Globalisation,Economic Growth,and Spillovers: A Spatial Analysis

Abstract

This article seeks to deepen our understanding of the globalisation–growth nexus as it extends the investigation to using a spatial econometric approach, hitherto rarely used in the globalisation literature. The objective of the article is to uncover not only the significant growth effects of globalisation but also its possible spillover effects onto neighbouring countries. Using a panel data set of 83 countries across a 30-year period and via a spatial autoregressive panel data method, this article estimates a standard growth model augmented with a parameter to capture the countries’ spatial dependence, whilst controlling for globalisation indices. The findings indicate a positive effect of economic globalisation, which is dependent upon the political settings in the countries under study. The spillover effects of globalisation across neighbouring countries are shown, both in geographical and institutional spheres. The article concludes with some policy recommendations.

JEL: F63, C31

Keywords

Globalisation Economic Growth Institutional Quality Spatial Analysis

1. Introduction and Background

‘It has been said that arguing against globalisation is like arguing against the laws of gravity,’ Kofi Annan, the former Secretary General of the United Nations, is once reported to have said.¹ Globalisation is apparently one of the many highly debated topics in the growth and development literature. Theoretically, it has many positive effects on growth via various mechanisms such as increased knowledge spillovers between countries, greater economies of scale, innovation potentials due to specialisation, effective allocation of domestic resources, diffusion of technology, improvement in factors productivity and augmentation of capital.

Notwithstanding the theoretical arguments, the empirical findings on the globalisation–growth nexus are still far from conclusive, as have been discussed by Grossman and Helpman (2015) and Samimi and Jenatabadi (2014). Generally, empirical studies on the effects of globalisation on growth can be divided into three general groups: first, studies with findings that are supportive of the positive effects of globalisation on growth; second, studies that postulate adverse effects of globalisation on growth; and finally, studies that argue that the positive growth effects of globalisation are dependent upon complementary policies.

Studies in this first group, for example, that of Dollar (1992), Sachs, Warner, Aslund, and Fischer (1995) and Edwards (1998), are able to show the positive growth effects of globalisation using various de facto indices of globalisation, namely, trade openness and foreign capital inflows. In contrast, studies arguing against the positive effects of globalisation on growth reject the existing evidence that according to them are weak and non-robust. For example, Rodriguez and Rodrik (2000) refute the findings of Dollar (1992), Sachs et al. (1995) and Edwards (1998), arguing that their evidence is weak due to the omission of some important growth indicators and the use of questionable trade openness indices. Alesina and Perotti (1994), Rodrik (1998) and Stiglitz (2004) too have expressed their reservations on the potential growth improvement driven by mechanisms related to globalisation. Finally, there are studies arguing that the positive growth effects of globalisation are dependent upon the presence of complementary policies in the globalising countries. For example, sufficient stock of human capital could enhance the positive effect of FDI, as shown by Borensztein, De Gregorio, and Lee (1998). In addition, structural policies relating to education, infrastructure, institutions, regulatory framework, among others, could be determining factors in the generation of positive globalisation effects (Calderón & Poggioa, 2010).

With regard to indicators of globalisation, arguably the most widely used indicator is the KOF index of globalisation, first introduced by Dreher (2006) and continuously updated by Dreher, Gaston, and Martens (2008). KOF is a comprehensive index of globalisation that comprises three dimensions, namely, economic, political and social globalisation. As is stated by Dreher (2006), this index in general captures the major ideas in a globalisation process such as creating new networks among economic actors worldwide, mediated by a variety of inflows such as capital, culture, goods, people, information and ideas. It is a process that erodes national boundaries, integrates national economies, cultures, technologies and governance, and produces a complex relation of mutual interdependence.

In his panel study on 123 countries for 1970–2000, Dreher finds that globalisation has positive effects on growth, especially in the economic and social dimensions. The political dimension, however, has no significant growth effect. Applying the KOF index in 21 African countries for 1970–2005, Rao and Vadlamannati (2011) find similar positive effects of globalisation on growth. The positive finding is also supported by Gurgul and Lach’s (2014) study on ten CEE economies. Samimi and Jenatabadi (2014) too find positive significant effects of economic globalisation in selected OIC countries, however, they argue that the effect is dependent upon the levels of human capital and financial development.

