Investigating the Role of Regional Economic Integration on Growth: Fresh Insights from South Asia

Abstract

Regional economic integration is the key to achieving prosperity and stability. However, intra-regional trade in South Asia accounts for not more than 5%–6% of their total trade. This study aims to examine the role played by regional economic integration in determining the economic growth of South Asian countries over the period 1980–2015. Since shocks in one country may affect another country in the region, this is taken into account in the article by employing methodologies that are robust to cross sectional dependence. Specifically, continuously-updated and bias-corrected (CupBC) of Bai et al. (2009) and Dumitrescu–Hurlin panel causality test (2012) have been employed to estimate long-run coefficients and determine the direction of relationship among the variables, respectively. The findings suggest that economic integration increases economic growth significantly in this region. However, contrary to popular belief, both democracy and human capital are negatively related to economic growth. Bidirectional causality is found between economic integration and democracy, regional integration and human capital, democracy and human capital and, democracy and labor. This study also presents several policy implications for South Asian countries.

Keywords

Integration growth South Asia cross-sectional dependence CUP causality

Introduction

The idea of pursuing international trade became globally widespread after World War II. Countries started to engage with each other more often, and eventually, due to less travel time and transport costs, they came to rely more on neighboring countries. In today’s world, economic integration between neighboring countries has become a major instrument of prosperity and stability. Though trade agreements were the basis of regional cooperation among countries in the past, today’s world is well-integrated both politically and economically (MacKay, 2005). When global coordination fails, it is regional integration that usually plays the role of building up the process of liberalization. Therefore, it has been rightfully argued that world politics and economy are more likely to be determined by regional integration in the coming years (De Lombaerde, 2006).

According to the World Bank report, South Asia is considered to be the fastest growing region in the world. Currently, South Asia is facing mounting challenges such as huge current account deficit, extreme poverty, high rate of unemployment, lower living standards, and slow progress in structural transformation. However, the major concern for South Asia is not any of the above-mentioned problems; rather, it is the poor regional integration among the countries in the region. Despite having many commonalities including colonial background and geography, economically, South Asia is the least connected region and this low level of economic integration poses a grave threat to the member states. Intra-regional trade accounts for not more than 5%–6% of their total trade. South Asia has a huge potential for trade growth, but due to some political, economic, and social factors, the region cannot realize its full potential in trade. Historically, South Asian countries have followed inward looking policies which is the reason for it being a latecomer in terms of regional integration (Kumar & Singh, 2009). It was Sri Lanka which first liberalized its economy in 1977, followed by India and others. This liberalization policy taken by Sri Lanka in the early 1980s paved the way for increased regional integration among the South Asian economies, and the initiation of South Asian Association for Regional Cooperation (SAARC) ensured a smoother progress (Regmi et al., 2017).

An economic and geopolitical organization, SAARC aims to promote the welfare and improve the standard of living of its people through higher economic growth and sociocultural development (Iqbal, 2006). The need for intra-regional trade among the south Asian countries was realized through the agreement of SAPTA (South Asian Preferential Trade Agreement) in 1993 and later by SAFTA (South Asian Free Trade Area) in 2004. But unlike other regions such as Southeast Asia or North America, SAARC could not make progress toward regional integration and trade (Raihan, 2014). Trade within this region accounted for less than 3% before the SAPTA was signed and it still remained below 6% in 2015 (Figure 1). Apparently, the trade agreements that have taken place till now had very little success in lifting the intra-regional trade above 10%. This regional trade has also been very minimal as compared to other regions such as Association of Southeast Asian Nations (ASEAN), EU, or even Latin America and Caribbean (Figure 2).

Figure 1.

Source: Asia Regional Integration Center, Asian Development Bank (2015).

Figure 2.

Source: Integration Indicators Database, Asia Regional Integration Center, Asian Development Bank (2015).

Nevertheless, there is scope for developing better integration in this region and enormous potentials exist for greater intra-regional trade, given the appropriate conditions. This would not only benefit the individual members but also significantly affects the growth of this region as a whole.

This study aims to assess the extent to which economic integration exerts its influence on growth in South Asian region. Three issues have been addressed specifically. First, South Asia is the fastest growing region in the world in terms of GDP; however, regional trade among the countries is still very low compared to ASEAN or EU. Economic growth is the key instrument to tackle critical issues affecting South Asia such as current account deficit (Özer et al., 2018; Yurdakul & Ucar, 2015), poverty (Fosu, 2009; Michálek and Výbošťok, 2019; Nansadiqa et al., 2019; Roemer & Gugerty, 1997; Santos et al., 2019), and unemployment (Lewis et al., 2019; Sahnoun & Abdennadher, 2019), and although indispensability of intra-regional trade to stimulate growth is well recognized by the previous studies, there still remains uncertainty about the direction or the degree of their relationship due to heterogeneity across countries (Campos et al., 2019). Hence, further work is required to identify the extent to which intra-regional trade exerts its influence on growth. This study focuses specifically on the growth impact of regional integration rather than the welfare impacts as in previous studies. The reason why we do so is because the efforts toward welfare can be significantly undermined if growth is not achieved at the same time.

Second, most of the South Asian countries have evolved from a common colonial background and have a shared history and culture. Hence, if a shock occurs in one country, it is likely to affect another country in this region, which is why studies dealing with South Asia must take account of cross-sectional dependence. Following the suggestions of Ehigiamusoe and Lean (2018), this study takes cross-sectional dependence into account and argues that results may show overestimation/underestimation if this factor is ignored. Finally, to date, very few empirical studies have applied continuously updated (CUP) estimation procedure by Bai et al. (2009) to examine growth–integration nexus and to the best of the author’s knowledge, as yet, no such empirical analysis has been conducted for South Asia. This estimator is robust against cross-sectional dependence, and it can take care of endogeneity and correlation bias. Therefore, this methodological approach of Bai et al. (2009) can be considered as our major contribution to the empirical literature on South Asia. Third, this study uses intra-regional share to total trade as a proxy of regional integration following Beckfield (2006). Previous studies used a dummy variable to observe whether economic integration, represented by a specific agreement like SAFTA or ASEAN Free Trade Area (AFTA), has any impact on growth (De Melo et al., 1992; Henrekson et al., 1997; Moinuddin, 2013; Taguchi, 2018), but using a dummy variable is assuming that effect may come from by just signing the agreement. The expected impacts, however, depend on the proper implementation of the agreement in the upcoming years as well. Only using a dummy variable will not result in any significant impact on growth (Berthelon, 2004).

Literature Review

Stylized Facts About Failure of Regional Integration in South Asia

One of the major reasons why South Asia is still behind other regions in terms of integration is because of the high tariff rate that exists within the countries. Both the intra-regional tariffs such as most-favored-nation (MFN) and effectively applied tariffs (AHS) are significantly higher for South Asia compared to other regions such as sub-Saharan Africa, North America, and East Asia and Pacific (Figure 3).

South Asian countries also have higher intra-regional trade costs. If we consider the non-tariff trade costs (% ad valorem equivalent, all goods) in 2015, India–Pakistan trade costs amounted to 158%, whereas it was only 133% for India–Brazil In 2011. On the other hand, bilateral cost for manufacturing goods was estimated to be 244% within the South Asian region, whereas South Asia–East Asia and Pacific amounted to 121%, expressed in ad valorem equivalent form (World Bank & UNESCAP, 2015). Many non-tariff measures are also in place between the countries including sanitary and phytosanitary measures, anti-dumping duty, etc.

Figure 3.

Source: World Integrated Trade Solution (WITS).

