Low-income countries’ linkages to BRICS: Are there growth spillovers?

Journal of Asian Economics 30 (2014) 1–14

Contents lists available at ScienceDirect

Journal of Asian Economics

Low-income countries’ linkages to BRICS: Are there growth spillovers?§ Issouf Samake a,*, Yongzheng Yang b a b

Western Hemisphere Department, International Monetary Fund, 700 19th Street, N.W., Washington, DC 20431, USA Asia and Pacific Department, International Monetary Fund, 700 19th Street, N.W., Washington, DC 20431, USA

A R T I C L E I N F O

A B S T R A C T

Article history: Received 30 July 2012 Received in revised form 7 August 2013 Accepted 6 September 2013 Available online 10 October 2013

This paper employs a global vector autoregression (GVAR) model to investigate business cycle transmission from BRICS (Brazil, Russia, India, China, and South Africa) to LICs through trade, FDI, technology, and exchange rates channels. Trade and financial ties between low-income countries (LICs) and BRICS have expanded rapidly in recent years. This gives rise to the potential for growth to spill over from the latter to the former. The estimation results show that there are indeed significant direct spillovers from BRICS to LICs. ß 2013 Elsevier Inc. All rights reserved.

JEL classification: C32 E17 F47 Keywords: Spillovers Low-income countries BRICS Global VAR

1. Introduction Research on business cycle transmission has regained attention in the wake of the recent global financial crisis. The International Monetary Fund (IMF) has, for instance, carried out several studies that examine spillovers from systemically important countries (the United States, European Union, Japan, and China) to the rest of the world.1 These studies almost exclusively focus on business cycle transmission among advanced and major emerging market economies, with limited attention to transmission to low-income countries (LICs), particularly that from major emerging market economies such as BRICS (Brazil, Russia, India, China, and South Africa). This is not surprising given that LIC-BRICS ties have become significant only recently. However, trade and financial relations between LICs and BRICS have grown so rapidly over the past decade that any analysis of LICs’ growth prospects would not be complete without taking into account the impact of the BRICS economies. The emergence of BRICS has brought about a significant redirection of LIC trade and financial ties toward these emerging markets. Bilateral trade between LICs and BRICS has grown exponentially in recent years, making BRICS collectively a trade partner that is comparable to the United States. BRICS FDI and development assistance, though remaining small relative to official development assistance (ODA) from traditional donors, have also grown rapidly and are making a significant impact

§ We thank Ayhan Kose, Montfort Mlachila, Nkunde Mwase, Cathy Pattillo and Nagwa Riad for helpful comment. Sibabrata Das and Ke Wang provided excellent research assistance. Any remaining errors and omissions are our own. * Corresponding author. Tel.: +1 2025699305; fax: +1 2025898983. E-mail addresses: [email protected] (I. Samake), [email protected] (Y. Yang). 1 Recently, Arora and Vamvakidis (2010) show significant spillovers from China to the rest of the world both in short and long-run.

1049-0078/$ – see front matter ß 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.asieco.2013.09.002

I. Samake, Y. Yang / Journal of Asian Economics 30 (2014) 1–14

2 Table 1 BRICS in the Global Economy, 1991–2015.a

1991–1994

2000–2004

2005–2009

2015

(In percent of world total; period average) Population BRICS Other EMEsb United States Euro Area

45.4 23.1 4.8 5.6

44.3 23.3 4.7 5.1

43.5 23.6 4.6 4.9

42.5 23.9 4.5 4.6

GDPc BRICS Other EMEs United States Euro Area

7.1 10.3 26.3 24.9

8.9 10.9 30.6 21.2

13.6 13.4 25.6 22.0

21.2 15.8 21.1 16.9

Exports BRICS Other EMEs United States Euro Area

3.8 13.0 13.3 34.6

6.9 15.8 12.0 30.8

10.6 18.6 9.7 29.0

17.2 18.8 9.3 23.2

Imports BRICS Other EMEs United States Euro Area

3.6 14.4 14.6 34.0

6.5 14.8 17.2 29.4

9.5 17.2 14.1 28.5

16.1 18.1 11.8 22.8

Sources: IMF and Work I Economic Outlook, October 2010. a IMF WEO Projections for 2015. b Emerging market economies excluding BRICS. c At market exchange rates.

in certain key sectors (e.g., infrastructure and resource extraction) of LIC economies. In addition, the rise of BRICS in the global economy could exert significant indirect effects on LIC economies via global goods and financial markets. In particular, rising global demand for commodities as a result of strong economic growth in BRICS and their rapid reserve accumulation could alter the terms of trade and the cost of financing for LICs in the global market. The relatively mild deceleration of LIC economic growth during the global financial crisis points to the potential benefits of their growing ties with BRICS. Most LICs were hit hard by the crisis, but growth often slowed less and recovered faster than anticipated. This milder impact was anticipated by some analysts at the onset of the global financial crisis. They argued that Sub-Saharan Africa, for instance, would be more resilient to the global financial crisis than conventional wisdom would suggest because of the region’s strong trade ties with BRICS, particularly China (Broadman, 2010). To the extent that BRICS’s growth suffered much less from the crisis than that of advanced economies, LICs’ ties with BRICS must have helped them lessen the impact of the global financial crisis, though the extent of this alleviation is unknown. More generally, as long as BRICS’s business cycles are not completely synchronized with those of advanced economies, their spillovers to LICs should help dampen the volatility of LIC growth. Moreover, the emergence of BRICS provides another potential source of sustained external demand and source of financing for low-income countries, raising the prospects for faster growth and poverty reduction in the long term (IMF, 2011). Against this background, this study uses global vector-autoregressive (GVAR) model to examine growth spillovers from BRICS to LICs. To make the analysis more manageable, we focus on the direct channels of spillover, trying to shed light on the extent and relative importance of the impact of technology, FDI, trade, and exchange rate shocks from BRICS on LICs.2 To the best of our knowledge, this is the first attempt to model the LIC-BRICS linkages through multivariate time series analysis using a GVAR model. Our intention is to engender more robust research in this important area of global economic linkages. The rest of the paper is organized as follows. Section II provides, as a background, some stylized facts on the role of BRICS in the global economy and their growing trade and financial ties with LICs. Section III presents the basic setup of the GVAR model estimated in this paper, and Section IV reports the estimation and simulation results. Section V concludes with a brief summary of the key insights and policy implications of this research. 2. BRICS in the World economy and LIC-BRICS linkages: some stylized facts The growing LIC–BRICS relations can be best understood in the context of BRICS’s increasing prominence in the global economy (Table 1). With a combined labor force of more than 1 billion people, BRICS have always had the potential to be key