Arguably, the mixed findings could be the result of different samples of countries and period specifications used in the studies, varying econometric techniques, as well as the presence of unobserved country-specific effects that bias the final results. As pointed out by Samimi and Jenatabadi (2014), the majority of the literature in the field of globalisation used trade or foreign capital volume as the de facto indices of globalisation to investigate its impact on economic growth. The issue with these de facto indices is that they do not proportionally capture trade and financial globalisation policies. Apart from trade and volume of capital inflows, the rate of protections and tariff also need to be accounted for since they are policy-based variables capable of reflecting the extent of trade restrictions in a country.

This article revisits the globalisation–growth nexus by extending the analysis into the spatial effect of globalisation using the spatial econometric estimation method. The spatial weight matrices used in this study comprise both geographical and institutional matrices. The use of geographical matrices is pretty unambiguous since globalisation processes are frequently shown to occur across countries located within the same clusters of areas, regions or economic clubs. Additionally, the geographical distance is widely used as a natural proxy for transportation costs and technological transfers, a common feature in the globalisation process. On the other hand, the use of an institutional matrix is of a somewhat recent vintage in the spatial studies, and apparently rarely applied in the globalisation–growth nexus. The use of institutional matrices is derived from the concept of institutional proximity discussed in Ahmad and Hall (2017), to distinguish a group of countries sharing similar institutional qualities.²

Against this backdrop, the research questions this study seeks to answer are: Does globalisation significantly determine growth? Is globalisation capable of generating a spillover effect on neighbours’ economic performance? What is the role of institutional quality in the globalisation–growth relationship? Does globalisation propagate its spillover effects to countries sharing similar institutional qualities? Whilst the first question is rather straightforward, the latter three dig deeper into possible globalisation spillover effects across neighbouring countries, notwithstanding the definitions of ‘neighbour’ either by geographical distance or via an institutional proximity of the countries under study.

Ultimately, this study seeks to contribute to our understanding on the globalisation process via a spatial econometric analysis with the aim of uncovering the effects of globalisation on growth and spillovers, which is a major contribution to the existing globalisation–growth literature. Apart from relying on geographical distance to capture the spillover effects of globalisation, this study also utilises the concept of institutional proximity in investigating the possible globalisation spillover effects across a group of countries with similar institutional characteristics. Finally, the panel data set of 83 countries for 1985–2014 used in this study is arguably extensively large and sufficiently able to yield robust answers to the above questions.

In general, the findings of this study indicate that economic globalisation has positive significant effects on growth, whereas political and social institutions do not. This result is consistent even when institutional quality is controlled for. Furthermore, economic globalisation is shown to be dependent on the complementary political settings in the countries under study. Economic globalisation is also shown to have indirect spillover effects supporting the growth performance of geographically closer countries or countries sharing similar institutional characteristics.

The study proceeds as follows: Section 2 discusses the globalisation–growth spatial model, followed by Section 3 discussing the data sources and estimation strategy. Section 4 interprets the results and Section 5 concludes.

2. Globalisation–growth spatial model

Consider a simple growth model based on Barro (1991) as follows:

g_{i t} = α_{i} + X_{i t} β + γ_{t} + ε_{i t}

(1)

where g_it is the average growth rate of GDP per capita in country i measured over a five-year interval; and X is a vector of explanatory variables that includes three globalisation indices and two institutional quality variables to reflect economic and political institutions, respectively, and some commonly used variables controlling for other growth determinants. Meanwhile α_i captures the unobserved country-specific effect, γ_t the time effects and ε_it represents the corresponding disturbance term where ε ~ N(0, σ²I). The control variables included in the vector X are commonly used determinants of growth, namely, initial level of real GDP per capita (in natural logarithmic form) proxied by the first-period real GDP per capita for each of the five-year intervals of our data set. This inclusion is meant to capture the convergence process, and the coefficient for initial GDP per capita is expected to be negative to show the catching up of countries to their steady-state growth level. The investment level, population growth rate, education (to reflect the level human capital) and inflation rate (as a proxy for macroeconomic policy) are among the control variables included in the growth model.

