A TALE OF TAX POLICIES IN OPEN ECONOMIES

• Author: Aurélien Eyquem,Stéphane Auray,Paul Gomme
• DOI: http://doi.org/10.1111/iere.12199
• Date: 01 November 2016
• Published date: 01 November 2016

INTERNATIONAL ECONOMIC REVIEW

Vol. 57, No. 4, November 2016

A TALE OF TAX POLICIES IN OPEN ECONOMIES∗

BYST´

EPHANE AURAY,AUR ´

ELIEN EYQUEM,AND PAUL GOMME1

CREST-Ensai and Universit´

e du Littoral Cˆ

ote d’Opale, France; CREST-Ensai, CNRS, and

GATE-Lyon Saint-Etienne, France; Concordia University and CIREQ, Canada

To evaluate fiscal policy reforms for Euro-area countries, this article develops and calibrates a small open economy

model. Debt reduction reforms require higher tax rates in the short term in exchange for lower rates in the long term

as the debt-servicing burden falls. Using the capital income tax to implement such a policy leads to welfare gains;

the consumption tax, a very small welfare gain; and the labor income tax, a welfare loss. Holding fixed the long-run

debt–output ratio, offsetting a lower capital income tax with either a higher labor income or consumption tax generally

yields welfare gains.

1. INTRODUCTION

The fiscal situation in Greece, Italy, Ireland, Portugal, and Spain is grim. France’s situation

is not much better although it has, to date, avoided a crisis. These observations all point to the

need for developed countries to get (and keep) their fiscal houses in order.

One need not be a “debt nutter” to think that there are benefits from reducing the size

of government debt, at least in the long run. In particular, given the real interest rate, every

percentage point drop in the debt–output ratio leads to a proportional decline in the primary

deficit in the long run. Although reducing the debt–output ratio requires raising taxes in the

short term, doing so allows for lower tax rates in the long run as the government debt servicing

burden drops.

We build several dynamic general equilibrium, macroeconomic models to assess the welfare

implications of various tax reforms. The benchmark model is an incomplete international asset

markets, small open economy model with two goods, home and foreign. To evaluate the im-

portance of some of the key features of the benchmark model, several related alternatives are

considered. The first is a complete markets version of the benchmark model. Relative to the

benchmark model, the key difference is that the Euler equation governing the accumulation of

net foreign assets is replaced by a risk sharing condition; see Section 2.2 for details. The second

variant is a one-good version of the benchmark model, similar in spirit to Mendoza (1991). The

final version is of a closed economy.

Two calibrations of the model are presented. The first calibrates to a high government debt

subgroup of Euro Area countries consisting of Greece, Italy, Ireland, Portugal and Spain (GIIPS

hereafter); the second calibration collects together Euro Area countries with lower government

debt levels (EA7 hereafter, composed of Austria, Belgium, Finland, France, Germany, Luxem-

bourg, and the Netherlands).

∗Manuscript received December 2014; revised May 2015.

1We thank the associate editor, Pierre Yared, as well as two anonymous referees for their comments. We also

thank Stefano Gnocchi, Enrique Mendoza, and Evi Pappa and participants of the Oviedo CFE conference, the 9th

ENSAI Economic Days, the T2M 2014 Conference, the 2014 North American Meeting of the Econometric Society,

and seminar participants at the Aix-Marseille School of Economics, Evry, CREST, and SUNY Albany. S. Auray

gratefully acknowledges financial support of the Chair ACPR/Risk Foundation: “Regulation and Systemic Risk.” The

usual disclaimer applies. Please address correspondence to: St´

ephane Auray, CREST-Ensai and Universit´

e du Littoral

Cˆ

ote d’Opale. ENSAI, Campus de Ker-Lann, Rue Blaise Pascal, BP 37203, 35172 BRUZ cedex, France. E-mail:

stephane.auray@ensai.fr.

