Trade, Remittances and Economic Growth in Nigeria: Any Causal Relationship? (original) (raw)

Trade, Remittances and Economic Growth in Nigeria: Any Causal Relationship?

Ebenezer Adesoji Olubiyi*

Abstract

This study examined the causal relationships among GDP, export, imports and remittances. The study, among others, investigated the validity of export-led and remittances-led growth hypotheses. Specifically, the study investigated the causal relationship between remittances and GDP, remittances and export and remittances and imports. Employing a VECM Granger Causality for data spanning between 1980 and 2012, imports and remittances significantly Granger-caused GDP in the short run. Also, there were reverse causalities running from GDP to export and imports. This implies that export-led growth hypothesis holds in Nigeria. Furthermore, there was a unidirectional causation running from remittances to GDP, implying that remittances matter for economic growth. But since the effect was more from the demand side, it could lead to inflationary pressure. The policy recommendation is that the authorities should intensify efforts on the export base of the economy. The monetary authorities should implement necessary policy to cool the pressure arising from conspicuous spending of remittances.

1. Introduction

The causal relationship among trade, remittances and economic growth is yet to receive unanimous agreement among researchers and policymakers. Until recently, researchers focused on the possible causal relationship between trade and growth. On the one hand, increase in exports leads to economic growth because export growth promotes specialization in the exports sector, boosts workers’ productivity, generates economies of scale and, hence, increase in ouutput (Lal and Rajapatirana, 1987). On the other hand, increase in economic growth facilitated by improved technology enhances trade (Lancaster, 1980 and Krugman, 1984).

Observably if the growth path of trade and GDP are determined by other unrelated factors, the potential for a causal relationship may be jettisoned (Pack, 1988). Also, limited access to foreign markets can thwart the potency of export-led growth (Adelman, 1984 and Buffie, 1992). However, if a country is input import dependent, then increased imports will expedite export supply response and economic growth. Furthermore, increase in output can lead to more import demand. Thus, growth, exports and imports appear to cause each other.

Recently, remittances have been identified as important international capital flows that are expected to boost output because of their ability to mitigate credit constraints, thereby increasing investment and consuption (Krugman, 1984 and Cooray, 2012). Observably, remittances may be a fungible resource or cause reverse capital flow, thereby wilting the potential growth effect (Serieux, 2010). Also, if imported consumption goods account for the large proportion of remittance spending and if they are not fully sterilized, it could lead to inflation (Vacaflores et al., 2012). Rising inflation weakens purchasing power and dwarfs economic growth. Thus, remittances cause both import and economic growth. Also, since remittances are denominated in foreign currency, they could generate Dutch Disease effects through currency appreciation with the resultant effect of a decrease in net export (Rodrik, 2007). This in turn, causes trade deficit and reduces economic growth. Altenatively, the monetary authorities could prevent such appreciation but at the expense of rising inflation. 1{ }^{1} This implies that not only trade, remittances and growth cause each other, but also that the direction of causation is unclear.

For more than a decade now, Nigeria’s GDP, trade and workers’ remittances have been growing consistently. Available evidence shows that in the early 1980s, trade and growth performance were not encouraging as degree of openness hovered

[1]

1. *Department of Economics, Lead City University, Ibadan Oyo State; Research Fellow, Trade Policy Research and Training Program, University of Ibadan; e-mail: biyimclincon@yahoo.co.uk ↩︎

between 31 per cent and 54 per cent while GDP growth was declining at an annual rate of -3.9 per cent. In the mid-1980s, degree of openness was within the range of 31 and 59 per cent. From 1992, degree of openness and growth rate have been on the increase… On the side of remittances, the inflow was less than 2525\25 25 million in 1980 and accounted for less than 1 per cent of GDP. In 2013, it has increased to 2121\21 21 billion and accounted for 7 per cent of GDP (World Bank, 2013). Although there is vast empirical evidence on the growth-trade nexus (see Egwaikhide, 1997; and Bankole et al., 1999) the omission of remittances in their models makes the result incomplete. Also, there is a plethora of evidence on the growth effects of remittances. The recent studies in this regard include Anyanwu and Erhijakpor (2010) and Cooray (2012). None of the studies considers the causality of trade and remittances. Therefore, the present study seeks to fill this lacuna. Section 2 discusses stylized facts about GDP growth, exports, imports and remittances, Section 3 reviews literature, Sections 4 and 5 discuss the methodology and results respectively while Section 6 concludes.

2. Trade, GDP and Remittances in Nigeria

The World Bank trade policy review of 2011, among others, documented Nigerian trade policy instruments covering both imports and exports. Tariffs, duty exception, concessions and other taxes such as port development levy, ECOWAS community levy and VAT on domestic and imported goods are prominently used to modify importation. Export expansion grant (EEG), Pioneer Status Scheme (PSS), the Nigerian Export-Import Bank (NEXIM) and export processing zones (EPZs) are some of the policy instruments employed to encourage exports.

Although these policies helped to improve trade and growth, the behavior over time is unstable. In Table 1, oil exports rose from 48.36 billion naira in the early 1980s (1980-84) to 131.25 billion naira in the late 1980s (1985-89) and grew at an annual average of 36.6 per cent.