Apart from tariff and non-tariff measures, lack of trust and political will, long sensitive list (list of products not including tariff concession), and absence of proper implementation of policies and infrastructure have slowed the progress of this region toward deeper regional integration. Effective implementation of policies to promote intra-regional trade has never really taken place in this region because of trust deficit among the members, especially between Pakistan and India (Behera, 2009). But if proper conditions can be created, then enormous potential exists for greater intra-regional trade in ways that would benefit individual members and South Asia as a whole. This would significantly affect the growth rate of this region.

Theoretical Framework

Hungarian economist Bela Balassa defined economic integration both as a process and as a state of affairs. Described as a process, economic integration requires eliminating discrimination between economic units that belong to different nations and as a state of affairs; it requires elimination of different forms of discrimination between the country nations (Balassa, 1961). Balassa described five stages of economic integration in his original work. The 1st stage is “Free Trade Area” (FTA) where products are allowed to cross borders without any import tariffs or any other barrier between the countries involved. The 2nd stage obviously involves more integration than the first stage by eliminating trade barriers internally and setting a common external tariff (CET). This stage is called “customs union” where countries behave like a single entity. The 3rd stage is common market which includes the previous two stages as well as factors of production such as labor and capital, which can also move freely. Economic integration is the 4th stage of integration where labor, capital, and products can move freely. It also requires common external policy and currency (such as the Euro). Some amount of sovereignty has to be sacrificed at this stage as it requires common fiscal, economic, social, and monetary policy (Balassa, 1961). EU is an economic union, but UK decided to opt out of the common currency policy in order to have control over currency policies. The 5th stage involves total/complete economic integration. However, Balassa’s stages now have been extended to include political union as a further development to the integration process.

Given the increased amount of regional trade agreements (RTAs) in recent years, the question of whether multilateralism should be promoted, or regionalism enhanced, has gained significant attention in international trade (Herz & Wagner, 2010). As opposed to regionalism, multilateralism is concerned with integration at the global level, that is, to enhance trade at global level. But regionalism, in general, debates about increasing trade among few neighboring countries. The argument for regionalism or RTA comes from gravity model of trade which asserts that the lesser the distance, the greater is the opportunity for trade. Countries form RTA because it reduces transaction costs as well as tariff barriers. RTA helps an economy to achieve economies of scale as demand does not remain constrained only by domestic market, therefore enabling the firms to operate under a larger market.

However, Viner (1950) demonstrated that trade agreement can be beneficial, but at the same time, it can be harmful for the society’s welfare. His analysis is based on two effects of trade: one is trade creation effect and the other is trade diversion effect. A potential upside of trade agreement lies in the fact that a country may import goods and services more cheaply from the partner country which can replace the domestic production that is much more expensive and this is known as trade creation effect. On the other hand, trade diversion takes place when production is shifted from external producers who are efficient, to the members who are inefficient. The overall effect depends on which one is stronger. If trade creation effect is stronger, it will stimulate welfare, but if trade diversion is stronger, then it is more likely to be harmful to the society’s welfare (Raihan, 2012). Viner’s approach, known as static analysis, has been criticized on the ground that this welfare approach cannot fully capture the effects of integration on welfare and can be referred to as “old regionalism,” whereas new regionalism, known as dynamic analysis, is concerned with “large scale economies, technological change, as well as the impact of integration on market structure and competition, productivity growth, risk and uncertainty, and investment activity” (Balassa, 1961).

One of the strongest proponents of multilateralism, Jagdish Bhagwati, in his analysis, noted that proliferation of preferential/regional trading arrangements creates a spaghetti bowl phenomenon where countries extend their preferences to countries with different trade agreements and thereby “clutters up trade with discrimination” (Bhagwati, 1995). This phenomenon is known as “spaghetti bowl syndrome of regionalism” and it was later referred to as “noodle bowl effect” in the context of Asia by Baldwin (2009).

Empirical Literature Review

The growth effect of regional integration has received considerable critical attention in the empirical literature. EU and ASEAN are the two trade blocs that have (almost) fully utilized regional integration to their benefits and have shown tremendous economic growth by removing trade barriers among the member countries. Therefore, this study is divided into three parts. The first part refers to the economic integration–growth nexus in European Union (EU); the second part deals with the ASEAN economies, and finally, the third part deals with the above-mentioned relationship in South Asian context.

Growth–Integration Nexus and European Union

The Maastricht treaty was established by EU in 1992, and with the introduction of the Euro, it became a monetary union in 1999. Since then, it has become a unique community of European countries tied through both political and economic partnership. European Union is a powerful trade block, and it has contributed greatly to the economies of Europe over the years. Trade deals with 70 other countries in addition to a single market; tariff-free trade within the union, human rights protection, right to receive emergency health cards in any member state, common EU greenhouse gas targets, and last but not the least, freedom of free movement between member states have made EU a role model for the whole world. The benefit to a country in joining EU was examined by Campos et al. (2019), and they found that it was positive and large. By using synthetic control method, they came to the conclusion that if European countries were not to integrate, per capita income of a European country would have been 10 percent lower in the first 10 years of joining EU. Their analysis showed that only Greece experienced lower growth and productivity after joining EU.

Badinger (2005) adopted a different measure of economic integration from previous studies by combining both European integration and General Agreement on Tariffs and Trade (GATT) liberalization. He then tested the growth effects of this measure via time series and dynamic panel approaches. He found that the average yearly growth rate of GDP for EU over the period 1950–2000 would have been 0.4 percentage points lower if there was no integration.

Despite the huge benefits of economic integration, UK decided to leave EU in 2016 over sovereignty and immigration issues. This sparked a heated debate among researchers, policymakers, and other stakeholders, with some arguing that it will be economically damaging for EU, and others have defended the move saying that the benefits of leaving outweigh the benefits of staying; for example, using quantitative trade model, Dhingra et al. (2017) found that living standard in the UK has reduced by 1.3% due to BREXIT and that is an optimistic scenario. From a pessimistic viewpoint, this loss amounts to 2.7%. A reduced-form approach also reveals that per capita income will be reduced by 6.3%–9.4% if Britain leaves EU. UK might also experience a decline in trade with EU members, and this reduced trade will cost the UK economy a heavy sum.

Trade among the EU members remarkably increased when trade restrictions were withdrawn (Van Reenen, 2016). Therefore, leaving the EU would result in higher trade costs and Britain will have to renegotiate all of its trade deals. Crafts (2016) asserts that UK’s GDP per capita would have been 8.6%–10.6% lower if UK were to opt out of the European Union. On the other hand, Bruno et al. (2016) found that Britain will lose about 22% of its FDI flows due to BREXIT.

Growth–Integration Nexus and ASEAN

The ASEAN has been hailed as the next European Union, due to faster implementation of its policies for boosting regional growth. Now it is considered one of the finest integrated regions in the world. Backward countries in ASEAN such as Vietnam, Cambodia, Laos, and Myanmar had shown tremendous growth after joining ASEAN block where trade barriers have been greatly reduced. The effect of AFTA came into effect in 1992 and since then, researchers and policymakers have become increasingly interested on the impact of this agreement on output expansion and welfare effects of ASEAN economies.

Chaipan et al. (2006), for example, found that AFTA would bring about 8.7% increase in real investment for Thailand and it will significantly improve the investment scenario of all the countries involved. For Indonesia, Gumilang et al. (2011) found that AFTA could increase output by 0.47%, whereas bilateral agreement can only increase output by 0.11%. Nguyen and Ezaki (2005), on the other hand, constructed a computable general equilibrium (CGE) model to examine how regional integration affects growth, income distribution, and poverty for Vietnam. Although it was found that regional integration improves income distribution of Vietnam, there can be adverse effects of economic integration on the economy. This is because tariff contributes largely in fiscal revenue generated by Vietnam, but greater integration means less tariff and, hence, less government revenue. This in turn implies less consumption and lower level of GDP. Their recommendation was to implement complementary tax policies to offset this decline in real GDP. CGE modelling approach is well recognized in empirical studies, but a major criticism about the approach is that critical relationships often do not have empirical justification in this approach; rather, they are purely theoretical. Moreover, the results can be sensitive to measurement error in case of variables that are hard to measure (Schiff & Winters, 2003).