1299

C

(2016) by the Economics Department of the University of Pennsylvania and the Osaka University Institute of Social

and Economic Research Association

1300 AURAY,EYQUEM,AND GOMME

The first set of reforms lowers the long-run government debt–output ratio by 10 percentage

points, a figure chosen as being sizable but not implausible. This reduction in the government

debt-to-output ratio is financed via one of the following: the consumption tax, the labor income

tax, or the capital income tax. To ensure stability of the debt dynamics, the government follows a

fiscal policy rule that links deviations of the primary surplus from its long-run target to deviations

of the debt-to-output ratio from its long-run target. Simply put, when the debt-to-output ratio is

above target, one of the tax rates is adjusted to push the primary surplus above its long-run value.

When the labor income tax rate is used to implement such a government debt reduction, the

change in policy distorts households’ decisions through the labor–leisure choice. In this case, all

four model variants deliver similar dynamics. The short-term increase in the labor income tax

rate reduces the after-tax real wage, leading households to cut back on their hours worked. In

turn, output and consumption also fall. This policy-induced recession lasts roughly a decade.

Naturally enough, the capital income tax rate affects the after-tax return to capital and so

affects capital accumulation. In this case, there are large differences across the four model

variants that can be traced to the behavior of the real interest rate. To understand these

differences, start with the simplest model, the one of a closed economy. A “no arbitrage”

condition dictates that the after-tax return to capital equals the return on government debt. The

initial increase in the capital income tax rate lowers the after-tax return to capital. The rate of

return equality condition then requires a drop in the real interest rate on government debt. In

the one-good open economy model, there is a third asset return to consider, that on international

bonds. Since the model is of a small open economy, the return on international bonds is close

to constant. Further, since it is a one-good model, the real exchange rate is, necessarily, fixed

(at one). Then the return on international bonds must equal that on domestic government debt,

which, in turn, must equal the after-tax return on capital. Because all rates of return are equal

and close to constant, the increase in the capital income tax rate necessitates a large drop in

the capital stock in order to increase the marginal product of capital so that the after-tax return

to capital is unchanged. Next, consider the benchmark model with incomplete international

markets and two goods. Although the world real interest rate is not very responsive to domestic

events, in this model there is a real exchange rate channel that affects the effective world real

interest rate. This real exchange rate effect leads to a greater response of domestic returns,

and so, on impact, the capital stock does not need to fall so much. The dynamics of the two

good, complete markets model are very close to those observed under incomplete markets in

the short run. In the long run, the cumulative effects of shutting down the international wealth

effects (under complete markets) lead to substantial differences in final steady states.2

The effects of the consumption tax operate through both the after-tax real wage, and asset

returns. Although it is generally known that the effects of an increase in the consumption tax

affect the labor–leisure choice in much the same way as an increase in the labor income tax

rate, the asset return channel is, perhaps, more subtle. In particular, what matters for asset

returns is the timing of the consumption tax. The debt reduction scenarios involve initially

raising the consumption tax, then gradually reducing it. Apart from the first period, future

consumption tax rates are lower than current tax rates. As a result, the temporal pattern of

the consumption tax acts as a subsidy to all asset returns. However, the size of this subsidy

effect is small relative to those seen under the capital income tax. Consequently, the short-term

dynamics of macroeconomic variables look broadly similar to those associated with the labor

income tax, although the subsidy effect cuts the length of the policy-induced recession roughly

in half.

To easily summarize the welfare results, we need to take a stand on which of the four mod-

els is our preferred model. Although the closed economy model provides useful insight to

the model’s mechanisms, Euro Area countries can hardly be described as closed economies.

Treating the GIIPS and EA7 regions as one-good open economies means fixing the real ex-

change rate, an assumption that does not fit the facts. The one-good model also predicts that a

2See Auray and Eyquem (2014) for a discussion of differences between complete and incomplete markets.