To account for spatial dependence in the growth process, Equation (1) is expanded with the error structure as:

ε_{i t} = λ W ε_{i t} + u_{i t}

(2)

where W is the spatial weight matrix capturing the spatial connections between the countries, λ is a spatial autoregressive parameter, ε_it is the spatially correlated errors, and u_it is the spatial disturbance term with i.i.d. properties. Equation (1) with the error process of Equation (2) is normally called the spatial error model (SEM), where spatial dependence operates via the residuals, since dependency is assumed to be present in the error terms due to the omission of some unobserved variables that could be spatially correlated. Nevertheless, by this definition, it also renders spatial spillovers a ‘nuisance’ factor that makes the spatial effect relatively less important in the model (Arbia et al., 2010).

To model a more substantive spatial effect, the spatial autoregressive model (SAR) is frequently used, as the following:

g_{i t} = α_{i} + X_{i t} β + ρ W g_{i t} + γ_{t} + u_{i t}

(3)

Equation (3) is an augmented model of Equation (1), with the presence of the term ρWg_it among the right-hand side variables. This term, called the spatially lagged dependent variable, captures countries’ spatial dependence in a more substantive manner, and shows that the growth rates of the home country depend, in part, on the weighted average of its ‘neighbours’’ growth rates.³ Apart from the variables capturing globalisation and institutional quality in the model, this term will be another variable of interest in this study. Its coefficient, rho (ρ), indicates the size of the growth spillovers between neighbours, and the sign of ρ if positive (negative) indicates countries with similar (dissimilar) levels of growth would cluster together. Although the SAR model seems to be preferred over the SEM, to decide which of the two models best suits our data, we refer to the Lagrange multiplier (LM) statistics that will be explained further in the estimation strategy discussion.

W in Equation (3) is the weight matrix to conceptualise the spatial dependence between countries and as earlier discussed, matrices based on geographical and institutional proximities are used. For geographical matrices, three measures are used: first, a simple binary contiguity matrix where countries are defined as neighbours if they share common borders; this matrix is denoted w_contig (therefore its element, w_ij = 1 if countries i and j have common borders, w_ij = 0 otherwise).⁴ Second, the k-nearest regions matrix, in which k is set to equal to five, beyond which spatial dependence is considered negligible; it is denoted as w_knn (its element w_ij = 1 if country j is located within the five nearest regions to country i, w_ij = 0 otherwise). Finally, an inverse-squared distance matrix based on the concept of exponential distance decay, and denoted as w_invsq.

The distance calculation for both the w_knn and w_invsq matrix is done via the great circle distance computation using the latitude and longitude coordinates of the countries’ capitals (Le Gallo & Ertur, 2003).⁵ For the matrix w_invsq, a cut-off distance is specified at a minimum threshold which will guarantee that each country in the sample will have at least one neighbour; therefore, the element of w_invsq is given by $w_{i j} = d_{i j}^{- 2} / \sum_{j} d_{i j}^{- 2}$ if $d_{i j}^{- 2} \leq \overset{─}{d^{- 2}}$ where d_ij is the great circle distance between the capitals of country i and j, $\overset{─}{d}$ is the critical distance cut-off after which the spatial effect is considered to be negligible, that is, w_ij = 0 if $d_{i j}^{- 2} > \overset{─}{d^{- 2}}$ .⁶

Although weight matrices based on geography have an exogeneity advantage and are able to avoid the identification problem, they may not be able to capture the true interdependence between the countries under study that may be shaped by non-geographical factors. To this end, this study posits that institutionally similar countries are expected to have similar degrees of globalisation, which would consequently support growth and spillovers among this group of countries. The institutional distance matrix is thus expected to uncover the impact of this institutional proximity on the globalisation–growth relationship. Nevertheless, the endogeneity of the institutional matrix could possibly bias the spatial estimators, if the matrix is constructed from time-varying institutional indicators scores, as Ahmad and Hall (2012b) have discovered. To avoid the identification problem due to this endogeneity issue, the historical determinants of institutions, such as legal origins, colonial origins and language characteristics are used to construct an institutional matrix. These historical determinants of institutions are undoubtedly time-invariant therefore exogenous to the model.