Between 1995 and 1999, oil export experienced a dramatic increase recording an average annual growth rate of 76.3 per cent during the period. In the 2000-2004 period, the country realized 12.89 trillion naira from oil export even though the average annual growth rate was not as high as the earlier period. Also from 2005 to 2009, oil export rose to 40.42 trillion naira while the average annual growth rate was 11.7 per cent. This implies that oil exports were increasing at a decreasing rate from 1995 to 2009. In the case of non-oil export, the country experienced growth decay to the tune of average annual percentage of 19.9. But in the succeeding periods, it rose dramatically to 8.9 billion naira and grew at an average of 49.6 per cent. From 1990 to 1995, non-oil export recorded 22.5 billion naira but grew at an annual average of 11.9 per cent. This implies that non-oil export suffered during this period. In the succeeding period when authorities embarked on trade liberalization, non-oil export rose to 5.3 billion naira with the annual growth being 52 per cent. However, the economy experienced export slowdown in 2000-2004 and 2005-2009 periods, in spite of 129.1 billion naira realized from non-oil export.

Unlike the case of exports, non-oil imports dominated total imports. Between 1980 and 1984 the amount of oil import outlay was 1.1 billion naira and rose dramatically to 12.6 billion naira in the 1985-89 period, constituting an average growth rate of 6.1 per cent compared to an annual average growth rate of 4.1 per cent in the earlier period.

Table 1: Pattern and growth rate of trade

Source: Computed using CBN Statistical Bulletin, 2013.

Oil import recorded a downward annual average growth rate as it fell from 56.1 per cent in the 1985-89 period to 44.1 per cent in the 1990-94 period and later fell to 33 per cent and 8.1 per cent in the 1995-99 and 2000-2004 periods respectively. From 2005 henceforth, oil import has been rising both in size and in growth. The highest average growth rate of non-oil import was experienced in the early 1980s. In the period that followed (1984-89), Nigeria did not import non-oil products so much and this informed the decline in the annual average growth from 68 per cent in the 1980-84 period to 26.7 per cent in the 1984-89 period. In the subsequent period, demand for non-oil import rose slightly but fell considerably in the early 2000s only to rise in the latest years of 2000s.

The pattern of trade shows an important behavioural characteristic that is similar to the dynamic of GDP. In the early 1980s, Nigeria experienced growth decline in oil and non-oil exports, with the non-oil being hard hit. Oil and non-oil imports experienced positive growth with non-oil import taking the lead. This situation partly accounted for the deteriorating current account balance experienced during this period. In the middle 1980s, non-oil export grew faster than non-oil import while oil import grew faster than oil export. However, the growth rate of non-oil export more than offset the growth rate of non-oil import while the growth rate of oil export also increased, thereby compensating for any loss in the current account balance. This suggests that trade-led growth policy appears to be effective in Nigeria.

This suggestion was supported by the declining GDP which stood at an average of 3.9 per cent between 1980 and 1984 while it grew at an average of 5.7 per cent in the mid-1980s (Table 2). In the late 1990s, when the growth rate of export fell, real GDP growth rate also plummeted at the rate of 2.5 per cent annually. When the economy improved and the trade liberalization was able to work relatively well, GDP grew at the rate of 6.2 per cent in the 2000-2004 period. However, the growth rate of both exports and imports declined during the period.

The growth of remittances also follows a similar pattern: declining in the 1980s and increasing later. However, unlike other variables, the slowdown of remittance inflows entered the third period (1990-94) but with a slight improvement. From the fourth period (1995-99), average growth has been positive. The highest annual average growth rate was experienced in 2005-2009 while it plummeted in 2010 and 2011. The reason for growth decay in the last two years was not unconnected with the aftermath of the financial crisis in the advanced countries,

The stylized fact shows that the dynamics of GDP, trade and remittances exhibit a pattern that suggests strong association among them. GDP was doing well when exports were relatively increasing and imports were relatively decreasing. The nature of trade dynamics appears to be driven by the trade policy situation, improving when the trade policy was favourable and decreasing during sanctions. Observably, when the growth rate of remittances was on the increase, GDP growth was falling and when it was falling, GDP was rising (Figure 1). This pattern of relationship could be informed either by the moral hazard effect, in which case the continuous increase of remittances leads to a low participation rate thereby reducing output orthe counter cyclicality effect, in which case, improved GDP growth leads to high income, and hence, low remittances.

3. Literature Review

There is a plethora of empirical evidence on the export-led growth hypothesis. Some relatively recent ones include Hamilton and Thompson (1994), Findlay and Watson (1996), Bhasin (1999), Hussain (1994), Akinlo and Odusola (1999), Bankole

Table 2: Performance of GDP and workers’ remittances

Source: Computed using CBN Statistical Bulletin, 2013 and IMF Balance of Payments Yearbook (CD-ROM 2012).

Figure 1: Trend of the growth rate of trade, remittances and GDP

Source: Computed using CBN Statistical Bulletin (online version, 2013) and IMF Balanace of Payments Yearbook (CD-ROM 2012)
et al. (1999), Baliamoune-Lutz (2011), Abdulai and Jacquet (2002), Oyejide (2007), Ahmed and Kwan (1991) and Mijiyawa (2013).