2 Samake and Yang (2011) explore indirect channels of spillovers from BRICS to LICs through global goods and financial markets given LICs’ high dependence on global demand and world commodity prices.

I. Samake, Y. Yang / Journal of Asian Economics 30 (2014) 1–14

3

players in the global economy. Rapid economic growth in recent decades has enabled BRICS to begin to realize this potential. While the rapidly growing LIC–BRICS ties have benefited from their great economic complementarity – two of the BRICS, China and India, are in great need of natural resources that many LICs are abundantly endowed with and can supply in exchange for competitively priced manufactures – BRICS’s growing weight in the world economy has been the fundamental force driving the LIC–BRICS economic relations. With a large population base and relatively low per capita income, BRICS’s role in the world economy is only likely to increase over time as they narrow their gaps with advanced countries in income levels. The world population in 2009 was estimated at 6.7 billion people. Almost three out of every seven people in the world today live in BRICS. Although the share of BRICS in world population is projected to decline over time, similar to that of the Unites States and the Euro area, it will remain a multiple of that of the United States and Euro area combined and would eventually supports economies that are commensurate to BRICS’s human resource base. Since the early 1990s, BRICS have more than doubled their share in global output. BRICS’s GDP (based on market exchange rates) is now the third largest in the world after the United States and the Euro Area. According to IMF projections, BRICS’s GDP will surpass that of the Euro area before 2015. BRICS exports appear have been the most dynamic in their integration into the world economy. Over the past two decades, BRICS’s share in world exports have nearly tripled, overtaking that of the United States and catching up rapidly with that of the Euro area. The growth of BRICS imports has been less spectacular but still very impressive – nearly doubling their share in world imports over the past two decades, and should catch up with the United States soon. Given BRICS’s strong growth and rapid integration into the global economy, it is not surprising that their trade and financial ties with LICs have bloomed. It must be noted, however, that these ties have also been re-enforced by improvements in macroeconomic management and the business climate in many LICs, while global commodity booms – which have partly resulted from BRICS’s economic growth – have provided a critical linkage between LICs and BRICS. Below are a few stylized facts about LIC-BRICS linkages (Figs. 1–3). FDI inflows from BRICS to LICs appear to be positively correlated with LIC growth. This correlation is stronger for most Asian countries and Sub-Saharan African countries, particularly among resource-rich countries such as Angola, Nigeria, Zambia, and the Republic of Congo. A large part of FDI from BRICS (mostly China) to LICs is concentrated in natural resources and infrastructure. Given that these countries have often come out of conflict only recently, there are great needs for reconstruction, and their economic growth is more likely to be associated with BRICS investment. Similarly, LIC trade with BRICS appears to be positively correlated with LIC growth. The correlation between BRICS-LIC trade and LIC growth is strongest for many Asian LICs, followed by some Sub-Saharan African countries, including Angola, Nigeria, Ethiopia, Uganda, and Tanzania. The correlation between FDI from BRICS and LIC trade with BRICS is insignificant. This is somewhat surprising, as one would expect FDI to provide positive spillover effects on host countries, most prominently through exports, which can often be facilitated by plugging host countries into global value chains. Spillovers can also happen when the entry of foreign firms stimulates competition and promotes efficiency in host industries. The insignificant correlation is most likely to reflect the early stage of BRICS FDI in LICs and the dominance of BRICS economic growth (rather than FDI) in driving trade. BRICS–LIC ties have significant regional dimensions as well as reflect the relative size of individual BRICS. BRICS trade ties with Africa and Asian LICs are far stronger than that with Latin American and Eastern European LICs. Among BRICS, China is by far the largest trade and investment partner for LICs in Africa, Asia, Middle East and North Africa, while Brazil is a dominant partner in Latin America. Russia’s prominence is seen in Eastern Europe and Central Asia.

3. Modeling BRICS–LIC linkages: a global VAR approach A key challenge in the business cycle literature is to come up with a consistent and accurate identification technique for modeling international spillovers of shocks. Common techniques include panel data analysis, single-country VAR models, large-scale macroeconomic models, factor models, dynamic factor models, and global models. It is well known that panel data analysis suffers from one-size-fits-all (single equation) and endogeneity problems. Single-country VAR models for global transmission generally require the estimation of a large number of parameters and hence face a degree of freedom constraint, especially for models involving LICs, which often have limited data with frequent structural breaks. Moreover, panel VAR models have limited ability to control for cross-country differences. Finally, large-scale macroeconomic models require a large number of behavioral equations and parameters (see Barrell et al., 2001). Dynamic factor models have remained until recently the most powerful and widely used econometric tool to analyze business cycles across countries or regions. However, factor models are criticized for being atheoretical and lacking a structural identification scheme. Additionally, even after controlling for ‘‘common’’ factors there are always important residual cross-country interdependencies due to policy and trade spillover effects that remain to be explained. Against this backdrop, we use a global vector autoregressive (GVAR) model to analyze BRICS spillovers to LICs. The GVAR approach is a recent development by Pesaran, Schuermann, and Weiner (2004), Pesaran and Smith (2006), De´es, di Mauro, Pesaran, and Smith (2007a) and De´es, Holly, Pesaran, and Smith (2007b). The GVAR model is a multivariate and multicountry framework that can be used to investigate cross-country and cross-region interdependencies. While it allows minimizing parameter requirements and can cover a large number of geographical areas, the GVAR model typically links individual countries or regions by including foreign-specific fundamentals. Unlike the factor models, the GVAR model introduces