The use of these historical determinants of institutions are based on the arguments of the following important studies: Acemoglu, Johnson, and Robinson (2001, 2002) on the impact of colonial origins; La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1998), La Porta, Lopez-de-Silanes, and Shleifer (2008) and Glaeser and Shleifer (2002) on the impact of legal origins to the current institutions; Alesina, Devleeschauwer, Easterly, Kurlat, and Wacziarg (2003) on linguistic fractionalisation roles in explaining the institutions, and the proposition of Easterly, Ritzen, and Woolcock (2006) that social cohesion (instrumented with linguistic homogeneity) leads to better institutions (see “theoretical arguments on the historical determinants of institutions” in Appendix for a summary of these studies’ theoretical propositions).

In brief, legal origins, colonial origins and language are perceived to be deep-determinants of the current level of institutions, and constitute the underlying framework of the institutional proximity concept. Institutional proximity thus can be defined as a situation when two or more countries in the sample share similar legal origins, colonial origins, and language, the historical factors that are expected to shape countries’ present-day institutions in a similar process over the long term. This natural process is assumed to eventually lead to the creation of a conducive economic environment supporting greater economic activities, increased bilateral trades, spillovers between countries of growth-promoting factors such as technology, human capital, information, etc., all of which are common features in the globalisation process.

Returning to the weight matrix W specification using institutional distance, the three institutional matrices are of a binary matrix, whose element w_ij = 1 if countries i and j share their legal origins, colonial origins and spoken language, and w_ij = 0 otherwise. Matrices based on legal origins are denoted as w_legor, colonial origins as w_color and language as w_lang. Finally, the elements of the main diagonal of all the geographical and institutional matrices are set equal to 0 by convention, since a country cannot be a neighbour to itself and all matrices are row standardised.

3. Data Sources and Estimation Strategy

A panel data set consisting of observations for 83 countries for a period of 30 years, from 1985 to 2014, is used in this study.⁷ All variables, with the exception of globalisation and the institutional variables which have an initial period value, are taken as the average over a five-year interval, therefore there are six five-year intervals in this study, with a total of 498 observations. Data on real GDP per capita growth, real GDP per capita (in natural logarithmic form), gross fixed capital formation as a share of GDP (as a proxy for investment), population growth rates, and inflation (measured by the GDP deflator) are obtained from the World Development Indicators (WDI) data set from the World Bank (2016). Finally, the education variable measured by the average years of total schooling for the population aged 15 and above is obtained from Barro and Lee (2013).

The variable of interests are the indices of globalisation, namely, economic globalisation, political globalisation and social globalisation, obtained from the KOF index (Dreher, 2006; Dreher et al., 2008). To capture the institutional quality in the countries under study, two institutional variables widely used in the growth literature are included: first, ‘law and order’ obtained from the International Country Risk Guide data set (PRS Group, 2014) to represent the level of economic institutions; and second the Polity 2 variable from the Polity IV data set (Marshall & Jaggers, 2014), to capture the level of political institutions. Summary statistics and variable definitions are in Tables A1 and A2, respectively.

For the institutional distance matrices, data for Legal Origins are obtained from La Porta et al. (1998), which identifies the legal origin of the company law or commercial code for each country from five possible sources, that is, English Common Law, the French Commercial Code, Socialist/Communist Laws, German Commercial Code or Scandinavian Commercial Code. For colonial origins, data are obtained from Central Intelligence Agency (2013) that classify former colonial rulers into Dutch, Spanish, Italian, United States, British, French or Portuguese. There are also countries in our sample that had never been colonised. Meanwhile, for the language similarity matrix, the language of the country is determined based on its official and second languages; when these languages have no neighbours (no other countries speaking them), the next-most common language spoken by immigrants is used, to meet the matrix requirement of at least one neighbour for each individual country. Language data are obtained from the CIA World Factbook (Wikipedia, 2016) cross-referenced against the Wikipedia page: ‘List of official languages by country and territory’ (Wikipedia, 2016).