Hamilton and Thompson (1994) developed a theoretical proposition that unilateral liberalization creates income convergence. In the long run, both unilateral and multilateral trade liberalization generates a positive impact on steady-state growth. Findlay and Watson (1996) show that China’s growth history depends on the market access for her export products. They pointed out that the rest of the world needs to cooperate with China in order to avoid instability. Bhasin (1999) investigated the export-led growth in Ghana using annual data for the period 1966 to 1996. Employing dynamic simultaneous equation models, the export-led growth was established. Furthermore, output growth causes export growth. Bankole et al. (1999) investigated the case of Nigeria using annual data from 1965 to1999. The authors employed a long-run and short-run dynamic model and the result validated the hypothesis. Abdulai and Jacquet (2002) were interested in the short- and long-run relationship among economic growth, exports, investment and labour force. Using cointegration and error correction estimation techniques, they discovered that there was longrun relationship among the variables. They also found causality running from export to GDP both in the short and long run. Baliamoune-Lutz (2011) further investigated the export-led growth but focused on the pattern and nature of trade. The ArellanoBond GMM technique for a panel of African countries in the period 1995-2008 suggested that the destination of export determines whether it will enhance growth or not. To this end, African export concentration in China contributed to the region’s growth. Also, imports contributed to the growth history rather than inhibited it. The work of Oyejide (2007) also demonstrated that exports matter for growth and so, promoting export expansion and diversification improves growth performance in Africa. Recently, Mijiyawa (2013) revisited growth drivers and showed that although changes in economic fundamentals matter for growth, exports appear to be the strongest growth driver in Africa. However, Ahmed and Kwan (1991) found no evidence in favour of the export-led growth hypothesis in 47 African countries.

Most of the empirical evidence seems to support the export-led hypothesis even though the degrees of effect differ across countries. In the case of the remittance-growth nexus, economic literature identified altruism, exchange, insurance, investment and inheritance as channels through which growth can be enhanced (Lopez-Cordova and Olmedo, 2006). 9{ }^{9} It was claimed that a significant portion of remitted funds are spent on consumption, but the nature of consumption by remittance receivers can as well reduce the consumption of non-recipients through the Dutch-Disease effect (Chami et al., 2003). Hence, the consumption effect of remittances on economic growth is unclear. If the Dutch-Disease effect prevails, remittances may reduce economic growth. In fact, Chami et al. (2003) in their study of panel data for 104 low and transition economies, using OLS and the Fixed Effects Model (FEM), find that an increase in the ratio of country’s remittances to GDP will reduce GDP growth. Their argument is that receivers substitute remittances for labour effort. Adams (1991) finds, in a sample of 74 Egyptians households, that the receipt on remittances increases the marginal propensity to invest (MPI) primarily in residences and land. Taylor (1992), Lopez and Selingson (1991) observe that in Latin American countries, remittances expenditure on housing, land and jewelry take a high percentage. Fayissa and Nsiah (2008) showed, in their unbalanced panel data for 37 African countries that remittances positively impacted on growth through investment.

Recently, Khathlan (2012) employed an autoregressive distributed lag (ARDL) test and an error correction model to investigate the long-run and short-run relationship between workers’ remittances and economic growth in Pakistan in the 1976-2010 period. He discovered that there was a positive and significant relationship between workers’ remittances both in the long and short run. In the same vein, Anyanwu and Erhijakpor (2010) employed a panel dataset on poverty and workers’ remittances for 33 African countries over the period 1990-2005. Their results showed that remittances reduce poverty level, depth and severity. Cooray (2012) focused on the impact of remittances on the economy of South Asia. In a panel dataset over the 1970-2008 period, remittances are found to have a significant positive effect on economic growth between 1976 and 2005. Its interaction with consumption and financial development further strengthens the growth effect. However, Jongwanich (2007) examined the case of Asia-Pacific spanning from 1993 to 2003 and discovered that remittances show marginal positive impact on growth, even though it reduces poverty level. But in the field survey conducted by Sofranko and Idris (2009) for 170 families in a small community in Pakistan, he found little evidence for the remittances effect of growth.

Some studies such as Akkoyunlu (2009), Siddiqui and Kemal (2006) and Ahmed and Uddin (2009) investigated the causal relationship between remittances and trade in Bangladesh between 1975 and 2005. Akkoyunlu (2009) developed and estimated an empirical model of Turkish migration to Germany and tested the model for the 1969-2004 period using the cointegration technique. The result shows that trade, GDP and remittances were interrelated. In Pakistan, Siddiqui and Kemal (2006) assessed the impact of trade liberalization and remittances shocks on growth using the computable general equilibrium (CGE) framework and found that an increase in trade liberalization and remittances reduce economic growth. The works of Ahmed and Uddin (2010) concentrated on the same causality for Bangladesh using annual data from 1976 to 2005. They found limited support in favour of export-led growth hypothesis because exports, imports and remittances cause GDP growth only in the short run.

Clearly, there is empirical evidence showing the causal relationship among remittances, trade and economic growth but the evidence is diverse and scanty. Research for Nigeria is missing despite its position in the remittances ranking in SSA coupled with a high degree of openness and relatively stable economic growth the country has been experiencing over a decade now. This is the empirical gap that this study seeks to fill.

