Government‐spending multipliers and the zero lower bound in an open economy

• Date: 01 November 2017
• Published date: 01 November 2017
• Author: Charles Olivier Mao Takongmo
• DOI: http://doi.org/10.1111/roie.12295

ORIGINAL ARTICLE

Government-spending multipliers and the zero lower

boundinanopeneconomy

Charles Olivier Mao Takongmo

University of Ottawa, Canada

Correspondence

Department of Economics, University of

Ottawa, 120 University, Social Sciences

Building, Room 9005, Ottawa, Ontario,

Canada, K1N 6N5.

Email: maotcharles@yahoo.fr

JEL Classification: E3, E4, F5, E62, F41,F44

Abstract

This paper assesses the size of the government-spending

multiplier in an open economy when the zero lower bound

(ZLB) on the nominal interest rate is binding. In a theoretical

framework, in a closed economy, other authors have shown

that when the nominal interest rate is binding the

government-spending multiplier can be very large (close to

four). Their theory helps illuminate the government-

spending multiplier in the ZLB, but it is difficult to match

that theory with the data. We argue that, in an open econ-

omy, another channel exists for the crowding-out effect via

the real excha nge rate. For an open economy , the

government-spending multiplier is not large owing to the

appreciation of the real exchange rate, induced by the appre-

ciation of aggregate demand that follows the increases in

government spending. To test the robustness of our open

economic model, we conduct the same analysis in a corre-

sponding closed economy model. The result from our closed

economy model confirms the result obtained in the other

work. Our theoretical results are consistent with the results

obtained in the empirical literature, which uses the vector

autoregressive method and the structural vector autoregres-

sive approach to measure the impact of government-

spending shock on the real gross domestic product and

revealed that the government-spending multiplier tends to be

lower in open economy.

1

|

INTRODUCTION

What is the size of the fiscal multiplier in an open economy when the nominal interest rate reaches the

zero lower bound (ZLB)? Using a theoretical framework in a closed economy, Christiano, Eichen-

baum, and Rebelo (2011) show that, when the nominal interest rate is binding, the government-

spending multiplier can be close to four. This theory helps us to understand the dynamics of an

1046

|

V

C2017 JohnWiley & Sons Ltd wileyonlinelibrary.com/journal/roie Rev IntEcon. 2017;25:1046–1077.

Received: 9 June 2014

|

Revised: 25 March 2017

|

Accepted: 28 March 2017

DOI: 10.1111/roie.12295

economy in ZLB following an increase in government spending, but it is difficult to match the Christi-

ano et al. (2011) theory with the data. For example, during the financial crisis and recession that fol-

lowed in 2007, interest rates in the United States and European countries reached their lowest levels.

Many significant budget plans emerged, including the American Recovery and Reinvestment Act

(ARRA) in the United States (U.S.$831 billion from 2009 to 2019) and the European Economic

Recovery Plan (EERP) in the European Union (e200 billion from 2008 to 2010). However, such gov-

ernment spending did not have the expected result. As Boskin (2012) pointed out, the ratio of debt to

gross domestic product (GDP) in the United States increased on average by 40.5 percent before 2008

and by 80 percent after 2008.

In this paper, we suggest that the real exchange rate is a channel that can explain why increasing

government spending in ZLB may not lead to a large government-spending multiplier. We propose a

theoretical open macroeconomics framework with market imperfections, wage and price rigidities, and

an endogenous smoothing monetary policy. In our framework, we introduce a shock to the discount

factor that pushes the nominal interest rate to its minimum level. We then compute the government-

spending multiplier in the ZLB. To test the robustness of our open economic model, we also conduct

the same analysis in a corresponding closed economic model.

In a closed economy and in the presence of ZLB, there is no crowding-out effect through interest

rates. We show that, in an open economy, another channel exists for the crowding-out effect via the

real exchange rate and the multiplier is not large. We show that increasing government spending

increases aggregate demand, which leads to appreciation of the real exchange rate that is greater than

the appreciation that we would have had if the lower nominal interest rate was not binding. The appre-

ciation of the real exchange rate then reduces the fiscal multiplier. Our closed economy framework

confirms that in a closed economy, the government-spending multiplier is very large (close to four).

That result is also obtained by Christiano et al. (2011) in a closed economy.

Our finding on the open economy is consistent with the results obtained in the empirical literature

(see, e.g., Perotti, 2005; Karras, 2012; Ilzetzki, Mendoza, & V

egh, 2013), which use the vector autore-

gressive (VAR) and the structural vector autoregressive (SVAR) approaches to show that the

government-spending effect on the GDP tends to be lower in an open economy. Even if those authors

use data that does not span the ZLB period, their results still provide an excellent idea of what we

could encounter during ZLB periods regarding the rule played by the real exchange rate.