TAX TALES 1301

region will either import goods or export goods but not both simultaneously; yet, these regions

have considerable imports and exports. Although complete international asset markets may

have desirable theoretical properties, it is evident to most observers that the GIIPS countries

have experienced considerable pain over the past few years; such pain seems inconsistent with

the risk sharing inherent to complete asset models. For the benchmark model, using the capital

income tax rate as the policy instrument yields the largest welfare gains, 0.16% of consumption

for the GIIPS group and 0.36% for the EA7 countries. The consumption tax is associated with

a small but positive welfare benefit whereas the labor income tax delivers a small welfare loss.

These welfare effects are computed over the full transition path. In contrast, “na¨

ıve” welfare

gains computed across steady states are considerably larger—as much as 2.7%.

The second set of policy reforms is inspired by the optimal taxation literature, which typically

finds that in the long run, capital income taxes should be close to zero. With this in mind,

leaving the long-run debt–output ratio unchanged, consider a 10 percentage point reduction

in the capital income tax rate. To satisfy the fiscal policy feedback rule, the capital income

tax cut is financed through either the labor income tax or the consumption tax. The short-run

dynamics of such a reform are dominated by the effects of the capital income tax rate cut. For

the benchmark model, there are sizable welfare gains associated with such a reform: For the

GIIPS group, the gains are either 0.24% (labor income tax) or 0.40% (consumption tax); for

the EA7, 0.58% (labor income tax) or 0.77% (consumption tax). Such results should dissuade

policymakers from a scenario like the following: Having raised the capital income tax rate to

bring down the government debt–output ratio, it may be tempting to maintain this increase

in the capital income tax rate, lowering instead one of the other taxes. This scenario delivers

the “worst of both worlds” since the benefits of using the capital income tax rate to lower the

debt–output ratio accrue from the long-term benefits of eventually lowering this tax rate.

In many of the reforms considered, we find that the welfare gains are larger for the EA7

countries than for the GIIPS. The chief reason is that taxation is initially higher in these

countries (and debt-to-output lower), generating larger potential efficiency gains from lower

long-run taxation in these countries.

Our article is broadly related to a variety of papers. As in the early contribution of Mendoza

and Tesar (1998), the model is unabashedly neoclassical, focusing exclusively on fiscal policy.

The first key difference relative to Mendoza and Tesar is that their model has only one good

(producible in both countries) while we distinguish between domestic and foreign goods. Sec-

ond, they have complete international asset markets although we consider both complete and

incomplete international asset markets. Third, although our tax replacement experiments are

in the same spirit as those conducted in Mendoza and Tesar, they do not consider the debt re-

duction experiments that we motivate with reference to contemporary fiscal crises. Our article

is also related to Mendoza et al. (2014). The two regions they study are the European “crisis”

countries (GIIPS) and the rest of the EMU. Their analysis is conducted within a one-good

model, and, consequently, real exchange rate movements are not considered. Their focus is

on changes in factor income tax rates that can reduce government debt-to-output levels in the

crisis region. They find that adjustments in capital income taxes cannot raise sufficient revenue,

a result that arises chiefly from variable capital utilization. An important technical distinction

between our work and that of Mendoza et al. is that they solve their model through log-linear ap-

proximations whereas we solve the nonlinear equations of our model, a procedure that delivers

more accurate transitional dynamics and is better suited to welfare comparisons of alternative

policies.

Like Mendoza and Tesar (1998) and Mendoza et al. (2014), our article is not about the

optimal structure of taxation, although it is broadly related to that body of work. The bulk of

the optimal taxation literature has focused on closed economy models; see Lucas and Stokey

(1983) or Chari and Kehoe (1999) among many others. The open economy dimension has

received less attention. Recently, Benigno and De Paoli (2010) characterized optimal fiscal

policy in a small open economy model with a single income tax and public debt; they consider

both steady-state and business cycle fluctuations. In a richer environment with capital, sticky