Preepremmote et al. (2018) examined the impact of economic integration on growth for every ASEAN economy. For economic integration, they used ASEAN economic integration index which comprises of single market, economic homogeneity and economic symmetry. The results suggested that growth is significantly increased due to economic integration for Indonesia, Malaysia, Philippines, Singapore, and Thailand. They also found that changes in economic integration have a greater role to play on growth than that of the degree of economic integration and, hence, continuous integration should be taken into account in policy formation.

In another analysis, Bong and Premaratne (2018) reported that regional integration indeed is a significant driver of economic growth in Southeast Asia. Their analysis revealed that a country can have immense economic benefits when it is integrated at regional level. Moreover, trade openness and capital can also determine economic growth. In addition, corruption hinders growth since it adversely affects stability and human and physical capital investment. In line with Barro (1996) and Tahir and Khan (2014), they found that human capital exerts a significant yet negative effect on economic growth.

Growth–Integration Nexus and South Asia

Effort to integrate SAARC economies is a great challenge because of political tensions between countries. Moreover, issues of nationalism and higher level of protectionism hold back the integration progress of this region. Numerous attempts therefore have been made to grasp the relevance of greater regional economic integration for economic growth in South Asia.

Whether SAFTA would lead to an increase in intra-regional trade has been subject to controversy since the beginning of this agreement; for example, using CGE model, Krueger et al. (2004) predicted that the effects of SAFTA are more likely to be small. Several reasons were highlighted for such outcome, and one of them was that if member countries do not grant preferential status to other member’s products, then they are not likely to gain large benefits from SAFTA. They said that the countries can achieve growth through bilateral trade rather than the regional trade because of political tension between India and Pakistan. This prediction was confirmed later by Taguchi (2018) who applied a gravity trade model analysis and found that trade creation effects were not identified in SAFTA, whereas the effects were verified under bilateral agreements. In his seminal work, Raihan (2012) applied Global Trade Analysis Project (GTAP) general equilibrium model and concluded that if SAFTA is to be effective, then along with trade liberalization, it is necessary to develop trade facilitation and infrastructure policies as well.

On the other hand, Kumar and Singh (2009) focused on the constraints faced by South Asian countries to integrate regionally. Examining the investment pattern of South Asia’s trade, they recommended that India should play a bigger role in order to boost regional integration in South Asia. They further state that India cannot do it alone; other neighboring countries also have to respond appropriately to the initiatives taken by Indian government, especially Pakistan and Bangladesh. However, their analysis was purely qualitative.

Moinuddin (2013) used empirical models to assess the trade effects of SAFTA and the factors that determine this agreement. He augmented the gravity model by including several variables and suggested that countries can benefit regionally and globally if trade restrictions are reduced. Using the same approach as that of Moinuddin (2013), Peiris et al. (2017) found that trade creation effects were significant prior to SAFTA and after SAFTA, but no evidence of trade diversion effects was found in their analysis. They also found that SAFTA is not likely to minimize trade with non-member countries as proponents of multilateralism (cooperation among three or more members) argue.

Trade facilitation refers to minimization of trade transaction costs that occur during the exports and imports. Perera et al. (2017) argued that trade facilitation can improve regional trade and thereby strengthen regional integration in South Asia. They believe that increasing the export competitiveness and promoting imports depend on timely deliberation of products. To reduce poverty lags among the countries, they proposed that South Asia should stimulate growth through trade.

Data and Methodology

This section focuses on model specification, data and variable description, and also on estimation. The empirical analysis has used software packages such as EVIEWS, STATA, and GAUSS. The countries included in this study are Afghanistan, Bangladesh, Bhutan, India, Nepal, Pakistan, and Sri Lanka.

Model Specification

As far as economic growth is concerned, the standard growth regression model must include both labor and capital as they are basic determinants of growth. Therefore, we start with a Cobb–Douglas production function which has constant returns to scale:

Y_{t} = A_{t} K_{t}^{α} L_{t}^{β}

(1)

Where Y stands for output/GDP, A denotes technological progress, K denotes phical capital, and L refers to labor force. Taking logarithm of Equation (1), we find:

\ln Y_{t} = Y^{'} + α \ln K_{t} + β \ln L_{t}

(2)

Where $Y^{'}$ denotes the growth rate of technology and equals ln(A_t). Now, we extend this basic production function to include the impact of regional integration on growth. The standard cross-country growth regression therefore can be written as:

Y_{i, t} = α + β {reg}_{i, t} + φ X_{i, t} + ε_{i, t}

(3)

Where Y denotes gross domestic product (GDP) for each country, reg refers to intra-regional trade for each country, X denotes the vector of control variables that include democracy, human capital, and other basic determinants of growth such as labor and capital, i refers to the individual countries, t denotes the study years, and ε refers to the error term.

Therefore, the econometric model of the above equation can be now written as follows:

\ln {gdp}_{i, t} = a_{0} + a_{1} \ln {reg}_{i, t} + a_{2} \ln {dem}_{i, t} + a_{3} \ln {edu}_{i, t} + a_{4} \ln {lab}_{i, t} + a_{5} \ln {cap}_{i, t} + ε_{i, t}

(4)

Where gdp stands for gross domestic product (constant 2010 US$), reg stands for regional economic integration, dem stands for democracy, edu stands for gross tertiary school enrolment (in percentage), lab indicates total labor force participation rate (in percentage term), and cap stands for gross capital formation.

Description of the Data

The data for seven South Asian countries over the 1980–2015 period has been considered for the study. The study period is chosen based on the availability of data. A linear interpolation has been performed in order to handle the missing values. Moreover, all variables have been transformed into logged form to normalize and reduce heteroscedasticity. A summary of data description and data source is given in Table 1.

Table 1.

Description of the Data and Data Source.

Variables	Abbreviation	Description	Source	Reference
Economic growth	gdp	Gross domestic product (constant 2010 US$)	UNSTAT	Sulaiman et al. (2015)
Regional economic integration	reg	Intra-regional trade share (%)	ADB	Beckfield (2006)
Democracy	dem	Polity 2	Polity IV database	Coppedge et al. (2018)
Human capital	tedu	Tertiary school enrolment (% gross)	WDI	Sulaiman et al. (2015)
Labor	lab	Labor force participation rate (%)	WDI	Haque et al. (2019)
Physical capital	cap	Gross capital formation	UNSTAT	Yusuf (2014)

Source: The author.

The dependent variable is economic growth, as measured by gross domestic product (constant 2010 US$). The data for this variable are obtained from the national accounts section of the United Nations statistics division. Apart from regional economic integration, several explanatory variables have also been added. The data sources of the independent variables along with their expected relationship with growth are described as follows:

Regional economic integration—following Beckfield (2006), economic integration or regional economic integration is measured by intra-regional trade share (in percentage) collected from Asian Development Bank (ADB) database. This refers to the percentage of intra-regional trade compared to total trade of the region. If the trade share is higher, it means members are largely dependent on regional trade. The expected relationship with growth is positive.

Democracy—democracy is measured by Polity2 from Polity IV database (Marshall et al., 2007). Among the existing measures of democracy, this index has been used widely (Coppedge et al., 2018). Polity2 is the modified version of Polity and it takes value from -10 (total autocracy) to +10 (total democracy). Based on empirical literatures, there can be three types of relationship between democracy and growth: positive (McCord, 1965), negative (De Schweinitz, 1964; Rao, 1984), or no relation at all (Bougharriou et al., 2019; Przeworski et al., 2000).