Usage of spatial econometrics is widespread in the analysis of cross-sectional models, however, the application of spatial analysis to panel data is still quite restricted mainly for two reasons: Theoretical models are very recent and computation implementation is difficult. We refer to Elhorst (2003, 2009, 2010) for the appropriate specifications of spatial panel models and Anselin, Bera, Florax, and Yoon (1996) for the test statistics to help with the estimation and testing processes. The presence of spatial autocorrelation in the residuals of the OLS estimation of Equation (1) is tested using Moran’s I test, and if it is present, the OLS estimates are no longer appropriate.⁸ The Moran scatterplot is also used to explore the spatial autocorrelation of the countries’ growth, regardless of the measures their distances from each other.

Having detected the presence of spatial effects, the appropriate form of the spatial model, either spatial error or the spatial lag model,⁹ is subsequently determined using the robust LM test (it is called robust because the existence of one type of spatial dependence does not bias the test for another type of spatial dependence). This robust characteristic is obviously important because the spatial model that fails this test in most cases when estimated with different weight matrices would be omitted. Finally, a spatial panel fixed-effect (FE) estimation technique based on Elhorst (2003, 2009) via a STATA command ‘spregfext’ prepared by Shehata and Mickaiel (2013) is employed. Spatial panel fixed-effect, and not random effect, estimation is considered due to the nature of the panel data set that assumes the presence of unobserved heterogeneity in the growth estimations as a result of omitted variables that may influence the growth process.¹⁰

4. Estimation Results and Discussion

The results of the OLS estimation in Table 1 below fit the stylised facts about the standard growth process, while at the same time controlling for the effects of globalisation. The presence of a conditional convergence process in the countries’ growth, at the rate of 0.8 per cent, is evidenced by the negative and statistically significant coefficient for initial GDP per capita. Similarly, the coefficients of the other growth determinants are also statistically significant with the expected signs. Meanwhile, among the globalisation indices, only economic globalisation is significant at the 1 per cent level.

Table 1
OLS Estimation Result of a Standard Growth Model Controlling Globalisation Variables

Initial GDP per capita –0.792 (0.150)*

Investment 0.174 (0.017)*

Population growth –0.567 (0.112)*

Education 0.104 (0.306)

Inflation –0.001 (0.001)

Economic globalisation 0.280 (0.106)*

Political globalisation 0.072 (0.070)

Social globalisation 0.055 (0.132)

Constant 2.991 (1.060)***

Source: The author.

Notes: Dependent variable is real GDP per capita growth. Standard error in parentheses; number of observations: 510, number of countries: 85, F-test p-value = 0.000, R-squared = 0.276, adjusted R-squared = 0.250; ***, ** and * denote significance at 1%, 5% and 10%, respectively.

Nevertheless, interpretation of the aforementioned results needs caution since there is a possibility of spatial autocorrelation in the error term leading to misspecification and bias. Moran’s I test statistics in Table 2 is thus referred, and the results indicate that the null hypothesis of no spatial autocorrelation in the residuals of the OLS regression is overwhelmingly rejected in all estimations regardless of the types of matrix, with the exception of w_legor. Similarly, as is seen in Figure 1, Moran scatterplots of the home country’s growth against the spatially lagged growth for the significant matrices also show a positive relationship between growth in a home country and growth levels in neighbouring countries. This is consistent regardless of distance measures. Thus, it can be perceived that not only do countries’ growth levels tend to cluster in space when distance is defined by geography but also when they are institutionally similar countries (when distance is defined using institutional proximity).

