Assessing Nonlinear Dynamics of Central Bank Reaction Function: The Case of Mozambique

• DOI: http://doi.org/10.1111/saje.12149
• Published date: 01 March 2017
• Author: Gerson Nhapulo,João Nicolau
• Date: 01 March 2017

ASSESSING NONLINEAR DYNAMICS OF CENTRAL BANK

REACTION FUNCTION: THE CASE OF MOZAMBIQUE

GERSON NHAPULO

†

AND JOA

˜O NICOLAU*

‡

Abstract

This paper sheds some light on the elements governing monetary policy-making during the

period 2000Q1–2015Q1 in Mozambique. We estimate a time-varying Taylor-type rule for the

BM, using a Markov-switching (MS) model and a Threshold model. The general finding is that

the behaviour of the BM can be characterised by two regimes. In regime 1, only changes in infla-

tion trigger a reaction by the monetary authority. This behaviour is prominent after the establish-

ment of the monetary policy committee in 2007 (CPMO). In regime 2, the BM reacts

aggressively both to cool off the economic activity and to curb inflationary pressures. Regime 2

occurred most frequently during 2000–2006, when the fiscal policy might have played an impor-

tant role in output stabilization. After the establishment of the CPMO, regime 2 occurred in the

context of a steep rise in fuel and food prices in 2007–2008 and in 2010. Both the MS model

and the Threshold model show similar asymmetric effects. We find evidence that inflation is

viewed more seriously by the monetary authorities when it is accompanied by a high output-gap

in the previous period, which triggers a more aggressive response from the monetary authorities.

JEL Classification: E58, E52, C54

Keywords: Taylor rule, Bank of Mozambique, Markov-Switching model, threshold autoregressive

model

1. INTRODUCTION

Since the early 90s, the economic literature in the realm of monetary policy actions in

terms of reaction function has been gaining momentum and leading economists use

monetary aggregates and short-term interest rates as policy instruments (see Aragon and

Portugal, 2010; Hutchison et al., 2013). Nevertheless, the question on whether rules or

discretion should be used to conduct the monetary policy is still unresolved. In practice,

modern policy rules, such as the McCallum rule and Taylor rule (McCallum, 1987; Tay-

lor, 1993), represent a constrained monetary authority’s discretion as long as they have

some flexibility in emphasizing alternative short-run inflation stabilization objectives (see,

for instance, Patra and Kapur, 2012).

McCallum (1987) suggested a rule whereby the monetary authority targets a nominal

income, using a monetary base as the main policy instrument. However, the ascendancy

* Corresponding author: ISEG Mathematics Rua do Quelhas, 6 Lisbon 1200-781. E-mail:

nicolau@iseg.ulisboa.pt

†

Bank of Mozambique, Research and Statistics Department, Central Bank of Mozambique.

‡

ISEG (Lisbon School of Economics & Management), Universidade de Lisboa and

CEMAPRE.

Jo~ao Nicolau was partially supported by the Project CEMAPRE - UID/MULTI/00491/2013

financed by FCT/MEC through national funds.

V

C2017 Economic Society of South Africa. doi: 10.1111/saje.12149

28

South African Journal of Economics Vol. 85:1 March 2017

South African Journal

of Economics

of the new Keynesian framework from the late 90s, inter alia, has implied a downgrading

of monetary aggregates and an upgrading of the Taylor rule in explaining the operational

conduct of monetary policy. Taylor (1993) proposed a rule in which the monetary

authority adjusts its policy instrument interest rate in reaction to deviations of output and

inflation in relation to their targets (see Chen and Werner, 2011; Hutchison et al., 2013).

Since 2000, the monetary authority of Mozambique, the Bank of Mozambique (BM),

has de jure adopted monetary targeting but, from 2011 it started to de facto switch to infla-

tion targeting lite (ITL) (this term was coined by Stone, 2003, and it is presented in section

(iv) in “Literature Review”). In this circumstance, Fernandes (2011) assessed the interest

rate transmission mechanism covering the sample period 1993M1–2009M4. Using the

marginal lending facility as a proxy of the policy interest rate, Fernandes found that devia-

tions in inflation and output gap take 2 and 4 months, respectively to bring about changes

in policy rate. Based on the Johansen cointegration test, Fernandes (2011) also estimated

the implied interest rate reaction function of the BM, and found evidence of more sensi-

tivity of the policy rate to inflation deviations than to the output gap and a high degree of

smoothing rate. That kind of monetary policy reaction function (MPRF), which we

address in more detail in section (i) of “Literature Review”, is referred to as a linear func-

tion given that it assumes constant and evenly weighted coefficients of target variables (i.e.

equal weights to deviations below and above the target). This approach, however, is often

criticised on the grounds that the monetary authorities may have asymmetric responses to

output stimulus and inflation stabilisation, thus, giving rise to a nonlinear monetary policy

reaction (see for instance, Aragon and Portugal, 2010; Sznajderska, 2014).

In view of this criticism, our analysis aims to provide an empirical assessment of the

nonlinear dynamics of the implied interest rate reaction function of the BM during the

sample period (2000Q1–2015Q1). Dynamic MPRFs can shed some light on the ele-

ments governing monetary policy-making over time, such as, whether inflation is given

more emphasis or not.

The paper is organised as follows. The next section provides the theoretical framework

and literature review. Section 3 describes the evolution of the monetary framework and

the main reforms in the financial sector in Mozambique in the post-independence

period. Section 4 outlines the estimation strategy and specifies the data and the variables

used. Section 5 describes the methodology and presents the main results. Finally, section

6 provides the concluding remarks.

2. THE MONETARY POLICY-MAKING

Central banks usually perceive price stability as the main goal of monetary policy, but

the mandates of the monetary authorities may comprise other objectives such as eco-

nomic growth and financial stability.

2.1 Theoretical Framework

One way to assess short to medium-run risks to price stability is to use the New Keynes-

ian Model (NKM) instead of the New Classical Model in which monetary policy is

essentially neutral to the real activity (Goodfriend and King, 1997; Romer, 2012).

Following Woodford (2007), one can specify the NKM model using three equations:

Aggregate supply (Phillips curve)

29South African Journal of Economics Vol. 85:1 March 2017

V

C2017 Economic Society of South Africa.