Human capital—the data for human capital, as proxied by tertiary school enrolment (percentage gross), are collected from World Development Indicator (WDI). The expected relationship is positive. However, Hanushek and Woessmann (2015) point out that human capital can be properly measured by what is actually learned at institutions, not by the length of study period. Hence, it is the quality of education that matters, not the quantity.

Labor—the data for labor proxied by labor force participation rate (in percentage) are obtained from WDI. This variable has been added in order to avoid the problem of omitted variable bias on the basis of Cobb–Douglas production function. The expected relationship with growth is positive.

Physical capital—physical capital is measured by gross capital formation following Akinola and Adeleke (2013), Kanayo (2013), and Yusuf (2014). The data for this variable have been extracted from the national accounts section of the United Nations statistics division. The expected sign of the coefficient for physical capital is also positive based on traditional growth theory.

Table 2.

Descriptive Statistics.

	cap	gdp	reg	dem	tedu	lab
Mean	22.57766	23.97729	2.579127	1.558581	4.139734	22.57766
Median	22.53970	24.07305	2.867490	2.772590	1.577230	4.090470
Maximum	27.44470	28.46390	5.183040	2.995730	3.291220	4.484300
Minimum	17.90980	18.82620	−4.002450	0.000000	−0.338751	3.921380
Std. Dev.	2.061464	2.141919	1.931196	0.926754	0.759372	0.168715
Skewness	0.149243	−0.253659	−1.327780	−1.244370	−0.028891	0.880361
Kurtosis	2.824434	2.856656	4.926467	3.431055	2.560647	2.617498
Jarque–Bera	1.259128	2.918142	113.0144	66.98620	2.061880	34.08773
Jarque–Bera Probability	0.532824	0.232452	0.000000	0.000000	0.356672	0.000000
Observations	252	252	252	252	252	252
Cross-sections	7	7	7	7	7	7

Source: The author.

Descriptive statistics are also reported in Table 2. From Jarque–Bera probability test, it is clear that three out of six variables are normally distributed. But violation of normality is not a major problem since we have sufficient observations (Ghasemi & Zahediasl, 2012).

Methodology

Cross-sectional Dependence Test

Cross-sectional correlation/dependence has to do with impact of shocks in one country on another country when both countries belong to the same region (South Asia in this case). It can be checked via several tests such as Breusch–Pagan (1980) LM, Pesaran (2004) scaled LM, Pesaran (2004) CD test, and a more recent test developed by Baltagi et al. (2012).

When N refers to fixed and T→∞, appropriate test is Breusch–Pagan (1980). The test statistic can be written as

LM = \sum_{i = 1}^{N - 1} \sum_{j = i + 1}^{N} T_{i} ̂_{i j}^{2} \to^{2} \frac{N (N - 1)}{2}

(5)

Here ρ̂_ij denotes correlation coefficient derived from each residual. However, this test cannot be applied when N tends to be infinite. Hence, Pesaran (2004) proposed a test which is applicable for infinite T and N, and it is also based on pairwise correlation coefficient ρ̂_ij:

{CD}_{scaled LM} = \sqrt{\frac{1}{N (N - 1)} \sum_{i = 1}^{N - 1} \sum_{j = i + 1}^{N} (T_{i}} ρ ̂_{i j}^{2} -^{} 1) \to N (0, 1)

(6)

But for N > T, Pesaran (2004) suggested a different test

CD = \sqrt{\frac{2 T}{N (N - 1)}} \sum_{i = 1}^{N - 1} \sum_{j = i + 1}^{N} ̂_{i j}

(7)

A more recent test is bias-corrected scaled LM test advocated by Baltagi et al. (2012). The test statistic is given by

{LM}_{BC} = \sqrt{\frac{1}{N (N - 1)}} \sum_{i = 1}^{N - 1} \sum_{j = i + 1}^{N} (T {\hat{}}_{i j}^{2} - 1) - \frac{N}{2 (T - 1)} \to N (0, 1)

(8)

Panel Unit Root Test

Cross-sectional correlation or in other words cross-sectional dependence means the residuals of entities or panels (e.g., countries in the panel data) are significantly correlated across entities. When residuals are correlated across the cross-sections, it simply means that shocks to one of the entities have impact on one or more of the others. Two generations of unit root tests can be distinguished depending on whether they allow for correlation across residuals of panel units or not (Hurlin & Mignon, 2007). First-generation tests do not allow cross-sections to be dependent while second generation tests do.

A second-generation unit root test called Pesaran (2007) unit root test has been applied in this study. This test is augmented version of basic Dicky–Fuller (CADF) regression, which is given by

Δ y_{i t} = a_{i} + b_{i} y_{i, t - 1} + c_{i} {\bar{y}}_{t - 1} + d_{i} Δ {\bar{y}}_{t}

(9)

Here, y̅_t–1 represents lagged level form of the mean value and $Δ {\bar{y}}_{t}$ also denotes mean but in first differenced form. Pesaran’s test is augmented version of Im, Pesaran and Shin test and is written as

CIPS (N, T) = N^{- 1} \sum_{i = 1}^{N} t_{i} (N, T)

(10)

Panel Cointegration Test

The next step following the unit root tests involves checking the long-run relationship among the variables, that is, whether they move together in the long run or not. Here we apply Westerlund (2007), which allows cross-sections to be dependent. Westerlund (2007) proposes four error correction-based test statistics. Two of them are “group mean” and other two are “pooled mean” estimation. Apart from allowing cross-sections to be dependent, another advantage of Westerlund (2007) is that it is also applicable in the heterogeneous panel. The model suggested by Westerlund (2007) is as follows:

Δ y_{i, t} = δ_{i}^{´} d_{t} + α_{i} y_{i, t - 1} + λ_{i}^{´} x_{i, t - 1} \sum_{j = 1}^{p_{i}} α_{i, j} Δ y_{i, t - j} + \sum_{j = 0}^{p_{i}} ϒ_{i, j} Δ x_{i, t - j} + μ_{i t}

(11)

Where α_i refers to the error correction term. Westerlund (2007) tests whether this term is different from 0 or not; for example, if this term equals 0 that would indicate no cointegration, but if it is less than 0, then it means there is cointegration. The alternative hypothesis for the first two test statistics (group mean tests) is that at least one unit is cointegrated. The other two test statistics, pooled panel tests, has the alternative that panel is cointegrated as a whole. Both the group mean and pooled panel tests just differ in their structure of alternative hypothesis, but the null hypothesis is the same.

Long-run Estimation

When evidence of cointegration is confirmed, the long-run coefficients are then estimated via a new econometric technique such as continuously-updated and bias corrected (CupBC), suggested by Bai et al. (2009). This estimator is robust when entities are cross-sectionally correlated. For estimating slope coefficients and latent common trend, Bai et al. (2009) adopted two procedures: CupBC and continuously updated and fully modified (CupFM) estimators. These estimators are constructed in such a way that they can take care of endogeneity and correlation bias. The two estimators differ in terms of estimating the bias.

In contrast to CUP estimation, the fully modified least squares (FM-OLS) of Hansen and Phillips (1990) and dynamic ordinary least squares (DOLS) of Saikkonen (1992) and Stock and Watson (1993) estimators cannot account for cross-sectional correlation. Therefore, if the countries are located in the same region and if a shock in one country affects another, then estimators other than FM-OLS and DOLS should be applied. The model developed by Bai et al. (2009) comes from the following equation:

y_{i, t} = x_{i, t}^{´} β + λ_{i}^{´} F_{t} + μ_{i, t}

(12)

where the dependent variable is a scalar, x_it consists of non-stationary regressors, and F_t denotes a vector of latent common factors.

