The emergence of regional business cycle in Africa—a reality or myth? A Bayesian dynamic factor model analysis

• Date: 01 January 2020
• DOI: http://doi.org/10.1111/twec.12888
• Published date: 01 January 2020

World Econ. 2020;43:239–273. wileyonlinelibrary.com/journal/twec

|

239

© 2019 John Wiley & Sons Ltd

Received: 7 February 2019

|

Revised: 20 August 2019

|

Accepted: 10 October 2019

DOI: 10.1111/twec.12888

ORIGINAL ARTICLE

The emergence of regional business cycle in

Africa—a reality or myth? A Bayesian dynamic

factor model analysis

Oyeyinka S.Omoshoro‐Jones

|

LumengoBonga‐Bonga

Department of Economics and Econometrics,University of Johannesburg, Auckland Park, South Africa

KEYWORDS

Africa, Bayesian method, business cycle, dynamic factor models, regional integration

1

|

INTRODUCTION

Africa is becoming increasingly integrated into the global economy mainly through trade linkages

(Gurara & Ncube, 2013; Kinfack & Bonga‐Bonga, 2015; Samake & Yang, 2014) with financial link-

ages playing a limited role (Canales‐Kriljenko, Hosseinkouchack, & Meyer‐Cirkel, 2014; Kose &

Riezman, 2001). To a large extent, the impressive economic growth in Africa since the 2000s can be

attributed to strong trade ties underpinned by economic cooperation with both large advanced (i.e., the

United States and European Union) and emerging market (Brazil, Russia, India, China—the BRICs

group) economies viewed as economic powerhouses in the world economy.1 These ratified bilateral

trade agreements have led to remarkable growth in Africa's total trade volume since the 1980s (see

Figure A1 in Appendix B). Conversely, prudent financial management and technology advancement

have also transformed the continent's financial landscape, leading to the emergence of the so‐called

frontier African market, a group of financially integrated African countries attracting sizeable (for-

eign) capital inflows and private investments (Deléchat, Ramirez, Wagh, & Wakeman‐Linn, 2010;

IMF, 2013; Nellor, 2008; The Economist, 2011; UNECA, 2015a).

Although trade remains a key driver of regional integration and economic growth in Africa, intra‐

African trade rose steadily from 6% in 1990 to 12% by 2011 and surging to about 16% in 2014 (AfDB,

2017; Arizala, Bellon, & Macdonald, 2018b), and total intra‐African trade is still low compared with

the total intra‐regional trade of EU of about 68%, 59% of Asia, 37% of North America and 20% of

Latin America (Afreximbank, 2018). In fact, the entire total intra‐Africa trade accounts for about 3%

(of which sub‐Saharan Africa accounts for 2%) of the total world trade. The performance of African

region in this regard compares dismally to the sizeable share of world trade accounted for by advanced

1 The close trade ties between Africa countries (almost all) and their major traditional (i.e., the United States and EU) and new

trading (China) partners is underscored by main economic cooperation, which includes African Growth and Opportunity Act

(AGOA), Everything But Arms (EBA) and Forum on China‐Africa Cooperation (FOCAC) with the United States, EU and

China, respectively.

240

|

OMOSHORO‐JONES aNd BONGa‐BONGa

countries (75%), EU (42%) and Euro area (31%), as shown in Figure A2 (see Appendix B).2 Hence, it

is important to understand the nature and drivers of business cycle co‐movement among African coun-

tries that are required to facilitate stronger economic integration and coordination of economic poli-

cies and align them with business cycles convergence at the regional level (see Arizala, Bellon, &

Macdonald, 2018a; Arizala et al., 2018b; Nzimande & Ngalawa, 2017).

More recently, the institutional willingness to improve the growth of Africa's trade share in the

world economy has led to the creation of an African continental free trade area (AfCFTA) consisting

of 49 (out of 54) African countries—the largest free trade zone since the creation of the World Trade

Organisation in 1995. The primary aim of the AfCFTA is to deepen regional integration by promoting

economic integration and higher intra‐African trade via the removal of tariff and nontariff barriers.

Other objectives of creating the AfCFTA are, among others, to allow the flow of cross‐border labour

and capital mobility; create an alternative large export market; foster regional industrialisation; and

make the region more resilient to negative external (global) shocks including spillovers from large

trading partners, and facilitate business cycle convergence (AfDB, 2017; Afreximbank, 2018; Fofack,

2018; UNCTAD, 2015). Another objective of the AfCFTA is to lay the foundation for the creation of

a continental customs union aimed at promoting regional economic integration in Africa.

Theoretically, the concerted political effort to deepen regional economic integration in Africa can

mitigate the adverse effects of external shocks (or spillovers). In the business cycle literature, trade

is arguably the most important driver of business cycle co‐movement and cross‐country spillovers

(Baxter & Kouparitsas, 2005; Imbs, 2004; Inklaar, Jong‐A‐Pin, & Haan, 2008), albeit the magnitude

of spillovers depends on the strength of bilateral trade ties (Kose & Yi, 2006). In this context, Frankel

and Rose (1998) show that higher bilateral trade allows cross‐country spillover of aggregate demand

shocks which lead to greater business cycle co‐movement as outputs become more correlated.

