/jpet.12532

ORIGINAL ARTICLE

How much should we fund the IRS?

Hanjoon M. Jung

1

|Meng‐Yu Liang

2

|C.C. Yang

2,3,4

1

Ma Yinchu School of Economics,

Tianjin University, Tianjin, China

2

Institute of Economics, Academia

Sinica, Taipei, Taiwan

3

Department of Public Finance, National

Chengchi University, Taipei, Taiwan

4

Department of Public Finance, Feng

Chia University, Taichung, Taiwan

Correspondence

Hanjoon M. Jung, Ma Yinchu School of

Economics, Tianjin University, Building

25‐B, No. 92, Weijin Road, Nankai

District, 300072 Tianjin, China.

Email: hj.jung@tju.edu.cn

Abstract

Tax enforcement not only detects tax evasion (the

direct effect) but also enforces voluntary compliance

(the indirect effect). This paper explores implica-

tions of the indirect effect for the optimal size of

the budget allocated to the Internal Revenue Service

(IRS). The conventional wisdom prescribes that the

size of the budget allocated to the IRS should always

fall short of the level where an additional dollar

allocated would return just an additional dollar of

revenue. In a model where taxpayers are subject to a

small amount of incomplete information, we show

that after taking into consideration the indirect

effect of tax enforcement, it is possible that the op-

timal prescription is opposite to the conventional

wisdom.

1|INTRODUCTION

As far as the collection of tax revenue is concerned, the true function of the Internal

Revenue Service's (IRS) enforcement may not lie in detecting tax evasion (the

direct effect) but in preventing tax evasion or enforcing voluntary compliance (the

indirect effect). This view is clearly expressed by the National Taxpayer Advocate

(2004,II‐4):

The IRS estimates that the annual net tax gap is about $255 billion. …Studies

show that examinations [audits] reduce the tax gap primarily through an in-

direct effect on voluntary compliance rather than direct collection from the

taxpayer under audit.

This paper explores implications of the indirect effect of tax enforcement for the size of the

budget allocated to the IRS.

1

In particular, we ask: How much should we fund the IRS?

Slemrod and Yitzhaki (1987, hereafter SY) investigate the same normative question as we

do in our paper. They prescribe that the size of the budget allocated to the IRS should always

fall short of the level where an additional dollar allocated would return just an additional dollar

of revenue. We call SY's prescription the

M

CM

R

<rule. This rule, dubbed as the conventional

wisdom, is also prescribed by Usher (1986), Kaplow (1990), Mayshar (1991), and Sanchez and

Sobel (1993).

2

In this paper we come up with a plausible case where the budget size allocated to the IRS is

associated with

M

CMR>=0

, indicating a complete waste of the IRS's enforcement activities

at the margin in terms of tax revenue collected. Despite the complete waste, we show that, once

there is a cutting down on the allocated budget size, a disaster in terms of voluntary compliance

(the indirect effect) will arise: Taxpayers move en masse from compliers to evaders. To prevent

the disaster from occurring, the optimal budget size allocated to the IRS satisfies the opposite of

the conventional wisdom; that is, the

M

CM

R

>rule holds at the optimum.

In their seminal work, Graetz, Reinganum and Wilde (1986, hereafter GRW) address

“strategic compliers,”whose size in the population is endogenously determined by taxpayers'

responses to the IRS's enforcement. This paper builds on and extends the GRW approach.

First, GRW consider a game in which taxpayers and the IRS move simultaneously. We

revise their simultaneous game to a sequential game.

Second, while GRW assume that the IRS's objective is to maximize the tax revenue collected

net of audit costs in the absence of the budget constraint, we assume that the IRS expends all

the audit budget allocated by the Congress in the presence of the budget constraint. Assuming

that the IRS maximizes its “profit”may leave large allocated budgets unspent in a fiscal year.

This contradicts the well‐established budget‐maximizing hypothesis of government bureaus put

forth by the classic work of Niskanen (1971).

Third, GRW consider a model in which taxpayers possess complete information, whereas

we consider a model in which taxpayers suffer from incomplete information in the vein of the

so‐called “global game,”as addressed by Carlsson and van Damme (1993), Morris and Shin

(1998), and Heinemann and Illing (2002), among others. When information is complete for

taxpayers, imposing budget constraints on the IRS will often give rise to multiple equilibria as

shown by Graetz et al. (1984), Kim (2005), Bassetto and Phelan (2008) and Sanchez Villalba

(2015). As a result, it encounters the difficult problem of equilibrium selection and succumbs to

the self‐fulfilling nature of agents' beliefs, which has little to do with economic fundamentals.

An important aim of the global game is to show, through its being tightly tied to economic

fundamentals, that multiplicity will vanish once the economy is perturbed away from the

complete‐information framework. Our analysis critically relies on the result of a unique

equilibrium.

1

We abstract from addressing the case where the IRS spends budget on improving the 3rd party reporting system. The

tax gap study by IRS (2016) shows that the compliance rate for income subject to substantial information reporting and

withholding (such as wages and salaries) is 99%, while the compliance rate for income subject to little or no in-

formation reporting (such as nonfarm proprietor income) is only 37%. The IRS finding reveals two things. First, to

prevent evasion, it is of critical importance to let taxable income be covered by the 3rd party reporting system as much

as possible. Second, the task to achieve the first thing is not easy, since the fact that there is a high evasion rate for

nonfarm proprietor income has been known for a while.

2

The intuition behind the conventional wisdom is provided in Section 5.1.

JUNG ET AL.

|

121