1.1

|

Literature review

Christiano et al. (2011), using a theoretical model, find that the government-spending multiplier can be

much larger than one (close to four) while the nominal interest rate reaches the ZLB. However, the

framework built in a closed economy cannot take into account the effect of government spending on

the real exchange rate or its effect on the level of the trade balance deficit. These effects can have a

real impact on the cost of imported goods and consumption and, therefore, on the multiplier of public

spending. The mechanism explaining the size of the government-spending multiplier in a closed econ-

omy in ZLB is described by Christiano et al. (2011) as follows:

Following an increase in government spending, there is an increase in production, marginal cost

and expected inflation. This causes a decrease in the real interest rate, and households consume

more. The increase in household spending increases output, marginal cost and expected infla-

tion. This further decreases the real interest rate and so on, which in turn leads to a significant

increase in production.

The theory may differ in an open economy.

MAO TAKONGMO

|

1047

Concerning the theory, the basic framework is developed by Mundell (1963). The model predicts

that in a small open economy with flexible exchange rates, a fiscal policy is ineffective

1

if capital

mobility is perfect. Indeed, an increase in government spending financed by borrowing creates an

excess demand for goods, which tends to increase income. This increases the demand for money and

the interest rate, attracting foreign capital. The exchange rate then appreciates, which in turn leads to

an equivalent decrease

2

in income through a trade imbalance. Even if the Mundell (1963) framework

is very restrictive,

3

the results described in the model are simple and understandable.

Concerning empirical analysis, using different empirical methodologies, many authors find

government-spending multipliers to be more or less close to one. The size of the multiplier depends on

the method, period and government-spending indicator being taken into account. However, it is clear from

empirical literature that the multiplier is approximately one. Fisher and Peters (2010) use as an indicator of

government spending the impact on income of the largest companies with government contracts in the

military sector. They find a multiplier of government spending equal to 1.5 over a period of 5 years.

The narrative approach

4

developed by Ramey and Shapiro (1998) identifies the response of the econ-

omy as a result of a sustained and unpredictable increase of exogenous government spending. Ramey

and Shapiro (1998) shows that the government-spending multiplier is sector specific. Ramey and Shapiro

(1998) finds that production and consumption decline following a government-spending shock.

5

1.2

|

Methodology

Our model is a dynamic stochastic general equilibrium model that builds on Ambler, Dib, and Rebei

(2004) and Amano and Shukayev (2012) and allows a ZLB environment. We allow our model to gen-

erate a time-variant discount factor that will help us push the interest rate to its lowest level. This is the

only type of shock in the absence of capital and a risk premium on capital that could push the interest

rate to its lower level (see Amano & Shukayev, 2012). We also consider an endogenous monetary pol-

icy. The policy states that “due to the shock, the monetary authorities set the nominal interest rate such

that it converges smoothly to the lowest level, but remains positive and differentiable at all points.”

This is a modified version of monetary policy used by Christiano et al. (2011) and it will help us use

an existing program to solve the model. When the nominal interest rate reaches the lowest level, the

government increases spending to stimulate economic activity. We create a ZLB environment in our

model, then generate data from the model and use the data to estimate the government-spending multi-

plier. The analysis is conducted in a closed economy as well as an open economy. This helps us shed

light on the differences in fiscal multipliers in the closed and open economy in a ZLB situation.

In Section 2, the government-spending multiplier is analyzed in a closed economy in the ZLB con-

text. In the case of a closed economy, the framework is as follows. The national aggregate good is pro-

duced by competitive firms using a continuum of differentiated national intermediate goods produced

by domestic firms in monopolistic competition. Households are characterized by different types of

work and supply labor in a monopolistic competition sense. The final good (the national aggregate

good) is used for consumption, government spending, and as input in the production of intermediate

goods. Prices and wages are sticky in the sense of Calvo (1983). Prices of intermediate goods and

wages are set in advance. There is a continuum of types of job offers with constant elasticity of substi-

tution. The labor is offered by a continuum of households in monopolistic competition on wages.

Section 3 presents the open economy framework. In the case of an open economy, the final good

is now produced by competitive firms using a quantity of national aggregate goods and a quantity of

imported aggregate goods. The national aggregate good is produced by competitive firms using a con-

tinuum of differentiated national intermediate goods produced by domestic firms in monopolistic com-

petition. The aggregate imported good is produced by national competitive firms using a continuum of

1048

|

MAO TAKONGMO