4. Methodology and Data

Given the fact that exports, imports, remittances and other controlled variables are extracted from secondary sources, a stationarity test was performed so as to determine their state of equilibrium. The Augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) tests were conducted in order to check the stationary property of the dataset (Perron, 1990). On the account that the variables are stationary at first difference, a test of cointegration was performed. The essence for cointegration is to allow the infusion of longrun and short-run information in the same model and therefore overcome some of the drawbacks associated with the loss of information that occurs from other attempts to address the problem of non-stationarity through differencing. Following Johansen (1988) and Johansen and Juselius (1990), the basics of cointegration and by extension, vector error correction, is demonstrated. Consider an unrestricted vector autoregressive (VAR) model up to lag kk in which the process YtY_{t} for a given Yt−k+1+,…,Y0Y_{t-k+1+}, \ldots, Y_{0} is defined by the following equation:

Yt=α+λ1Yt−1+λ2Yt−2+…………+λk−1Yt−k+1+εt,t=1,2,3,……,TY_{t}=\alpha+\lambda_{1} Y_{t-1}+\lambda_{2} Y_{t-2}+\ldots \ldots \ldots \ldots+\lambda_{k-1} Y_{t-k+1}+\varepsilon_{t}, t=1,2,3, \ldots \ldots, T

where εt\varepsilon_{t} is assumed to be an independently and identically distributed pp-dimensional Gaussian error term with mean zero and variance matrix σ,Yt\sigma, Y_{t} is a vector integrated of order one (I(1))(\mathrm{I}(1)) variables and α\alpha is a vector of constants. If YtY_{t} is non-stationary, then Equation 1 can be expressed in (first) differenced error correction as shown in Equation 2:

ΔYt=α+ℑ1ΔYt−1+ℑ2ΔYt−2+…………+ℑk−1ΔYt−k+1+λkYt−k+εt\Delta Y_{t}=\alpha+\Im_{1} \Delta Y_{t-1}+\Im_{2} \Delta Y_{t-2}+\ldots \ldots \ldots \ldots+\Im_{k-1} \Delta Y_{t-k+1}+\lambda_{k} Y_{t-k}+\varepsilon_{t}

where ΔYt\Delta Y_{t} is Yt−Yt−1Y_{t}-Y_{t-1} and ℑi\Im_{i} is defined as −(1−λ1−λ2−…………−λk);i=1,2,3,4,…,k-\left(1-\lambda_{1}-\lambda_{2}-\ldots \ldots \ldots \ldots-\lambda_{k}\right) ; i=1,2,3,4, \ldots, k
The term λkYt−k\lambda_{k} Y_{t-k} differentiated Equation 2 from the traditional (basic) VAR model. The coefficient matrix λ\lambda contains information about the long-run relationships between the variables. This coefficient matrix could possess full rank, in which case, the vector process YtY_{t} is stationary, that is, λ=ρ\lambda=\rho where ρ\rho represents full rank matrix. If λ\lambda possesses no rank, then λ\lambda is a null matrix and so λ=0\lambda=0. In this case, Equation 2 corresponds to a traditional differenced vector time-series model. The third possibility is when λ\lambda possesses rr rank, where rr lies between zero and ρ\rho, that is, 0<r<ρ0<r<\rho. This implies that there exists rr cointegrating vectors

with the property that λ=γβ′\lambda=\gamma \beta^{\prime} where γ\gamma and β\beta are ρ×r\rho \times r matrices and the cointegrating vector β\beta has the property that Yi′βY_{i}^{\prime} \beta is stationary even though YtY_{t} is non-stationary. In this case, Equation 2 becomes an error correction model.

The likelihood ratio of cointegrating rank rr can be tested using trace test (TT) and minimum Eigen value (MEV) (Johansen, 1988 and Johansen and Juselius, 1990). The trace test is compared with the null hypothesis that there are at most rr cointegrating vectors against the alternative of more than rr cointegrating vectors and the trace statistic is computed using the following formula:

Trace =−T∑i=r+1ρln⁡(1−φi)\text { Trace }=-T \sum_{i=r+1}^{\rho} \ln \left(1-\varphi_{i}\right)

and φ^r+1…φ^ρ\widehat{\varphi}_{r+1} \ldots \widehat{\varphi}_{\rho} are ρ−r\rho-r smallest estimated Eigen value. The MEV gives information about the likelihood ratio test statistic for the null hypothesis of rr cointegrating vectors against the alternative of r+1r+1 cointegrating vectors and it is given by Equation 4:

ϕmax⁡=−T[ln⁡(1−φi)]\phi_{\max }=-T\left[\ln \left(1-\varphi_{i}\right)\right]

The next step is to determine the numbers of variables that should enter into a long-run equilibrium relationship. This is usually done by testing linear restrictions on the long-run coefficients after they have been normalized. The hypothesis of long-run exclusion of each variable is tested using a likelihood ratio test with χ2\chi^{2} distribution and kk degrees of freedom where kk is the number of restriction(s) tested. The decision rule is that if the test statistic is less than 0.05 critical value then the affected coefficient is significant and so the variable will be included in the long-run equilibrium relationship. It turns out that the number of cointegrating relationships will result in a corresponding number of residual series which are the error correction terms that will then be used in the subsequent vector error correction model (VECM). The estimable VECM is specified as follows:

ΔYt=γ11ε1,t−1+γ12ε2,t−1+γ13ε3,t−1+∑t=1mπ11ΔYt−1+∑t=1mπ12ΔXt−1+∑t=1mπ13ΔMt−1+∑t=1mπ14Rt−1+α1ΔXt=γ21ε1,t−1+γ22ε2,t−1+γ23ε3,t−1+∑t=1mπ21ΔYt−1+∑t=1mπ22ΔXt−1+∑t=1mπ23ΔMt−1+∑t=1mπ24Rt−1+α2ΔMt=γ31ε1,t−1+γ32ε2,t−1+γ33ε3,t−1+∑t=1mπ31ΔYt−1+∑t=1mπ32ΔXt−1+∑t=1mπ33ΔMt−1+∑t=1mπ34Rt−1+α3ΔRt=γ41ε1,t−1+γ42ε2,t−1+γ43ε3,t−1+∑t=1mπ41ΔYt−1+∑t=1mπ42ΔXt−1+∑t=1mπ43ΔMt−1+∑t=1mπ44Rt−1+α4\begin{aligned} & \Delta Y_{t}=\gamma_{11} \varepsilon_{1, t-1}+\gamma_{12} \varepsilon_{2, t-1}+\gamma_{13} \varepsilon_{3, t-1}+\sum_{t=1}^{m} \pi_{11} \Delta Y_{t-1}+\sum_{t=1}^{m} \pi_{12} \Delta X_{t-1}+\sum_{t=1}^{m} \pi_{13} \Delta M_{t-1}+\sum_{t=1}^{m} \pi_{14} R_{t-1}+\alpha_{1} \\ & \Delta X_{t}=\gamma_{21} \varepsilon_{1, t-1}+\gamma_{22} \varepsilon_{2, t-1}+\gamma_{23} \varepsilon_{3, t-1}+\sum_{t=1}^{m} \pi_{21} \Delta Y_{t-1}+\sum_{t=1}^{m} \pi_{22} \Delta X_{t-1}+\sum_{t=1}^{m} \pi_{23} \Delta M_{t-1}+\sum_{t=1}^{m} \pi_{24} R_{t-1}+\alpha_{2} \\ & \Delta M_{t}=\gamma_{31} \varepsilon_{1, t-1}+\gamma_{32} \varepsilon_{2, t-1}+\gamma_{33} \varepsilon_{3, t-1}+\sum_{t=1}^{m} \pi_{31} \Delta Y_{t-1}+\sum_{t=1}^{m} \pi_{32} \Delta X_{t-1}+\sum_{t=1}^{m} \pi_{33} \Delta M_{t-1}+\sum_{t=1}^{m} \pi_{34} R_{t-1}+\alpha_{3} \\ & \Delta R_{t}=\gamma_{41} \varepsilon_{1, t-1}+\gamma_{42} \varepsilon_{2, t-1}+\gamma_{43} \varepsilon_{3, t-1}+\sum_{t=1}^{m} \pi_{41} \Delta Y_{t-1}+\sum_{t=1}^{m} \pi_{42} \Delta X_{t-1}+\sum_{t=1}^{m} \pi_{43} \Delta M_{t-1}+\sum_{t=1}^{m} \pi_{44} R_{t-1}+\alpha_{4} \end{aligned}

where YY is the log of real GDP; XX is the log⁡\log of real export; MM is the log⁡\log of real import; and RR is the log⁡\log of real remittances.
Real GDP is nominal GDP deflated by the GDP deflator (with base year 2002). Real export is total export deflated by export price index while real import is total import deflated by import price index. Remittances are defined as workers’ remittances which are part of immigrant income sent back home for the purpose best understood by the sender. This variable was deflated by the GDP deflator. All the variables, between 1980 and 2012 were sourced from the World Development Indicators (WDI) 2012, except workers’ remittances which was sourced from the IMF Balance of Payments Yearbook (2012). All the variables were expressed in logarithmic transformation so as to reduce the problem of heteroscedasticity (Gujarati, 1995).

5. Result and Discussion

The presentation of results begins with an investigation of the stationarity of the series. Table 3 reports the results obtained from the ADF unit root test while Table 4 shows the results of the PP unit root test. As the results show, all the variables were found to be stationary at first difference when a constant was included and when constant and trend were included. This implies that all the variables are integrated of order 1.

Table 3: Augmented Dickey-Fuller test

Table 4: PP test

Table 5: Cointegration test

∗,∗∗{ }^{*},{ }^{* *} indicate the number of cointegrating equations at 1%1 \% and 5%5 \% respectively.

Following the results of the ADF and PP tests, the next step is to test the nature of cointegration of the variables. The Schwartz Information Criterion (SIC), the Hernan-Quinn Information Criterion (HQIC) and the Akaike Information Criterion (AIC) provide information about the maximum lag length to be included in the VAR system. Table 5 shows that there are two and one cointegrating equations in the system at the 1 per cent and 5 per cent level of significance respectively. Thus, there are three cointegrating equations in the system, suggesting that the series was driven by at least three common trends. Thus, the residuals from the first three equations of the VAR were saved and used as the error correction terms in the subsequent tests for Granger causality.

Table 6 contains statistical information about the possible causalities of real GDP (LNRGDP), real export (LNREXP), real import (LNRIMP) and real remittances (LNRREM) together with the error correction terms, namely Σ1, T−1,Σ2, T−1\Sigma_{1, \mathrm{~T}-1}, \Sigma_{2, \mathrm{~T}-1}, and Σ3, T−1\Sigma_{3, \mathrm{~T}-1}. It must be recalled that these error correction terms represents the possible long run causality.

The result suggests that real export, import and remittances Granger cause real GDP in the short run. Notably, real GDP causes real exports and real imports in the short run but did not cause real remittances. This implies that the causal relationship between real GDP, exports and imports are bidirectional while the causal relationship between real GDP and real remittances are unidirectional. The result is in line with the findings of Abdulai and Jacquet (2002) in the case of Côte d’Ivoire and Bankole et al. (1999) in Nigeria where the export-led growth hypothesis was established. Meanwhile, since oil accounts for more than 95 per cent of Nigeria’s total export, it is clear that resource endowment, rather than economies of scale dictate the growth effect of exports.