The estimators can be applied to mixed I(1)/I(0) factors and mixed I(1)/I(0) regressors. The CUP estimators minimize the following function:

CLS (β, F) = \sum_{i = 1}^{N} \sum_{t = 1}^{T} \frac{1}{N T^{2}} {(y_{i t} - x ’}_{i t} β + λ_{i}^{´} F_{t})^{2}

(13)

Hence, the CUP estimator for (β, F) can be defined as

(β_{CUP}^{^}, F_{CUP}^{^}) = {argmin}_{β, F} CLS (β, F)

The next two equations represent the solution to (β̂_CUP, F̂_CUP)

B^{̂} = {(\sum_{i = 1}^{N} x_{i}^{´} M_{F ̂} x_{i}^{´})}^{- 1} \sum_{i = 1}^{N} x_{i}^{´} M_{F ̂} y_{i}

(14)

F^{̂} V_{N T} = [\frac{1}{N T^{2}} \sum_{i = 1}^{N} y_{i} - x_{i} \hat{β})^{'}] F^{̂}

(15)

Both the equations are needed while solving for β̂ and F̂ iteratively. An estimator of Λ is described as

\hat{Λ} = T^{- 2} {\hat{F}}^{´} (Y - X \hat{β}) .

Even though β_CUP is consistent, both serial correlation and endogeneity together generate an asymptotic bias. To correct that asymptotic bias, Bai et al. (2009) considered CupBC and CupFM estimators. Bias is corrected by β̂_CupBC at the final stage, whereas it is corrected at each stage by β̂_CupFM. This study implements only CupBC along with the DOLS to see whether ignoring cross=sectional dependence has any implications or not.

Causality Test

After attaining long-run relationship, causality test helps us to see whether there is any unidirectional or bidirectional causality. Here we employ Dumitrescu and Hurlin panel causality test (2012). Their panel statistics have the ability to increase the power of Granger non-causality when both time and cross-sectional units are small. When cross-sectional dependence exists, this test produces unbiased results (Anoruo & Elike, 2015).

We consider the following linear model:

y_{i, t} = α_{i} + \sum_{k}^{K = 1} ϒ_{i}^{(k)} y_{i, t - k} + \sum_{k = 1}^{K} β_{i}^{(k)} x_{i, t - k} + ε_{i, t}

(16)

where K denotes optimal lag length. The individual effects denoted by α_i are fixed in time dimensions and for all cross-sectional units, there is identical lag order. Further, ϒ_i^(k) and β_i^(k) differ across groups.

Individual Wald statistics in average are given by

W_{N, T}^{H_{n c}} = 1 / N \sum_{i = 1}^{N} W_{i, T}

(17)

The null hypothesis indicates here that there is no causal relationship. However, there are two subgroups for alternative hypothesis: one states that x to y have causal relationship and another one identifies no relationship. According to Dumitrescu and Hurlin (2012), all coefficients are heterogeneous across cross-sections.

Results and Discussions

If cross-sectional correlation is not taken into account, this will have some detrimental consequences (Baltagi, 2015). Therefore, we first start with cross-sectional dependence tests since shocks in one country might affect another country in South Asia. Table 3 reports CSD results of the respective variables in our study. All the tests are able to reject the null hypothesis of no cross-sectional dependency at conventional level of significance for all the variables. This implies that there is cross-sectional dependence, and we must incorporate it while conducting further analysis.

Since the variables in this study are correlated cross-sectionally, tests that assume cross-sections to be independent are not applicable anymore. This study, therefore, makes use of a second-generation test developed by Pesaran (2007). Table 4 gives the result of this CIPS test.

Both “Intercept with no trend” and “Intercept and trend” specifications have been considered for the analysis. The data have unit root at non-stationary level while considering the assumption of intercept only, but the same is also true for intercept-trend model. But when first difference is taken, the data appear to have no unit root with either intercept or intercept-trend. Hence, it is reasonable to conclude that all variables are integrated of order one or I(1) here.

Since all the variables are integrated of the same order, we can now proceed to estimate long-run relationship among the variables. Here we use Westerlund (2007) Panel Cointegration test which is second-generation cointegration test. The result of Westerlund (2007) cointegration test based on 500 bootstrap replications is reported in Table 5.

Table 3.

Test Results for Cross-sectional Dependence of the Variables.

Variables	Breusch Pagan LM	Pesaran ScaledLM	Bias-Corrected Scaled LM	Pesaran CD
gdp	570.1892^α	84.74173^α	84.64173^α	22.80980^α
reg	406.4931^α	59.48287^α	59.38287^α	8.695502^α
dem	132.3306^α	17.17869^α	17.07869^α	5.700439^α
tedu	569.8559^α	84.22740^α	84.12740^α	23.65500^α
lab	224.6628^α	31.42586^α	31.32586^α	−2.124329^α
cap	494.8132^α	73.11096^α	73.01096^α	20.54350^α

Source: The author.

Note: ^αrejects the null of no cross-sectional correlation or dependence at 1% level.

Table 4.

Panel Unit Root Test Results.

VariablesName	CIPS Test
VariablesName	Intercept	Intercept + Trend	Intercept		Intercept + Trend
gdp	−1.481	−2.774	∆gdp	−5.806^α	−6.004^α
reg	−2.049	−2.899	∆reg	−4.198^α	−4.337^α
dem	−2.366	−2.654	∆dem	−5.351^α	−5.362^α
tedu	−1.204	−2.407	∆tedu	−4.451^α	−4.382^α
lab	−0.703	−1.845	∆lab	−2.728^α	−2.863^α
cap	−2.389	−2.995	∆cap	−5.561^α	−5.791^α

Source: The author.

Notes: ^α and ^β Reject null hypothesis of non-stationarity at significance level 1% and 5%, respectively. Maximum lag order is set at 1.

Table 5.

Westerlund (2007) Cointegration Test Result.

Statistic	Value	p-Value	Robust p-Value
Deterministic specification: constant
Gt	−2.398	0.755	0.286
Ga	−9.456	0.962	0.026
Pt	−5.872	0.646	0.300
Pa	−8.879	0.780	0.074
Deterministic specification: constant and trend
Gt	−3.405	0.131	0.042
Ga	−5.645	1.000	0.570
Pt	−11.961	0.000	0.012
Pa	−9.088	0.973	0.078

Source: The author.

Notes: The rule 2(T/100)^2/9 has been used to determine lags and leads. The width of the Bartlett kernel window is set at 3. Bootstrapped p-values are obtained using 500 replications.

Considering constant only, when we use robust p-values and thereby making allowance for cross-sectional dependence, the null of no cointegration is rejected for 2 out of 4 statistics. When deterministic specification is constant and trend, 3 out of 4 statistics can reject null of no cointegration. This suggests that variables are cointegrated in the long run.

Now, to estimate long-run coefficients, we apply CupBC estimator which performs well for the small samples and is considered asymptotically normal and unbiased. We have included the results of DOLS for the sake of comparison. Table 6 shows the results obtained from CupBC estimation along with the results of DOLS.

Our main variable of interest is regional economic integration (reg). The CupBC estimation reveals that regional economic integration can increase economic growth significantly. To be more specific, if there is a 1% increase in economic integration, it will push real GDP of South Asian countries up by 3.5%. This result corroborates the findings of the previous studies in this field including Bong and Premaratne (2018) and Campos et al. (2019). The result suggests that creating a well-integrated and sustainable region should be at the forefront of economic policy. The economic issues must delink from the political issues in South Asia. Indeed, South Asia as a whole can develop at a faster rate if the member nations start working together to emerge as an integrated economy. DOLS reveals almost similar result although coefficient is slightly higher.