[(Fig._1)TD$IG]

4

I. Samake, Y. Yang / Journal of Asian Economics 30 (2014) 1–14

Fig. 1. LICs and BRICS linkages: 2000–2007 average (in percent)

observed country-specific foreign variables in individual country models to deal with pervasive dependencies in the global economy in a flexible manner. Typically, the GVAR model explicitly specifies international business cycle spillovers with following characteristics: (i) domestic variables are related to corresponding trade-weighted foreign variables to match the international trade pattern of the country under consideration; (ii) non-zero pair-wise correlation in residuals between countries and equations are allowed, to capture a certain amount of dependence from idiosyncratic shocks; (iii) common observed shocks can be introduced; and (iv) country idiosyncratic factors are controlled for. The GVAR model used in this paper is built on the work of PSW (2004). It covers the five BRICS countries (Brazil, Russia, India, China, and South Africa) and 30 LICs (of which 13 are from Africa, 8 from Asia, 6 from Middle-East, and 3 from Latin America and the Caribbean). Among the 30 LICs, there are 6 oil producers and 5 other commodity exporters (see Appendix Table A1). To make the GVAR exercise more manageable, the five BRICS countries are grouped into a single economy in the analysis. Each LIC and BRICS model is linked to the others by including foreign-specific variables related to its international trade patterns. In addition, global variables representing international factors are included in each of the country models and all the country-specific variables and observed global factors (such as oil prices) are treated endogenously. Basically, suppose there are N + 1 LICs and BRICS, indexed by i = 0, 1, . . ., N, where country 0 serves as the numeraire country (which is BRICS for simplicity, but could be any other country). Country-specific variables include real GDP (log of real GDP, PPP-based,

[(Fig._2)TD$IG]

I. Samake, Y. Yang / Journal of Asian Economics 30 (2014) 1–14

5

Fig. 2. Trade with BRICS, by region, 2000–2008 (U.S. dollar, million)

2000 = 100); trade (log of volume of trade with BRICS for each LIC); FDI from BRICS (proxied by time-varying shares of FDI from BRICS to each country multiplied by total FDI received by country);3 and Inflation (log of CPI index, 2000 = 100). The model is estimated over the period 1970–2010 using annual data.4 The time series data for BRICS was constructed by cross-section weighted averages using the average Purchasing Power Parity GDP weights, computed over 2003–2007. The BRICS-specific variables include aggregate real GDP (log of real GDP, PPP-based, 2000 = 100), aggregate demand (log of weighted average import volume, 2000 = 100), and real effective exchange rate (log of weighted average real effective exchange rate index, 2000 = 100). All models include the common global variables: international crude oil price (US$ per barrel), global demand, and U.S. Fed Fund rates. Several variables have been selected to capture the spillovers from BRICS to LICs based on the theoretical and empirical literature. Technology shock is considered to be a key channel of spillover and is proxied here by shocks to BRICS’s GDP per capita. The Ricardian model predicts that comparative advantage arises from technology differences between countries. Technology disturbances affect the marginal product of factors of production, influence investment opportunities within each country/region, and alter trade patterns of final goods. Changes in international specialization in production could also result in higher real aggregate income and welfare. The demand shock is proxied by changes in trade. The Heckscher–Ohlin–Samuelson theory predicts that comparative advantage results from differences in factor endowments across countries. As such, countries/regions export goods that intensively use factors of production with which they are relatively abundantly endowed, and import goods that use

 Total BRIC FDI to LIC at t   Total FDI recieved by LICX it . FDI inflows to LIC X it ¼ Total World FDI to LIC at t Source data are International Monetary Fund, WEO database; OECD database; International Monetary Fund, Direction of trade statistics; International Monetary Fund, World Economic Outlook International Financial Statistics Database. 3 4

[(Fig._3)TD$IG]

I. Samake, Y. Yang / Journal of Asian Economics 30 (2014) 1–14

6

Fig. 3. Trade with BRICS, by type of exporters, 2000–2008 (U.S. dollar, million)

intensively factors that are relatively scarce at home. Trade linkages generated under such frameworks can lead to both demand- and supply-side spillovers across countries, resulting in a higher degree of synchronization of output across countries. On the other hand, if stronger trade linkages facilitate increased specialization of production across countries, and if sector-specific shocks are dominant, then the degree of co-movement of output could fall (see Baxter and Kouparitsas, 2005). Shocks to FDI are justified in the context of standard business cycle literature. This is supported by the fact that LIC–BRICS ties could increase their investment correlations and raise investment growth in LICs because of their potentially higher returns on capital. However, the empirical research has not yet produced conclusive findings on the impact of FDI on growth. While some authors have attributed this situation to methodological problems, others have identified threshold effects. For instance, Kose, Prasad, and Taylor (2009) show that there are certain ‘‘threshold’’ levels of financial and institutional development that an economy needs to attain, over which financial flows, including FDI, could have significant growth effects. Recently, Dabla-Norris, Honda, Lahre`che, and Verdier (2010) find that FDI can significantly contribute to growth if host countries have sound economic fundamentals and macroeconomic stability. The exchange rate is an important channel of economic and financial transmission, and movements in BRICS’s real exchange rates would affect relative prices of tradables versus nontradables in their economies. Thus, real exchange rate changes in BRICS should affect LIC exports, trade patterns, and business synchronization. Formally, the general individual model for the 31 LICs/BRICS is presented in the form of a VAR X*(1,1) model as follows:5  X it ¼ aio þ ai1 t þ fi X it1 þ Li0 Xit þ Li1 Xit1 þ G i0 Dt þ G i1 Dt1 þ uit

5

Given data limitation and the number of parameters to be estimated, this paper limits the lag length to 1.