Table 2

Moran’s I Test for Spatial Autocorrelation in OLS Regression with Different Matrices and Robust LM Test for Spatial Error versus the Spatial Lag Model

	w_contig	w_knn	w_invsq	w_legor	w_color	w_lang
Moran’s I	3.281***	4.689***	5.023***	1.023	4.600***	4.194***
Spatial error
LM test	9.679***	19.557***	22.679***	0.561	17.064***	15.599***
Robust LM test	0.043	0.533	0.125	0.375	1.011	1.683
Spatial lag
LM test	12.151***	28.753***	26.056***	1.965	17.467***	14.853***
Robust LM test	2.515	9.729***	3.502*	1.779	1.414	0.936

Source: The author.

Notes: ***, ** and * denote significance at 1%, 5% and 10%, respectively. w_contig = contiguity matrix, w_knn = five-nearest region matrix, w_invsq = inversed squared distance matrix, w_legor = legal origins matrix, w_color = colonial origins matrix and w_lang = language matrix.

Figure 1

Source: The author.

Therefore, Equation (1) can be considered misspecified and must be modified to include a spatial dependence term. Returning to Table 2, the robust LM test statistics indicate that the SEM is apparently inappropriate as it fails in all estimations across different matrix specifications. Therefore, the SAR or spatial lag model is the preferred model to explain the spatial dependence of growth in the countries under study.

Returning to this study’s research questions, how do the globalisation growth effects fare in these two distinct spatial settings? Table 3 shows the results of the SAR model of Equation (3), via spatial fixed-effects estimations with five weight matrices found to be significant in the earlier Moran’s I tests. In Table 4, the similar estimations are repeated with the presence of institutional quality variables.

Table 3

Spatial Fixed-effect Estimation of the Globalisation–growth Model with Various Matrices Without the Institutional Variables

	w_contig		w_knn		w_invsq		w_color		w_lang
Rho, ρ	0.257***	(0.056)	0.372***	(0.076)	0.390***	(0.065)	0.330**	(0.147)	0.397***	0.089)
Economic globalisation	0.650***	(0.219)	0.559**	(0.230)	0.577***	(0.222)	0.610***	(0.231)	0.682***	(0.233)
Political globalisation	0.165	(0.213)	0.147	(0.224)	0.161	(0.219)	0.196	(0.223)	0.217	(0.215)
Social globalisation	0.032	(0.212)	0.072	(0.214)	0.004	(0.216)	0.027	(0.219)	–0.113	(0.231)
Initial GDP per capita	–4.007***	(0.693)	–3.772***	(0.757)	–3.821***	(0.709)	–4.031***	(0.761)	–3.885***	(0.768)
Investment	0.195***	(0.034)	0.195***	(0.034)	0.192***	(0.033)	0.198***	(0.036)	0.197***	(0.034)
Population growth	–0.602**	(0.305)	–0.609**	(0.305)	–0.646**	(0.302)	–0.595*	(0.309)	–0.606*	(0.317)
Education	0.285	(1.149)	–0.022	(1.189)	0.256	(1.153)	0.123	(1.224)	0.247	(1.187)
Inflation	–0.001***	(0.000)	–0.002***	(0.000)	–0.001***	(0.000)	–0.001***	(0.000)	–0.001***	(0.000)
Constant	25.573***	(4.306)	24.479***	(4.666)	24.559***	(4.357)	25.901***	(4.688)	24.472***	(4.843)
No of observations	498		498		498		498		498
No of countries	83		83		83		83		83
Wald p-value	0.000		0.000		0.000		0.000		0.000
F-test p-value	0.000		0.000		0.000		0.000		0.000
R-squared	0.554		0.558		0.566		0.541		0.552
Adjusted R-squared	0.454		0.459		0.468		0.438		0.451
LLF	–967.798		–965.086		–960.954		–974.714		–968.754
AIC	3.007		2.975		2.926		3.092		3.019

Source: The author.

Notes: The dependent variable is real GDP per capita growth; robust standard errors are in parentheses; the Wald test is for the null hypothesis that rho, ρ = 0 ~ χ²(9); ***, ** and * denote significance at 1%, 5% and 10%, respectively.