It is important to note that, in the aftermath of the recent global economic recession and financial

crisis during the period 2007–12, the role of a regional factor on the global business cycle has gained

lots of interest in recent empirical studies mostly due to the fast recovery of many developing countries

compared with advanced countries. This phenomenon may be indicative of the declining role of the

global factor on regional factors—decoupling hypothesis (Kose, Otrok, & Prasad, 2012).3 Subsequent

studies have confirmed the growing importance of regional factors vis‐à‐vis the declining global fac-

tors on the business cycle of developing countries as a result of a significant increase in intra‐regional

trade among these countries (Aastveit,Bjørnland & Thorsrud, 2016; Dong & Wei, 2012; Helbling et

al., 2007; Hirata, Kose, & Otrok, 2013; Kose et al., 2012; Mumtaz, Simonelli & Surico, 2011).

In this context, recent developments in Africa can cause the emergence of a regional (or “an

African”) business cycle, for example the dramatic surge in intra‐regional trade and financial flows

largely driven by regional integration initiatives such as the creation of the continental free trade

area to promote a higher degree of business cycle co‐movement (UNECA, 2015a, 2015b). Regional

business cycles can also emerge as a result of region‐specific shocks driven by the implementation of

similar policies in a region (Degiannakis, Duffy, Filis, & Livada, 2016; Fatás & Mihov, 2008; Drazen,

2000). Likewise, cross‐border spillovers propagated in large economic powerhouses such as Nigeria

2 See, also, UNCTADStats (2018).

3 See Pesce (2017) for a detailed survey of the theoretical and empirical literature on the decoupling and recoupling

(convergence) hypotheses. The decoupling hypothesis argued that the business cycles of emerging market economies have

become more independent of the advanced economies cycles due to a growing regional integration among the former. By

implication, economic performance of the emerging market economies is less reliant on the growth performance of advanced

countries.

|

241

OMOSHORO‐JONES aNd BONGa‐BONGa

or South Africa widely perceived as regional growth drivers can translate into more synchronised

national business cycles at a region level (IMF, 2012; World Bank, 2016).

Therefore, in this paper, we examine the evolution of regional interdependencies across seven African

(A‐7) countries consisting of Egypt, Kenya, Ghana, Nigeria, South Africa, Tanzania and Uganda. These

countries account for about 60% of Africa's aggregate GDP. The choice of these countries is because they

can be potential sources based on the size of their economies (in billions US$; WDI, 2016); their significant

role as regional growth drivers (AfDB, 2017; Afreximbank, 2017, 2018; Fofack, 2018); and their preference

as destinations of private capital inflows (especially in the form of foreign direct investment), official devel-

opment assistance and investment‐linked loans targeting natural resources (IMF, 2009; UNCTAD, 2015;

UNECA, 2015a, 2015b). The other motivation for the choice of these countries is consideration of potential

sources of inward spillovers to neighbouring countries markets (Arizala et al., 2018b; Nellor, 2008).4

Our analytical exercise is more relevant and timely for three reasons. First, the evolution of the struc-

tural composition among the A‐7 countries could have important implications for future regional policy

coordination. For example, the dominance of regional, country‐specific or idiosyncratic factors under-

lying the national business cycle of the A‐7 countries can be a catalyst for the emergence of a common

regional business cycle, as outputs become more correlated. Second, the knowledge on the nature of

the business cycle of these frontier markets can help policymakers to effectively design, coordinate and

implement (future) regional policies. Third, a better understanding of the impact of external (global) or

internal (regional or country‐specific) shocks as well as driving (external and internal) forces behind

cycle fluctuations in these countries can explain the extent to which economic gains or crisis in these

frontier markets would be transmitted in the region through trade and/or financial linkages.

Furthermore, so far, only a handful of empirical studies have examined the nature (or degree) of

business cycle co‐movement between African countries with close trade ties or geographical proximity.

The majority of these studies employ a standard dynamic factor model using annual real GDP data as the

only approximate measure of the national business cycle due to paucity of high‐frequency data in most

African countries. Among these, Kabundi and Loots (2007) find country‐specific factors as the main

drivers of business cycle co‐movement between the 11 countries of the Southern African Development

Community (SADC) regional bloc between 1980 and 2002, except the South African business cycle

which is largely influenced by the global factor (proxied as the G7 countries). Nzimande and Ngalawa

(2016) assess the feasibility of creating a common currency area that allows the use of a single currency

among 16 members of the SADC using data sample period 1988–2014, and find little evidence to

support this regional initiative. This is due to the dominance of both country‐specific and idiosyncratic

factors driving the national business cycles, in most cases. More recently, Claassen, Loots, Kabundi, and

Viviers (2016) find that the business cycles of oil‐exporting and low‐income African countries have de-

coupled from those of the advanced (G‐7) countries due to the significant influence of idiosyncratic fac-

tors on the business cycles of these two groups. Nonetheless, global factor dominates the co‐movement

between the middle‐income African countries and the G‐7 cycles, in the aftermath of the recent global

crisis. Elsewhere, Canova and Ciccarelli (2012) use a factor augmented VAR (FAVAR) model estimated

over the period 1980–2009 to analyse the business cycle co‐movement between few Mediterranean

Basin countries (i.e., Egypt, Tunisia, Algeria and Morocco) and find the co‐existence of a transitory and

reversible convergence/divergence among their national business cycles with a large dominant idiosyn-

cratic factor causing cyclical fluctuations in domestic output, consumption and investment.

4 Most of the African countries (or frontier markets) considered in this paper have been found to be potential conduit for

outward and/or inward transmission of global and regional shocks as well as financial contagion spillovers, thus heightening

the continent's exposure to both external and internal shocks/spillover; for example, see Dabla‐Norris et al. (2015),

Giovannetti and Velucchi (2013), and Canales‐Kriljenko et al. (2013).