(1)

I. Samake, Y. Yang / Journal of Asian Economics 30 (2014) 1–14

7

where aio is a Ki  1 vector of fixed intercepts; fi is a Ki  Ki matrix of coefficients associated with lagged domestic variables, Xit1; Gi0 and Gi1 are K i  Ki matrices of coefficients associated with, respectively, contemporaneous and lagged foreign variables, X*it and X*it1; and Dt denotes a mx1 vector of (common) global factor variables. Xit and Dt are assumed to be weakly exogenous, Gi0 and Gi1 are ki  m matrices of coefficients associated with, respectively, contemporaneous and lagged global (weakly) exogenous factor variables, Dt and Dt-1; and uit is a ki  1 vector of country-specific shocks, assumed serially P uncorrelated with zero mean and a non-singular covariance matrix, ii = (sii,1s), where sii,1s = cov(uilt, uist), or more compactly, uit  i.i.d.(0, Sii). Finally, as in PSW (2004), we allow the idiosyncratic shocks, uit, to be correlated across countries/ regions, to a limited degree. That is, for i 6¼ j (X Eðuit u0jt0 Þ ¼ covðuit ; u0jt0 Þ ¼

ij

0

for t ¼ t 0 for t 6¼ t 0

We impose that ki, the number of country-specific variables, is the same across countries. However, this paper recognizes that (i) not all LICs are equally important in the global economy; (ii) country-specific shocks might be crosssectionally correlated due to geographic or contagion effects that are not totally caused by common factors; and (iii) markets might not be equally important or present in LICs countries. In light of these considerations, this paper adopts an approach developed by PSW, which uses country-specific weights, wij, as opposed to common weights, wj, in constructing cross-country averages (Appendix Table A2). More specifically, PSW uses the following formula to weigh country-specific foreign variables:

Xit ¼

N M X X wi jt X jt þ wi jt X jt j¼0

j¼0

The weight wijt (for j = 0, . . ., N) captures the relative importance of country j for the ith country in the world economy. For instance, if i = 1, w1jt captures the relative importance of countries j = 1, 2, . . ., N for country i = 1. This formulation allows us to better control for shocks from BRICS and other observed factors that impact on LICs’ endogenous variables. More importantly, it also helps assign appropriate weights to various foreign shocks. At the same time, these time-varying weights control for any breaks induced by the fast growing BRICS–LIC trade and financial ties. Given the size of the GVAR, the model cannot be estimated using standard VAR techniques because it would require more estimated parameters than available observations. The assumption of weak exogeneity of country-specific foreign variables allows for constructing country-specific variables using time-varying trade weights. This greatly reduces the number of unrestricted parameters to be estimated. The corresponding vector error correction model of Eq. (1) that serves as the basis for the generalized impulse response function (GIRF) is:

DX it ¼ cio  ai b0i ½Z it1  g i ðt  1Þ þ Li0 DXitj þ cio DZ it1 þ G i0 Dt þ G i1 Dt1 þ uit

(2)

where Z it ¼ ðX 0it ; Xit0 Þ, is a k  ri matrix of rank ri, and bi is a ðk þ ki Þ  r i matrix of rank ri. The estimation is undertaken by taking into account the integration properties of the series; as a consequence it is possible to find cointegrating relationships among domestic and foreign variables. Even if the GVAR model is atheoretical in spirit, it can incorporate economic-based structural restrictions. After the estimation of the countrymodels, their corresponding estimates are connected through link matrices and then stacked together in order to build the GVAR model. Having tested that the majority of our series are I(1), the cointegrating VARX* country models are estimated. More specifically, each cointegrating VARX* model is estimated subject to the reduced rank restriction (Johansen, 1992, 2005). Therefore, the cointegration rank is derived by employing both the trace and maximum eigenvalue statistics. Trace statistics have been chosen for the rank selection, given that they yield better small sample results compared to the maximum eigenvalue statistics. The asymptotic distribution of the trace and maximum eigenvalue statistics depend on whether the intercept and/or the coefficients on the deterministic trend are restricted or unrestricted. In the current application, the reduced rank regressions are used in the case of unrestricted intercepts and restricted trends (e.g. case IV in Pesaran, Shin and Smith, 2000). In particular, rank tests are conducted at the 95% significance level.6 The reduced rank procedure neither indicate exactly which cointegrating relationships are found (i.e. which of the variables are linked together), nor does it identify them. In order to exactly identify the cointegrating matrix, r contemporaneous restrictions on each cointegrating vector are imposed. These are implemented by normalizing each cointegrating vector, thus following a purely atheoretical approach. Subsequently, in cases where cointegration is found, each country VARX* model is estimated under its vector error correction (VECMX*) form.

6

Rank tests statistics are available upon request.

[(Fig._4)TD$IG]

I. Samake, Y. Yang / Journal of Asian Economics 30 (2014) 1–14

8

Shock to Technology from BRICS ± 2 S.E.

Shock to Trade from BRICS ± 2 S.E.