Table 6: Vector error correction (VEC) Granger causality result

Note: Figures in brackets are tt-values.

The table shows that there is a feedback effect from import to economic growth. This should not be surprising because Nigeria is input import dependent. That is, the capital and intermediate goods are imported from abroad. Thus, the availability of these resources and the ease with which it can be accessed tends to dictate the growth behaviour of the economy. Even where local content requirement is binding, input imports serve as complement. This result supports the findings of Baliamoune-Lutz (2011) who found that imports are growth-enhancing variable in Africa. Our result also confirms the causal effect of remittances on economic growth, thereby supporting the findings of Cooray (2012). As discussed earlier, remittances cause growth through consumption, investment and reduction of the problem of liquidity/credit constraints and by extension, poverty (Anyanwu and Erhijakpor, 2010). Thus, our result suggests that the consumption pattern arising from remittances appears to be productive. Alternatively, the share of remittances spent on productive activities such as investment tends to more than offset the share spent on non-productive activities.

Further, there exists reverse causality running from real GDP to export and to imports. The reason for reverse causality from GDP to trade can be traced to some policy shifts such as the movement from import substitution industrialization in the mid-1980s to export orientation and trade liberalization. Although there could be traces of import substitution policy that encourages production of some goods on the basis of either comparative production advantage or industrialization, it must be noted that consumption of imported goods arising from increased income tends to outweigh the consumption of import substituting products. The causal effect of GDP on export could arise from government’s fiscal stance that favours the exports sector. For instance, government provides an export grant to expand non-oil export and this is complemented with the provision of subsidy which expectedly reduces the cost of production. Also, increased GDP creates a signalling effect and by implication, attracts foreign investment in the form of physical, financial and human capital that are needed in the export sector.

Although remittances cause GDP, the reverse causality was insignificant, albeit, rightly signed. This result is somewhat surprising because in most empirical evidence, remittances were argued to be countercyclical, falling when the economy is improving and rising during economic downturn, and in fact, the trend analysis supports this. Observably, reasons for the absence of reverse causality could partly be explained by moral hazard and partly by humanitarian consideration or self-insurance purposes. It can also be that remittance is a fungible resource, in which case ‘white elephant projects’ benefit from fiscal expansion and are grossly induced by corruption and capital flight. Incidence of resource fungibility, corruption and capital flight are evidenced in Nigeria and this could dampen the counter-cyclicality effect of remittances.

The long-run causality captured by the estimated errors from the VAR model shows that long-run real GDP, long-run export and long-run import cause short-run real GDP and long-run import causes short-run import. However, there is no reverse causality among the variables. This implies that the direction of short-run and long-run causality appears to be unidirectional. There are some reasons that can be advanced for this. First, the incessant change of political structure between 1980 and 1999 was usually followed by economic distortions, arising from abandonment of one economic policy and the revival of another, which in turn would be short lived, giving way to a brand new policy. For instance, SAP was discontinued after three years of inception and a guided regulation was implemented. This policy led to financial crisis which eventually swallowed some banks. Also, the country was battling with socio-ethnic challenges which scared away intending investors, thereby attenuating the low growth time path.

The situation was arrested partly by money-financed fiscal expansion but at the expense of an inflation spiral and exchange rate volatility. Therefore, within ten years, the economy embarked on at least three different types of economic policies. This policy-creation-policy-destruction phenomenon did not allow the system to complete the adjustment process at any period in time.

6. Conclusion and Policy Implications

This study examined the causal relationship among GDP, export, imports and remittances. The study intended, among others, to test the validity of export-led growth hypothesis and extend the empirical literature by investigating the causal relationship between remittances and GDP, export and imports. The result from the VECM Granger causality test shows that Nigeria still maintains the export-led growth hypothesis even though exports are still dominated by oil products. Similar results were obtained by Bankole et al. (1999) in the case of Nigeria, Bhasin (1999) in the case of Ghana and Abdulai and Jacquet (2002) in the case of Côte d’Ivoire. But beyond Bankole et al.'s result, the causal relationship between GDP and exports was bidirectional, likewise the causal relationship between real GDP and imports. This implies that exports, imports and growth cause each other in Nigeria. However, factors that could enhance export-led growth may not be the same as factors that facilitate growth-led export. For instance, government policy may be effective in the case of the latter while foreign demand and comparative natural resource advantage drive the former. Therefore, to the extent that such causal relationship is favourable, it is advisable to propose and implement policies that will stimulate the export sector in particular and economic growth in general. The ‘one exports product one local government’ strategy is a good export policy and it should be implemented effectively and efficiently.

The bidirectional causation of real GDP and import implies that imports matter for economic growth in Nigeria, likewise growth is very important for import. Again, the factors that account for the latter are different from the former. Nigeria’s real sector depends heavily on input imports and so, as long as access is restricted the real sector will be notably hampered. Thus, to the extent that this causation is favourable, the government should make input import relatively accessible. However, if the exchange rate strongly dictates the decision to import, then government may be helpless, particularly under a floating exchange rate regime. If this is the case, the monetary authority could intervene by using appropriate monetary instruments to dwarf the negative effect.