The impact of democracy on economic growth is negative and statistically significant at 10% level, as found by CupBC. This result is supported by Sirowy and Inkeles (1990) who stated that developing countries cannot achieve democracy and economic development together since it is considered a luxury for them. This also lends support to the authoritarian model, advocated by De Schweinitz (1964) and Rao (1984), which states that autocratic countries can handle internal conflicts very well and therefore it can be considered as growth enhancing in some cases. On the other hand, democratic countries are more concentrated on maintaining internal order to win elections but not on growth; hence, they focus more on the development for the short run and ignore the development that is needed for the long term (Huntington, 1970). The argument for negative effect of growth can also be found in Olson (1982) who believed that democracy is negatively affected when there are too many organizations pursuing their own interest. Moreover, the principle reason why democracy is adopted is to have equality, but Persson and Tabellini (1994) demonstrated that democracy is harmful for growth due to high level of redistribution. The effect turns out to be positive but insignificant once cross-sectional dependence is ignored.

Table 6.

Results from Different Models.

Variables	CupBC	DOLS
reg	0.035036418^α	0.036734^α
dem	−0.028881765^λ	0.017122
tedu	−0.15209644^α	0.111735^β
lab	−0.0011404202	−0.214227
cap	0.033010753^λ	0.655633^α

Source: The author.

Notes: ^α, ^β, and ^λ Refer to significance level at 1%, 5% and 10% level, respectively. Constant and trend has been taken as the deterministic specification for both the models. Lags and leads are fixed at 1 for DOLS.

Strong negative effect of human capital on growth is found via CupBC estimation. Although this result does not match many of the earlier studies, it is fully in line with Barro (1996), Tahir and Khan (2014) and Bong and Premaratne (2018). This negative effect of human capital on growth is largely due to heterogeneity in educational spending across the countries studied (Pelinescu, 2015). In South Asia, human capital development gap by region is the second largest after sub-Saharan Africa, as indicated by Global human capital report (2017). In developing human capital, Sri Lanka dominated the ranking in this region, but it is nowhere in the top 50; rather, it stood at 70th position. Other South Asian countries such as Nepal stood at 98th, India at 103rd, Bangladesh at 111th, and Pakistan was at the 125th position out of 130 countries. Moreover, majority of population with less than primary education lives in South Asia, as demonstrated by World Bank report (2018). The report also mentioned that South Asia still has 322 million uneducated adults. Hence, apart from failing to boost intra-regional trade, this region also had very little success in developing human capital. Another reason why negative effect of human capital was observed in this study could be that it is the quality of the education that matters, not the quantity as we have measured here by attainment of specific educational level (Hanushek & Woessmann, 2015). The effect of human capital on growth, however, becomes positive once cross-sectional dependence is ignored.

In line with Galor and Zang (1997), the result did not show any significant increase in economic growth due to increased labor force participation. The result is robust against both CupBC and DOLS approaches. This might be explained by South Asia’s labor market which is characterized by stagnant growth in employment and lower participation of women across all sectors. In addition to that, the rate of structural transformation across the countries of this region has always been slow. From CupBC estimation, it is also clear that physical capital significantly and positively influences economic growth. In particular, if physical capital increases by 1%, this will lead to a 3.3% increase in economic growth. Compared to CupBC, DOLS gives a much higher positive value for the physical capital, as it ignores the impact of cross-sectional dependence.

There is a tendency that some economic variables have common movements, that is, they can have effects on each other. Therefore, determining the direction of relationship is extremely necessary, especially for the policymakers. Dumitrescu and Hurlin (2012) panel causality test has been implemented here to ascertain the direction of movement, either unidirectional or bidirectional, between the variables. Schwarz information criterion (SIC) has been used as optimal lag criterion. The results obtained from Dumitrescu and Hurlin (2012) test are set out in Table 7.

Table 7.

Dumitrescu and Hurlin (2012) Panel Causality Test Result.

Direction of Causality	W-stat	Z-bar Stat	Prob.
gdp→reg	1.5324	0.7763	0.6850
reg→gdp	0.4460	−1.0346	0.2680
gdp→dem	0.9282	−0.2308	0.8990
dem→gdp	26.1571	0.9343	0.1310
gdp→tedu	5.1291	6.7716	0.0610
tedu→gdp	0.6440	0.7045	0.5730
gdp→ lab	4.4025	2.5780	0.2550
lab→gdp	0.7253	0.5690	0.5970
gdp→cap	4.5652	5.8316	0.3120
cap→gdp	41.8253	2.4769	0.0320
reg→dem	8.39e+05	8.26e+04	0.0000
dem→reg	409.7109	38.6950	0.0010
reg→tedu	3.8193	4. 5883	0.0070
tedu→reg	176.8067	15.7657	0.0000
reg→lab	29.6361	1.2768	0.1380
lab→reg	0.6957	−0.6183	0.7070
reg→cap	1.6401	0.9558	0.4310
cap→reg	1.7389	1.1206	0.5480
dem→tedu	2.3531	2.1443	0.0500
tedu→dem	2.08e+06	2.05e+05	0.0010
lab→dem	1.22e+04	1201.5873	0.0000
dem→lab	37.3269	2.0340	0.0370
cap→dem	0.8896	0.2951	0.8770
dem→cap	0.3526	1.1903	0.2790
tedu→lab	27.6547	1.0818	0.1670
lab→tedu	2.0781	1.6860	0.1860
tedu→cap	1.8315	1.2749	0.5000
cap→tedu	4.7646	6.1639	0.0290
cap→labor	3.5896	1.6483	0.4560
lab→cap	1.3543	0.4794	0.6580

Source: The author.

The result reveals that both the economic integration and democracy variable have causal relationship with each other. It means that past and present values of economic integration affect the present value of democracy and vice versa. The same result can also be inferred for regional integration and human capital, democracy and human capital, and democracy and labor. It can also be observed from Table 7 that there is unidirectional or one-way causality running from economic growth to human capital, physical capital to economic growth and to human capital.

Conclusion and Policy Recommendations

With India leading the way, South Asia is growing faster than any other region in the world, but compared to others, member states trade less with each other. In an economically suffering region from almost all perspectives, intra-regional trade fluctuates around 4%–5% of their total trade. This is disturbing since regional integration is seen as a strong driver in meeting SDG targets and it can have massive impacts on economies. This article tries to address this issue by examining how economic integration plays its role in fostering growth for South Asia and to what extent. This study considers the case of cross-sectional dependence and uses a unique CUP estimation procedure by Bai et al. (2009) which has been applied only a few times in the studies of growth–integration nexus.

The study finds that economic growth of SAARC or South Asian economies can be well explained by economic integration. This result confirms the previous studies. Since South Asia is a least integrated region and economic integration has a huge positive effect on growth, policymakers and relevant authority must put business ahead of politics. Not fully utilizing SAARC has some serious negative trade and economic consequences for the economies involved and therefore, greater cooperation between them is needed. However, this in turn requires much deeper integration than mere removal of tariffs or non-tariff barriers. Apart from developing stronger trade relationship, proper implementation of SAARC can reduce extreme and absolute poverty, which still exists in many parts of this region. The member countries have high level of mistrust among them, so much so that solving the integration conundrum seems counterproductive to their political agenda. Therefore, it is necessary to strengthen the political cooperation by initiating different bilateral as well as multilateral agreements under the forum among the countries. Promoting SAARC can also boost up cross-border movement of goods and services. Therefore, countries should reduce sensitive lists and make positive list-based approach. Being the largest country among the nations, India should play a much bigger role in utilizing SAARC to the fullest. Higher bilateral trade costs should also be reduced. There are several practical and other limitations to implement SAARC; hence, reforms within this forum are much needed, such as relaxation of rules of origin. Proper atmosphere must be created in order to have productive dialogues between counties, especially between India and Pakistan. Closer and deeper regional union is mandatory which may bring major changes to the entire political and economic structure of this region.