.3

1.6

1.2

.2

0.8 .1 0.4 .0

0.0

-.1

-0.4 11

12

13

14

15

16

17

18

19

20

11

12

Shock to Exchange Rate from BRICS ± 2 S.E.

13

14

15

16

17

18

19

20

19

20

Shock to FDI from BRICS ± 2 S.E.

.7

1.0 0.8

.6

0.6 .5 0.4 .4 0.2 .3

0.0

.2

-0.2 11

12

13

14

15

16

17

18

19

20

11

12

13

14

15

16

17

18

Bootstrap Mean Estimates and 90 percent Bootstrap Error Bound, with 1000 replications. Fig. 4. Africa LICs: generalized impulse response of output, 2011–2020 (one standard error shock, responses in percent)

4. Empirical results We analyze the dynamic properties of the GVAR model by means of the GIRF proposed in Koop and Pesaran (1996) and further developed in Pesaran and Shin (1996).7 The analysis focuses on the effects of shocks to BRICS (foreign) variables on the LICs by region/economic group (Africa, Asia, Middle East, Latin America and the Caribbean, and oil exporters).8 In particular, we investigate the impact of one unit (a positive standard error) of shocks to BRICS technology, trade, real exchange rate, and FDI. Each GIRF (Figs. 4–9) shows the dynamic response (up to 10 years from 2011) of each domestic variable of an LIC group to a unit shock originated from BRICS. The solid lines in figures show the bootstrap mean estimates of GIRFs and dashed lines denote 90 percent confidence intervals computed using bootstrap with 1000 replications. The GVAR includes 123 endogenous variables (4 times 30 LICs plus 3 times BRICS) and its maximum lag order is equal to 2. Hence, the companion VAR(1) form has 246 eigenvalues. Stability tests show that their moduli are all less than or equal to unity. The model stability is also evidenced by the fact that the GIRFs settle down generally quickly. Nonetheless, our analysis will mostly focus on the medium term following shocks during which estimates appear to be more efficient than the outer years. Furthermore, the moduli of 72 eigenvalues are equal to unity, consistent with evidence of unit roots of the model; hence, simulated shocks will tend to exert permanent effects on endogenous variables in the global system. Next, of the remaining 174 eigenvalues falling inside the unit circle, 132 are complex, suggesting that GIRFs will display cyclical behavior. Finally, for the three largest eigenvalues among those that have moduli less than unity, we expect to observe convergence toward a steady-state equilibrium. The generally good significant level of estimates is associated with annual data frequency, which tends to produce lower volatility of estimates than monthly frequency data.

7 Given the forward-looking nature of the generalized impulse response analysis, we assume BRICS will keep the current trade momentum and hence the weights of the most recent five years are used. Analysis using fixed trade weights of the last ten years (2000–09) shows some sensitivity of FDI results, but overall results remain generally unchanged. Similar results generally hold when the average weights of 1970–2009 are used. The results differ quite significantly when the average weights of 1970–1979 are used. Analysis using time-varying weights has been criticized on the grounds that changes in trade weights tend to be counteracted by the co movement of macroeconomic variables so that country-specific variables computed using fixed and variable trade weights are often very close (De´es et al., 2007a,b). 8 The GVAR could also directly control for the effects of business cycles in advanced countries on LICs, but at the cost of a reduced degree of freedom and hence less efficient estimates. In the current setup, the ‘‘other’’ foreign variables (excluding shock emanating from BRICs) account for the effects of the rest of the World (including advanced countries).

[(Fig._5)TD$IG]

I. Samake, Y. Yang / Journal of Asian Economics 30 (2014) 1–14

Shock to Technology from BRICS ± 2 S.E.

9

Shock to Trade from BRICS ± 2 S.E.

1.4

2.0

1.2

1.6

1.0 1.2 0.8 0.8 0.6 0.4

0.4 0.2

0.0 11

12

13

14

15

16

17

18

19

20

11

12

13

Shock to Exchange Rate from BRICS ± 2 S.E.

14

15

16

17

18

19

20

19

20

Shock to FDI from BRICS ± 2 S.E.

1.0

.5 .4

0.8 .3 0.6

.2

0.4

.1 .0

0.2 -.1 0.0

-.2 11

12

13

14

15

16

17

18

19

20

11

12

13

14

15

16

17

18

Bootstrap Mean Estimates and 90 percent Bootstrap Error Bound, with 1000 replications.

Fig. 5. Asia LICs: generalized impulse response of output, 2011–2020 (one standard error shock, responses in percent)

[(Fig._6)TD$IG] Shock to Technology from BRICS ± 2 S.E.

Shock to Trade from BRICS ± 2 S.E.

2.0

1.6

1.5

1.2 0.8

1.0

0.4 0.5 0.0 0.0

-0.4

-0.5

-0.8

-1.0

-1.2 11

12

13

14

15

16

17

18

19

20

11

12

13

14

15

16

17

18

19

20

19

20

Shock to FDI from BRICS ± 2 S.E.

Shock to Exchange Rate from BRICS ± 2 S.E. 1.6

1.0 0.8

1.2

0.6 0.8 0.4 0.4 0.2 0.0

0.0

-0.4

-0.2 11

12

13

14

15

16

17

18

19

20

11

12

13

14

15

16

17

18

Bootstrap Mean Estimates and 90 percent Bootstrap Error Bound, with 1000 replications.

Fig. 6. Middle east LICs: generalized impulse response of output, 2011–2020 (one standard error shock, responses in percent)

[(Fig._7)TD$IG]

I. Samake, Y. Yang / Journal of Asian Economics 30 (2014) 1–14

10

Shock to Technology from BRICS ± 2 S.E.

Shock to Trade from BRICS ± 2 S.E.