Unfortunately such policy may favour import demand of consumption goods. The price elasticity of import in Nigeria is relatively inelastic and this suggests that a large change in price will only result in a very small change in import demand. The reason for the inelastic import price is that Nigerians have higher preference for imported products (Ogwumike and Olubiyi, 2009; Olubiyi, forthcoming), and the relevant inputs needed for industrial growth are imported (Ogwumike and Olubiyi, 2009; Olubiyi, forthcoming; Adenikinju and Chete, 1999). Thus, to the extent that preferences are high and firms wish to produce, a large price increase may lead to a mild decrease in import (Adenikinju and Chete, 2002). Exchange rate policy that favours input import will also favour imported consumption goods. Therefore, there is a need to investigate the respective degree of response of input import and imported consumption goods to changes in the exchange rate. If the latter is inelastic while the former is elastic, then it is unambiguously advisable to leverage on exchange rate. If both are inelastic but the former is more inelastic than the latter, exchange rate policy can still work. The unidirectional causation from remittances to real GDP implies that remittances matter for growth in Nigeria. This is not surprising since a substantial share was used for household consumption, and human capital development (Adepoju, 2007). The growth enhancing effect of remittances in Nigeria is well articulated in Olubiyi (forthcoming) where it was established that remittances impacted positively on consumption, investment and imports with the sum of coefficients of consumption and investment more than offset that of the imports. Thus, the growth effect of remittances appears to be from the demand and the supply sides. Meanwhile, if the demand side effect exceeds the supply side, it will lead to inflation, thereby causing macroeconomic instability. To dampen this effect, the monetary authority should closely monitor the spending behaviour of remittance receivers and use appropriate monetary policy instruments to cool the economic overheating arising from excessive demand. Alternatively, the authorities can discourage ostentatious spending through attractive saving policy. From the supply side, policies that will enhance productive investment, particularly in the real sector should be harnessed.

Notes

1. The monetary authority can prevent inflation arising from monetary expansion by retiring some liquidity through the sale of domestic bonds.

References

Abdulai, A. and P. Jacquet (2002), ‘Export and Economic Growth: Cointegration and Causality. Evidence for Cote d’Ivoire’, African Development Review, Vol. 14, No. 2, pp. 1-17.

Adams, R. (1991), ‘The Economics Uses and Impact of International Remittances in Rural Egypt’, Economic Development and Cultural Change, Vol. 39, pp. 695-722.

Adelman, I. (1984), ‘Beyond Export-led Growth’, World Development, Vol. 12, pp. 937-49.
Adenikinju, A. and L. Chete (1999), ‘Trade Liberalization, Market Structure and Productivity in Nigeria’, Nigerian Journal of Economic and Social Sciences, Vol. 41, No. 3, pp. 385-404.

Adepoju, A. (2007), ‘Migration in Sub-Saharan Africa’, Background Paper commissioned by the Nordic Africa Institute for the Swedish Government, White Paper on Africa.

Ahmed, J. and A. Kwan (1991), ‘Causality between Export and Economic Growth: Evidence from Africa’, Economic Letters, Vol. 37, pp. 243-48.

Ahmed, H. and M. Uddin (2009), ‘Export, Imports, Remittances and Growth in Bangladesh: An Empirical Analysis’, Trade and Development review, Vol. 2, No. 2, pp. 79-92.

Akinlo, A. and A. Odusola (1999), ‘An Alternative Framework for Assessing the Impacts of Nigeria’s Naira Depreciation on Output and Inflation’, Final Report, African Economic Research Consortium, AERC Workshop, Accra, Ghana.

Akkoyunlu, S. (2009), ‘Trade, Aid, Remittances and Migration’, KOF Working Papers/KOF Swiss Economic Institute, ETH Zurich, 229.

Anyanwu, C. and E. Erhijakpor (2010), ‘Do International Remittances Affect Poverty in Africa?’, African Development Review, Vol. 22, No. 1, pp. 51-91.

Baliamoune-Lutz, M. (2011), ‘Growth by Destination (Where You Export Matters): Trade with China and Growth in Africa Countries’, African Development Review, Vol. 23, No. 2, pp. 202-18.

Bankole, S., A. Lawanson and O. Lawanson (1999), ‘Impact of Manufactured Goods’ Exports on Economic Growth: A Dynamic Econometric Model for Nigeria’, African Journal of Economic Policy, Vol. 11, No. 2, pp. 1-26.

Bhasin, V. (1999), ‘Export and Economic Growth: Empirical Evidence from Ghana’, African Journal of Economic Policy, Vol. 6, pp. 31−5031-50.

Buffie, E. (1992), ‘On the Condition for Export-led Growth’, Canadian Journal of Economics, Vol. 25, pp. 211-25.
Central Bank of Nigeria (CBN) (2013), Statistical Bulletin. Available online at www.cenbank.org/documents/Statbulletin.asp.
Chami, R., C. Fullenkamp and S. Jahjah (2005), ‘Are Immigrant Remittances Flows a Source of Capital for Development?’, IMF Staff Paper, Vol. 52, No. 1, pp. 55-81.

Cooray, A. (2012), ‘The Impact of Migrant Remittances on Economic Growth: Evidence from South Asia’, African Development Review, Vol. 20, No. 5, pp. 985-98.

Egwaikhide, F. (1997), ‘Trade Policy Reform in Nigeria: A Historical Outline in External Trade and Economic Development in Nigeria’, Nigerian Economic Society, Ibadan.