From the bidirectional causality between human capital and economic integration and also between human capital and democracy, it is clear that if left unchecked, gaps in human capital development may compromise regional integration and political stability since the relevant authorities will not have the capacity to act in the best interests of this region. Therefore, South Asian economies should focus more on capacity building and skills development rather than just increasing the number of people in schools and universities.

Results from CUP estimation suggest that both democracy and human capital negatively affect growth. The reason why this might have happened is subject to further analysis and investigation. Since this study is limited by the lack of latest data for several variables, it is recommended that future research be undertaken by including various factors such as human capital measurement (e.g., quality of education) and institutional quality. Further works may also attempt to identify the heterogeneity of RTA effects across the member countries.

Footnotes

Acknowledgments

The author would like to thank Dr Avik Sinha and the anonymous reviewers for their valuable comments and suggestions to improve the manuscript.

Declaration of Conflicting Interests

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

References

Akinola

G. W.

Adeleke

(2013). Savings, gross capital formation and economic growth Nexus in Nigeria (1975–2008). IOSR Journal of Economics and Finance (IOSRJEF) , 1(2), 19–25.

Anoruo

Elike

(2015). Human capital-economic growth nexus in Africa: Heterogeneous panel causality approach. International Journal of Economics and Financial Issues , 5(4), 1017–1023.

Badinger

(2005). Growth effects of economic integration: Evidence from the EU member states. Review of World Economics , 141(1), 50–78.

Bai

Kao

(2009). Panel cointegration with global stochastic trends. Journal of Econometrics , 149(1), 82–99.

Balassa

(1961). Towards a theory of economic integration. Kyklos , 14(1), 1–17.

Baltagi

B. H.

(2015). The Oxford handbook of panel data . OUP Catalogue, Oxford University Press. ISBN 9780199940042.

Baltagi

B. H.

Feng

Kao

(2012). A Lagrange Multiplier test for cross-sectional dependence in a fixed effects panel data model. Journal of Econometrics , 170(1), 164–177.

Baldwin

(2009). The spoke trap: hub and spoke bilateralism in East Asia . Institute of International Studies. https://dx-doi-org-s.web.bisu.edu.cn/10.1093/acprof:oso/9780199235889.003.0003

Barro

R. J.

(1996). Determinants of economic growth: A cross-country empirical study (No. w5698). National Bureau of Economic Research.

10.

Beckfield

(2006). European integration and income inequality. American Sociological Review , 71(6), 964–985.

11.

Behera

N. C.

(2009). SAARC & beyond: Civil society and regional integration in South Asia. South Asia Centre for Policy Studies. In Rahman

S. H.

Khatri

Brunner

H. P.

(Eds.), Regional integration and economic development in South Asia . Edward Elgar Publishing.

12.

Berthelon

(2004). Growth effects of regional integration agreements . Documentos de Trabajo (Banco Central de Chile), (278), 1. https://si2.bcentral.cl/public/pdf/documentos-trabajo/pdf/dtbc278.pdf

13.

Bhagwati

(1995). US trade policy: The infatuation with FTAs (No. 726). Columbia University, Discussion Paper Series.

14.

Bong

Premaratne

(2018). Regional integration and economic growth in Southeast Asia. Global Business Review , 19(6), 1403–1415.

15.

Bougharriou

Benayed

Gabsi

F. B.

(2019). The democracy and economic growth nexus: Do FDI and government spending matter? Evidence from the Arab world. Economics: The OpenAccess. Open-Assessment E-Journal , 13(2019-19), 1–29. http://www.economics-ejournal.org/economics/journalarticles/2019-19

16.

Breusch

T. S.

Pagan

A. R.

(1980). The Lagrange multiplier test and its applications to model specification in econometrics. The Review of Economic Studies , 47(1), 239–253.

17.

Bruno

Campos

Estrin

Tian

(2016). Gravitating towards Europe: An econometric analysis of the FDI effects of EU membership (CEP Technical Paper). Brexit Analysis, 3. http://cep.lse.ac.uk/pubs/download/brexit03_technical_paper.pdf

18.

Campos

N. F.

Coricelli

Moretti

(2019). Institutional integration and economic growth in Europe. Journal of Monetary Economics , 103, 88–104.

19.

Coppedge

Gerring

Lindberg

S. I.

Skaaning

S. E.

Teorell

Altman

Knutsen

C. H.

(2018). V-Dem [Country-Year/Country-Date] Dataset v8 . Varieties of Democracy (V-Dem) Project.

20.

Crafts

(2016). The growth effects of EU membership for the UK: A review of the evidence (CAGE Working Paper, 280). University of Warwick.

21.

De Lombaerde

(Ed.). (2006). Assessment and measurement of regional integration . Routledge.

22.

De Melo

Montenegro

Panagariya

Mundial

(1992). Regional integration old and new (Vol. 985). Country Economics Department, World Bank.

23.

De Schweinitz

(1964). Industrialization and democracy: Economic necessities and political possibilities . Free Press of Glencoe.

24.

Dhingra

Huang

Ottaviano

Paulo Pessoa

Sampson

Van Reenen

(2017). The costs and benefits of leaving the EU: Trade effects. Economic Policy , 32(92), 651–705.

25.

Dumitrescu

E. I.

Hurlin

(2012). Testing for Granger non-causality in heterogeneous panels. Economic Modelling , 29(4), 1450–1460.

26.

Ehigiamusoe

K. U.

Lean

H. H.

(2018). Do economic and financial integration stimulate economic growth? A critical survey (Economics Discussion Papers, No 2018-51). Kiel Institute for the World Economy. http://www.economics-ejournal.org/economics/discussionpapers/2018-51/file

27.

Fosu

A. K.

(2009). Inequality and the impact of growth on poverty: Comparative evidence for sub-Saharan Africa. The Journal of Development Studies , 45(5), 726–745.

28.

Galor

Zang

(1997). Fertility, income distribution, and economic growth: Theory and crosscountry evidence. Japan and the World Economy , 9(2), 197–229.

29.

Ghasemi

Zahediasl

(2012). Normality tests for statistical analysis: A guide for non- statisticians. International Journal of Endocrinology and Metabolism , 10(2), 486.

30.

Gumilang

Mukhopadhyay

Thomassin

P. J.

(2011). Economic and environmental impacts of trade liberalization: The case of Indonesia. Economic Modelling , 28(3), 1030–1041.

31.

Haque

A. U.

Kibria

Selim

M. I.

Smrity

D. Y.

(2019). Labor force participation rate and economic growth: Observations for Bangladesh. International Journal of Economics and Financial Research , 5(9), 209–213.

32.

Hanushek

E. A.

Woessmann

(2015). The knowledge capital of nations: Education and the economics of growth . The MIT Press.

33.

Henrekson

Torstensson

(1997). Growth effects of European integration. European Economic Review , 41(8), 1537–1557.

34.

Herz

Wagner

(2010). Multilateralism versus regionalism!? (Discussion Paper No. 89). BGPE.

35.

Huntington

S. P.

(1970). Political order in changing societies. VRÜ Verfassung und Recht in Übersee , 3(2), 257–261.

36.

Hurlin

Mignon

(2007). Second generation panel unit root tests (Working Papers). HAL. https://halshs.archives-ouvertes.fr/halshs-00159842

37.

Iqbal

M. J.

(2006). SAARC: Origin, growth, potential and achievements. Pakistan Journal of History and Culture , 27(2), 127–140.

38.

Kanayo

(2013). The impact of human capital formation on economic growth in Nigeria. Journal of Economics , 4(2), 121–132.