1.00

.8

0.75

.6

0.50 .4 0.25 .2 0.00 .0

-0.25 -0.50

-.2 11

12

13

14

15

16

17

18

19

20

11

12

Shock to Exchange Rate from BRICS ± 2 S.E.

13

14

15

16

17

18

19

20

19

20

Shock to FDI from BRICS ± 2 S.E.

.5

.3

.4 .2 .3 .2

.1

.1

.0

.0 -.1 -.1 -.2

-.2 11

12

13

14

15

16

17

18

19

20

11

12

13

14

15

16

17

18

Bootstrap Mean Estimates and 90 percent Bootstrap Error Bound, with 1000 replications.

Fig. 7. Latin America LICs: generalized impulse response of output, 2011–2020 (one standard error shock, responses in percent)

[(Fig._8)TD$IG] Shock to Technology from BRICS ± 2 S.E.

Shock to Trade from BRICS ± 2 S.E.

1.0

1.8

0.8

1.6

0.6

1.4

0.4

1.2

0.2

1.0

0.0

0.8

-0.2

0.6 11

12

13

14

15

16

17

18

19

20

11

12

Shock to Exchange Rate from BRICS ± 2 S.E.

13

14

15

16

17

18

19

20

19

20

Shock to FDI from BRICS ± 2 S.E.

2.0

1.0 0.8

1.6 0.6 1.2

0.4

0.8

0.2 0.0

0.4 -0.2 0.0

-0.4 11

12

13

14

15

16

17

18

19

20

11

12

13

14

15

16

17

18

Bootstrap Mean Estimates and 90 percent Bootstrap Error Bound, with 1000 replications.

Fig. 8. Low-income oil exporters: generalized impulse response of output, 2011–2020 (one standard error shock, responses in percent)

[(Fig._9)TD$IG]

I. Samake, Y. Yang / Journal of Asian Economics 30 (2014) 1–14

Shock to Technology from BRICS ± 2 S.E.

11

Shock to Trade from BRICS ± 2 S.E.

.8

1.6

.6

1.2

.4 0.8 .2 0.4

.0

-.2

0.0 11

12

13

14

15

16

17

18

19

20

11

12

Shock to Exchange Rate from BRICS ± 2 S.E.

13

14

15

16

17

18

19

20

19

20

Shock to FDI from BRICS± 2 S.E.

1.0

1.0 0.8

0.8

0.6 0.6 0.4 0.4 0.2 0.2

0.0

0.0

-0.2 11

12

13

14

15

16

17

18

19

20

11

12

13

14

15

16

17

18

Bootstrap Mean Estimates and 90 percent Bootstrap Error Bound, with 1000 replications.

Fig. 9. Other LICs: generalized impulse response of output, 2011–2020 (one standard error shock, responses in percent)

The GIRFs show that spillovers from BRICS generally manifest in significant improvements in LIC output over time, but there are considerable variations in the impact across different channels of transmission as well as across different LIC regions (Figs. 4–9). Among the four identified channels of spillovers, trade shocks dominate.9 The trade channel accounts for around 60 percent of the impact of BRICS on LIC growth – it is the most significant and persistent channel of transmission for all regions. In general, trade shocks lead to positive short- and long-run growth effects except in the European and Middle East LICs, where the short-run response is negative, possibly due to competition in third markets for imports (e.g., agricultural products). In terms of magnitude of the impact across the regions, the greatest responses are seen in LICs in Asia and Middle East, consistent with the strong trade links between BRICS and these countries. The response in African LICs is also quite strong, reflecting growing trade ties that these countries have forged with BRICS in recent years. Furthermore, the overall impact on oil exporters is considerably larger, almost twice as large as on non-oil exporters. This is hardly surprising given the strong complementarity between major BRICS (China and India) and oil exporters, discussed earlier on. A positive shock to the real exchange rate (appreciation) in BRICS is associated with positive growth effects on LICs. Given that LIC-BRICS ties are dominated by trade in magnitude, exchange rate movements are likely to exert their impact through trade. As such, like trade shocks, these effects appear to be positive and particularly strong in oil exporters among LICs.10 Surprisingly, however, real exchange rate appreciation is associated with reduced growth in LICs in Latin America and the Caribbean in the short run. This could reflect the fact that the selected Latin America and the Caribbean countries have limited exports destined for BRICS (except Brazil) and a real appreciation in BRICS leads to higher import costs, worsening the terms of trade and growth. Results show that the impact of technological shocks is generally significant and leads to higher long-run growth in LICs. Consistent with some recent studies of the relationship between BRICS (China in particular) and LICs, short-term responses in LICs are sometimes negative, but they tend to shift to a significant positive impact in the longer horizon. As observed by Ajakaiye and Kaplinsky (2009), such initial negative impact could be related to BRICS’s (China in their case) competition in third markets for exports and imports.11 Asian LICs stand out in terms of benefiting from technology shocks from BRICS. This probably reflects the closer integration of Asian LICs into global manufacturing supply chains, in which BRICS (particularly

9 The shocks are a one-standard error increase in the volume of total BRICS imports (in logs) from LICs. The same magnitude applies to other shocks discussed below. 10 Statistically significant but the magnitude is small compared with trade shocks. 11 Recently, Arora and Vamvakidis (2010) discussed that China’s exports could have negative direct effect on net exporter countries.