Fayissa, B. and C. Nsiah (2008), ‘The Impact of Remittances on Economic Growth and Development of Africa’, Department of Economics and Finance Working Paper Series.

Findlay, C. and A. Watson (1996), ‘Economic Growth and Trade Dependency in China’, Chinese Economies Research Centre, DP 96/5, University of Adelaide.

Gujarati, D. (1995), Basic Econometrics, 3rd edn, McGraw Hill, New York.
Hamilton, N. and C. Thompson (1994), ‘Export Promotion in a Regional Context: Central America and Southern Africa’, World Development, Vol. 22, pp. 1379-92.

Hussain, I. (1994), ‘Results of Adjustment in Africa’s Selected Countries’, Finance and Development, 6-9 June.
IMF (2011), Balance of Payments Yearbook, International Monetary Fund, Washington DC.
IMF (2012), Balance of Payments Yearbook, International Monetary Fund, Washington DC.
Johansen, S. (1988), ‘Statistical Analysis of Cointegration Vectors’, Journal of Economic Dynamics and Control, Vol. 12, pp. 231-54.

Johansen, S. and K. Juselius (1990), ‘Maximum Likelihood Estimation and Inference on Cointegration: With an Application to Demand for Money’, Oxford Bulletin of Economics and Statistics, Vol. 52, pp. 169-210.

Jongwanich, J. (2007), ‘Workers’ Remittances, Economic Growth and Poverty in Developing Asia and the Pacific Countries’, UNESCAP Working Paper, WP/07/01.

Khathlan, K. (2012), ‘The Link between Remittances and Economic Growth in Pakistan: A Boon to Economic Stability’, British Journal of Economic Management & Trade, Vol. 2, No. 3, pp. 167-85.

Krugman, P. (1984), ‘Import Protection as Export Promotion’, in H. Kierzkowski (ed.), Monopolistic Competition in International Trade, Oxford University Press, Oxford.

Lal, D. and S. Rajapatirana (1987), ‘Foreign Trade Regimes and Economic Growth in Developing Countries’, The World Bank Research Observer, Vol. 2, No. 2, pp. 189-217.

Lancaster, K. (1980), ‘Intra-industry Trade under Perfect Monopolistic Competitions’, Journal of International Economics, Vol. 10, pp. 151-75.

Lopez-Cordova, E. and A. Olmedo (2006), ‘International Remittances and Development: Existing Evidence, Policies and Recommendations’, INTAL/ITD Occasional Paper No. 1.

Lopez, J. and M. Selingson (1991), ‘Small Business Development in El Salvador: The Impact of Remittances’, in S. DiazBriquets and S. Weintraub (eds.), Migration, Remittances and Small Business Development: Mexico and Caribbean Basin Countries, Development and International Migration in Mexico, Central America, and the Caribbean Basin, Vol. 4, Westview Press, Boulder, CO.

Mijiyawa, A. (2013), ‘Africa’s Recent Economic Growth: What Are the Contributing Factors?’, African Development Review, Vol. 25, No. 2, pp. 289-302.

Ogwumike, O. and E. Olubiyi (2009), ‘Remittances and Economic Development in Sub-Saharan Africa’, Indian Journal of Economics, Vol. LXXXX, No. 356, pp. 1-28.

Olubiyi, E. (Forthcoming), ‘Migrants’ Remittance Flows and Use: A Case of Nigeria’, Network of Migration Research on Africa (NOMRA), Lagos, Nigeria.

Oyejide, T. (2007), ‘Relaxing Export Supply Capacity Constraints and Promoting Export Expansion and Diversification in Africa: The Role of Special Export Development Strategies’, African Development Review, Vol. 19, No. 1, pp. 96-122.

Pack, H. (1988), ‘Industrialization and trade’, in H. Chenery and T. Srinivasan (eds.), Handbook of Development Economics, Vol. 1, Elsevier, Amsterdam.

Perron, P. (1990), ‘Testing for a Unit Root in a Time Series with a Changing Mean’, Journal of Business and Economic Statistics, Vol. 8, pp. 153-62.

Rodrik, D. (2007), ‘The Real Exchange Rate and Economic Growth: Theory and Evidence’, Brookings Papers on Economic Activity, Fall.

Serieux, A. (2010), ‘Remittances and Reverse Flows in Developing Countries’, The Ideal Working Paper Series, Paper No. 02/ 2010 .

Siddiqui, R. and A. Kemal (2006), ‘Remittances, Trade Liberalization, and Poverty in Pakistan: The Role of Excluded Variables in Poverty Change Analysis’, PIDE Working Papers 2006:1.

Sofranko, A. and K. Idris (2009), ‘Use of Overseas Migrants’ Remittances to the Extended Family for Business Investment: A Research Note’, Rural Sociology, Vol. 64, No. 3.

Taylor, J. (1992), ‘Remittances and Inequality Reconsidered: Direct, Indirect and Intertemporal Effects’, Journal of Political Modelling, Vol. 14, No. 2, pp. 187-208.

Vacaflores, D., R. Kishan and J. Trinidad (2012), ‘Remittances, International Reserves, and Exchange Rate Regimes’, Working Paper Series, Department of Finance and Economics, Texas State University, Texas.

World Bank (2012), World Development Indicators, World Bank, Washington DC.
World Bank (2013), World Development Indicators, World Bank, Washington DC.
WTO (2011), ‘Trade Policy Review: Report by the Secretariat Nigeria’, WT/TPRS/247, World Trade Organization, Washington DC.