39.

Krueger

Pinto

R. C. B.

Thomas

(2004). Impacts of the South Asia free trade agreement. In Policy Analysis Workshop, Public Affairs, 869, 2003–2004. https://lafollette.wisc.edu/images/publications/workshops/2004-SAFTA.pdf

40.

Kumar

Singh

(2009). India’s role in South Asia trade and investment integration (Working Paper Series on Regional Economic Integration 32). Asian Development Bank.

41.

Lewis

Veronica

C. M.

Francis

Isaac

(2019). Effects of gross domestic product and inflation rate on unemployment rate in Ghana: Comparative analysis of multiple regression and Covariance Matrix Models. American Journal of Applied Mathematics , 7(1), 5–12.

42.

MacKay

J. (Ed.)

. (2005). Regional economic integration in a global framework . European Central Bank. https://www.ecb.europa.eu/pub/pdf/other/regionaleconintegrationglobalframework2005en.pdf

43.

Marshall

M. G.

Jaggers

Gurr

T. R.

(2007). Polity IV project: Political regime characteristics and transitions, 1800–2006 . Center for International Development and Conflict Management, University of Maryland.

44.

McCord

W. M.

(1965). The springtime of freedom: Evolution of developing societies (No. 321.7). Oxford University Press.

45.

Michálek

Výbošťok

(2019). Economic growth, inequality and poverty in the EU. Social Indicators Research , 141(2), 611–630.

46.

Moinuddin

(2013). Fulfilling the promises of South Asian integration: A gravity estimation (ADBI Working Paper 415). http://hdl.handle.net/11540/1177

47.

Nansadiqa

Masbar

Majid

M. S. A.

(2019). Does economic growth matter for poverty reduction in Indonesia. East African Scholars Journal of Economics, Business and Management , 2(2), 46–52.

48.

Nguyen

Ezaki

(2005). Regional economic integration and its impacts on growth, poverty and income distribution: The case of Vietnam. Review of Urban And Regional Development Studies , 17(3), 197–215. https://doi.org/10.1111/j.1467-940X.2005.00104.x

49.

Olson

(1982). The rise and decline of nations: Economic growth, economic rigidities and stagflation . Yale University Press. https://www-jstor-org-443.web.bisu.edu.cn/stable/j.ctt1nprdd

50.

Özer

Žugić

Tomaš-Miskin

(2018). The relationship between current account deficits and growth in Montenegro: ARDL bounds testing approach. Journal of Central Banking Theory and Practice , 7(3), 5–24.

51.

Peiris

T. U. I.

Azali

Habibullah

M. S.

Hassan

(2017). Is SAFTA desirable? Evidence from static theory. Colombo Business Journal , 8(1), 43–61.

52.

Pelinescu

(2015). The impact of human capital on economic growth. Procedia Economics and Finance , 22, 184–190.

53.

Perera

Siriwardana

Mounter

(2017). Trade facilitation, economic development and poverty alleviation: South Asia at a glance. Poverty, Inequality and Policy , Gabriel Ilie Staicu, IntechOpen. https://dx-doi-org-s.web.bisu.edu.cn/10.5772/intechopen.69948

54.

Persson

Tabellini

(1994). Representative democracy and capital taxation. Journal of Public Economics , 55(1), 53–70.

55.

Pesaran

M. H.

(2004). General diagnostic tests for cross section dependence in panels (Cambridge Working Papers in Economics 0435). Faculty of Economics, University of Cambridge.

56.

Pesaran

M. H.

(2007). A simple panel unit root test in the presence of cross‐section dependence. Journal of Applied Econometrics , 22(2), 265–312.

57.

Phillips

P. C.

Hansen

B. E.

(1990). Statistical inference in instrumental variables regression with I(1) processes. The Review of Economic Studies , 57(1), 99–125.

58.

Preepremmote

Santipolvut

Puttitanun

(2018). Economic integration in the ASEAN and its effect on empirical economic growth. Journal of Applied Economic Sciences , 13(4), 922–934.

59.

Przeworski

Alvarez

R. M.

Alvarez

M. E.

Cheibub

J. A.

Limongi

(2000). Democracy and development: Political institutions and well-being in the world, 1950–1990 . Cambridge University Press.

60.

Raihan

(2012). SAFTA and the South Asian countries: Quantitative assessments of potential implications (MPRA Paper 37884). University Library of Munich.

61.

Raihan

(2014). Next steps to South Asian Economic Union: A study on regional economic integration (Phase II) Commissioned by the SAARC Secretariat . Asian Development Bank. http://hdl.handle.net/11540/4721

62.

Rao

(1984). Democracy and economic development. Studies in Comparative International Development (SCID) , 19(4), 67–81.

63.

Regmi

R. K.

Devkota

S. C.

Upadhyay

M. P.

(2017). Impact of SAFTA on South Asian trade. Asian Economic and Financial Review , 7(3), 232.

64.

Roemer

Gugerty

M. K.

(1997). Does economic growth reduce poverty? (CAER II Discussion Paper 4). Harvard Institute for International Development.

65.

Sahnoun

Abdennadher

(2019). Causality between inflation, economic growth and unemployment in North African Countries. Economic Alternatives , (1), 77–92.

66.

Saikkonen

(1992). Estimation and testing of cointegrated systems by an autoregressive approximation. Econometric Theory , 8(1), 1–27.

67.

Santos

M. E.

Dabus

Delbianco

(2019). Growth and poverty revisited from a multidimensional perspective. The Journal of Development Studies , 55(2), 260–277.

68.

Schiff

Winters

L. A.

(2003). Regional integration and development . The World Bank. http://documents.worldbank.org/curated/en/263681468780873762/Regional-integration-and-development

69.

Sirowy

Inkeles

(1990). The effects of democracy on economic growth and inequality: A review. Studies in Comparative International Development , 25(1), 126–157.

70.

Stock

J. H.

Watson

M. W.

(1993). A simple estimator of cointegrating vectors in higher order integrated systems. Econometrica: Journal of the Econometric Society , 61(4), 783–820.

71.

Sulaiman

Bala

Tijani

B. A.

Waziri

S. I.

Maji

I. K.

(2015). Human capital, technology, and economic growth: Evidence from Nigeria. Sage Open , 5(4). https://doi.org/10.1177/2158244015615166

72.

Taguchi

(2018). Trade impacts of South Asian Free Trade Agreements: The case of Sri Lanka (MPRA Paper 85238). University Library of Munich.

73.

Tahir

Khan

(2014). Trade openness and economic growth in the Asian region. Journal of Chinese Economic and Foreign Trade Studies , 7(3), 136–152.

74.

Van Reenen

(2016). Brexit’s long-run effects on the UK economy (pp. 367–383). Brookings Papers on Economic Activity.

75.

Viner

(2014). The customs union issue . Oxford University Press.

76.

Westerlund

(2007). Testing for error correction in panel data. Oxford Bulletin of Economics and Statistics , 69(6), 709–748.

77.

World Bank and UNESCAP. (2015). International trade costs (database) . World Bank.

78.

World Bank. (2018). Poverty and shared prosperity 2018: Piecing together the poverty puzzle . World Bank Group. http://documents1.worldbank.org/curated/en/104451542202552048/pdf/Poverty-and-Shared-Prosperity-2018-Piecing-Together-the-Poverty-Puzzle.pdf

79.

Yurdakul

Ucar

(2015). The relationship between current deficit and economic growth: An empirical study on Turkey. Procedia Economics and Finance , 26, 101–108.

80.

Yusuf

S. A.

(2014). The analysis of impact of investment in education on economic growth in Nigeria: Veracity of association of staff union of University of Nigeria’s agitation (MPRA Paper 55524). University Library of Munich.