I. Samake, Y. Yang / Journal of Asian Economics 30 (2014) 1–14

12

China) play a critical role. LICs in Middle East also benefit significantly from technology shocks, as do non-oil commodity exporters. Finally, the FDI channel also matters, but compared with the other three channels of spillovers, its impact on LIC growth is more modest even in the long run. While the empirical evidence on the growth benefits of FDI in general is inconclusive, this finding may reflect the relatively small volumes of BRICS’s FDI in LICs – the ‘‘threshold’’ effects are yet to manifest (Kose et al., 2009). Despite the recent surge in BRICS’s FDI to LICs, it remains relatively small compared to total FDI inflows to LICs. Nevertheless, the impact of BRICS FDI could be much larger in individual LICs that have received larger volumes of BRICS FDI. 5. Concluding remarks We have estimated a GVAR model to assess growth spillovers from BRICS to LICs. Using GIRFs, we were able to show the dynamic responses of LIC output to various shocks originating from BRICS, including trade, FDI, technology, and exchange rate. This exercise demonstrates the usefulness of the GVAR model in studying the LIC–BRICS economic linkages, which have been transformed in the past decade as a result of rapid economic growth in BRICS and improved economic performance in LICs. Our GVAR results show that direct spillovers from BRICS to LICs are significant and persistent, exerting considerable impact on LIC growth. Bilateral trade is the most powerful channel of transmission, but exchange rate movements, productivity innovations in BRICS, and FDI flows from BRICS also matter, albeit more modestly. The spillovers spread to all LIC regions, with commodity-exporting countries and those in Asia experiencing the strongest impact. The response in African countries is also quite strong, reflecting the rapid expansion of their trade with BRICS in recent years, particularly in commodities. Looking forward, increasing LIC–BRICS trade and financial ties will only strengthen their business cycle synchronization over time. As long as BRICS business cycles are not fully synchronized with those of advanced countries, these growing ties should help dampen growth volatility in LICs. This also means that LICs will have to pay increased attention to macroeconomic developments in BRICS in assessing their macroeconomic policy stance. Moreover, continued expansion of BRICS economies could alter LICs’ growth potential in the long run, serving as new drivers of long-term growth. Some caveats are worth noting. The estimated spillovers represent broad orders of magnitude rather than definitive estimates and should be treated with caution. Limited lengths of time series and frequent structural breaks in LIC and BRICS economies reduce the robustness of the estimates. Furthermore, neither BRICS nor LICs are homogenous entities. Bilateral relations thus often vary greatly from country to country, and any assessment of bilateral spillovers and policy implications at the country level should therefore be based on a close examination of these relations. Appendix A See Tables A1 and A2. Table A1 List of LICs and selected economic indicators.a Country

Oil exporters

Angola Burkina Faso Cameroon Congo, Republic of Cote D’lvoire Ethiopia Ghana Kenya Nigeria Senegal Tanzania Uganda Zambia Bangladesh Cambodia Mongolia Myanmar Nepal Pakistan Sri Lanka Vanuatu Vietnam Yemen, Republic of Azerbaijan Kyrgyz Republic

1

Commodity exporters 1

1

1

1

1

1 1

FDI inflows from BRICSb

Trade with BRICSc

Real GDP growth

1.5 0.4 0.5 1.2 0.4 0.8 0.6 0.1 1.2 0.1 0.7 0.7 1.4 0.2 1.1 1.6 0.5 0.0 0.4 0.3 1.4 1.0 0.4 2.9 0.9

22.4 9.3 6.4 20.8 5.6 17.2 13.5 10.0 16.0 10.2 15.0 9.1 5.3 16.9 7.4 58.4 24.7 57.6 9.6 16.3 7.7 14.4 29.3 14.0 33.8

11.8 5.2 3.6 4.1 -0.3 7.1 5.2 4.0 9.3 4.2 6.8 7.3 5.0 5.8 9.6 6.6 12.9 3.8 5.1 5.1 2.8 7.6 4.2 15.0 4.5

I. Samake, Y. Yang / Journal of Asian Economics 30 (2014) 1–14

13

Table A1 (Continued ) Country

Oil exporters

Sudand Uzbekistan Bolivia Papua New Guinea St. Lucia

1

Commodity exporters 1 1

FDI inflows from BRICSb

Trade with BRICSc

1.5 0.2 0.6 0.2 3.5

41.7 30.6 30.5 6.0 18.4

Real GDP growth 7.5 6.0 3.4 1.8 1.9

Source: IMF, Direction of Trade Statistics, 1960–2009. a Unweighted average in 2000–2007. b Inward FDI flows to GDP ratio (in percent). c LIC’s trade (export plus import values) with BRICS to its total trade in percent. d Sudan is assumed to belong to the Middle East, Noth Africa, and Eastern Europ country group.

Table A2 LIC, BRICS, and Rest of the World (RoW) average trade weights, 2003–2007.a,b,c Total LICs (in the sample)

Total LICs

BRICS

Total BRICS

RoW

Sum

(a)

Brazil (1)

Russia +(2)

India +(3)

China +(4)

S.A. +(5)

=(b)

(c)

d=a+b+c

Africa Angola Burkina Cameroon Congo, Rep CIV Ethiopia Ghana Kenya Nigeria Senegal Sudan Tanzania Uganda Zambia

0.024 0.010 0.005 0.027 0.002 0.014 0.010 0.005 0.004 0.004 0.045 0.009 0.005 0.004

0.032 0.006 0.013 0.023 0.008 0.010 0.026 0.005 0.062 0.024 0.006 0.002 0.002 0.001

0.001 0.004 0.001 0.000 0.011 0.008 0.005 0.007 0.002 0.001 0.001 0.009 0.006 0.000

0.011 0.021 0.010 0.020 0.018 0.060 0.049 0.081 0.051 0.050 0.035 0.078 0.059 0.022

0.244 0.095 0.045 0.273 0.024 0.135 0.095 0.049 0.041 0.036 0.446 0.093 0.047 0.044

0.001 0.004 0.009 0.000 0.007 0.007 0.003 0.016 0.044 0.001 0.001 0.017 0.002 0.020

0.290 0.130 0.078 0.316 0.067 0.221 0.177 0.157 0.200 0.113 0.489 0.198 0.116 0.087

0.686 0.861 0.917 0.656 0.930 0.766 0.813 0.838 0.795 0.884 0.467 0.793 0.879 0.909

1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000

Asia Bangladesh Cambodia Mongolia Myanmar Nepal Pakistan Srilanka Vanuatu Vietnam

0.008 0.008 0.043 0.019 0.005 0.009 0.004 0.003 0.013

0.005 0.000 0.001 0.000 0.002 0.004 0.002 0.000 0.002

0.007 0.002 0.201 0.002 0.002 0.008 0.011 0.000 0.012

0.090 0.003 0.001 0.084 0.589 0.027 0.147 0.032 0.012

0.084 0.085 0.428 0.192 0.054 0.090 0.042 0.032 0.132

0.004 0.001 0.001 0.000 0.001 0.196 0.001 0.049 0.001

0.190 0.091 0.632 0.278 0.649 0.326 0.204 0.112 0.159

0.801 0.901 0.325 0.702 0.346 0.665 0.792 0.884 0.828

1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000

Middle-East Azerbaijan Yemen Kyrgyz Uzbekistan

0.002 0.019 0.010 0.009

0.007 0.015 0.002 0.001

0.121 0.008 0.301 0.247

0.010 0.122 0.004 0.006

0.024 0.191 0.098 0.093

0.041 0.004 0.001 0.012

0.203 0.341 0.406 0.359

0.795 0.640 0.584 0.632

1.000 1.000 1.000 1.000

Latin America Bolivia Papua N.G. St Lucia

0.003 0.005 0.001

0.341 0.001 0.251

0.001 0.001 0.000

0.002 0.016 0.001

0.027 0.054 0.011

0.004 0.000 0.005

0.375 0.072 0.268

0.623 0.923 0.731

1.000 1.000 1.000

Source: IMF, Direction of Trade Statistics, 1960–2009. a Trade weights are computed as shares of exports and imports displayed in rows by region and by country in the sample. b 2008 and 2009 are excluded from the trade weights because of relative large volatility in trade dynamics in these periods due to the 2008 fuel and food price shocks and the 2009 global financia I crisis. c The displayed weight matrix is just indicative, the estimation and simulation used time-varying weights given the fast growing I trade patterns of BRICS–LICs.

References Ajakaiye, O., & Kaplinsky, R. (2009). China in Africa: A Relationship in Transition. European Journal of Development Research (2009), 21, 479–484. Arora, B. V., & Vamvakidis, A. (2010). China’s Economic Growth: International Spillovers, IMF Working Paper No. 10/165. Washington, DC: International Monetary Fund. Barrell, R., Dury, K., & Holland, D. (2001). Macro-models and the medium term – The NIESR experience with NiGEM. National Institute of Economic and Social Research, Preliminary http://ecomod.net/conferences/ecomod2001/papers_web/Barrell_Bruss4.PDF.

14

I. Samake, Y. Yang / Journal of Asian Economics 30 (2014) 1–14

Baxter, M., & Kouparitsas, M. A. (2005). Determinants of business cycle comovement: a robust analysis. Journal of Monetary Economics, 52(1), 113–157. Broadman, G. H. (2010). Economic Drivers of China’s Foreign Policy Toward Sub-Saharan Africa. Dabla-Norris, E., Honda, J., Lahre`che-Re´vil, A., & Verdier, G. (2010). FDI Flows to Low-Income Countries: Global Drivers and Growth Implications, IMF Working Paper No. 10/132. Washington, DC: International Monetary Fund. De´es, S., di Mauro, F., Pesaran, M. H., & Smith, L. V. (2007). Exploring the international linkages of the Euro area: a Global VAR analysis. Journal of Applied Econometrics, 22, 1–38. De´es, S., Holly, S., Pesaran, M. H., & Smith, L. V. (2007). Long-run Macroeconomic Relations in the Global Economy, European Central Bank, Working Paper, No. 750 May 2007. International Monetary Fund. (2011). New Growth Drivers for Low-Income Countries: The Role of BRICs Available via the internet: http://www.imf.org/external/pp/ longres.aspx?id=4534. Johansen, S. (1992). Cointegration in partial systems and the efficiency of single-equation analysis. Journal of Econometrics, 52(3), 389–402. Johansen, J. (2005). Interpretation of cointegrating coefficients in the cointegrated vector autoregressive model. Oxford Bulletin of Economics and Statistics, 67(1), 93–104. Koop, G., & Pesaran, M. H. (1996). Impulse response analysis in nonlinear multivariate models. Journal of Econometrics, 74(1), 119–147. Kose, M. A., Prasad, E. S., & Taylor, A. D. (2009). Thresholds in the Process of International Financial Integration, NBER Working Paper Series, Working Paper 14916 http:// www.nber.org/papers/w14916. Pesaran, M. H., Schuermann, T., & Weiner, S. M. (2004). Modeling regional interdependencies using a global error-correcting macroeconometric model. Journal of Business and Economics Statistics, 22, 129–162. Pesaran, M. H., & Shin, Y. (1996). Cointegration and speed of convergence to equilibrium. Journal of Econometrics, 71(1/2), 117–143. Pesaran, M. H., Shin, Y., & Smith, R. J. (2000). Structural analysis of vector error correction models with exogenous I(1) variables. Journal of Econometrics, 97(2), 293– 343. Pesaran, M. H., & Smith, R. P. (2006). Macroeconomic Modeling with a Global Perspective pp. 24–49). (vol. 74(1)Manchester School, University of Manchester. Samake, I., & Yang, Y. Z. (2011). Low-Income Countries’ BRIC Linkage: Are There Growth Spillovers?, IMF Working Paper WP/11/267 Washington, DC: International Monetary Fund.