Table 4

Spatial Fixed-effect Estimation of the Globalisation–growth Model with Various Matrices in the Presence of Institutional Variables

	w_contig		w_knn		w_invsq		w_color		w_lang
Rho, ρ	0.247***	(0.057)	0.367***	(0.077)	0.383***	(0.067)	0.306**	(0.151)	0.405***	(0.091)
Economic globalisation	0.539**	(0.225)	0.437**	(0.237)	0.462**	(0.228)	0.498**	(0.241)	0.559**	(0.240)
Political globalisation	0.074	(0.203)	0.045	(0.209)	0.066	(0.206)	0.102	(0.209)	0.115	(0.201)
Social globalisation	0.100	(0.209)	0.143	(0.212)	0.075	(0.211)	0.096	(0.217)	–0.034	(0.224)
Law & order	–0.075	(0.104)	–0.057	(0.108)	–0.073	(0.106)	–0.074	(0.115)	–0.105	(0.112)
Polity 2	0.169**	(0.075)	0.182**	(0.075)	0.175**	(0.073)	0.174**	(0.079)	0.189**	(0.078)
Initial GDP per capita	–3.924***	(0.690)	–3.678***	(0.752)	–3.734***	(0.706)	–3.951***	(0.756)	–3.778***	(0.763)
Investment	0.196***	(0.033)	0.196***	(0.033)	0.193***	(0.032)	0.200***	(0.035)	0.198***	(0.033)
Population growth	–0.649**	(0.288)	–0.657**	(0.286)	–0.693**	(0.285)	–0.645**	(0.291)	–0.657**	(0.297)
Education	0.148	(1.110)	–0.174	(1.147)	0.106	(1.111)	0.009	(1.178)	0.059	(1.145)
Inflation	–0.002***	(0.000)	–0.002***	(0.000)	–0.002***	(0.001)	–0.001***	(0.000)	–0.001***	(0.000)
Constant	25.212***	(4.372)	24.016***	(4.739)	24.177***	(4.410)	25.528***	(4.750)	24.051***	(4.878)
No of observations	498		498		498		498		498
No of countries	83		83		83		83		83
Wald p-value	0.000		0.000		0.000		0.000		0.000
F-test p-value	0.000		0.000		0.000		0.000		0.000
R-squared	0.563		0.569		0.575		0.550		0.563
Adjusted R-squared	0.462		0.469		0.477		0.447		0.463
LLF	–962.745		–959.221		–955.429		–969.554		–962.328
AIC	2.971		2.929		2.885		3.053		2.965

Source: The author.

Note: The dependent variable is real GDP per capita growth; standard errors are in parentheses; the Wald test is for the null hypothesis that rho, ρ = 0 ~ χ²(11); ***, ** and * denote significance at 1%, 5% and 10%, respectively.

The variables of interest are the three indices of globalisation. Apparently, the results in Table 3 of the spatial fixed-effect estimations mirror that of the OLS estimation, where economic globalisation is the only KOF index significant at the 1 per cent level. The level of significance is consistent across the geographical and institutional matrices. The latter is undoubtedly an important finding that is able to support our earlier proposition that countries sharing similar institutional characteristics (in this case similar colonial origins and language homogeneity) are expected to cluster along similar levels of globalisation, and consequently to generate a spillover effect towards growth. Meanwhile, in Table 4, the economic globalisation index retains its significance in the presence of variables capturing institutional quality in the countries under study, although the level of significance is now slightly weaker at 5 per cent, compared to 1 per cent when it is estimated independently of institutional quality in Table 3. This is arguably additional evidence for the study finding positive globalisation effects dependent upon complementary policies, as earlier discussed.

This study finds that institutional quality is apparently an important factor supporting the positive growth effect of globalisation, at least in the case political institutions. Specifically, Polity 2 is found to be significant at the 5 per cent level with the expected sign, but rule of law is not and has the wrong sign. This is consistent across all estimations with different weight matrices. This finding also support the proposition of political prominence by Acemoglu, Johnson, and Robinson (2005) over property rights institutions proposed by North (1990) and many similar studies thereafter.

Another variable of interest is spatially lagged growth; its coefficient ρ (rho) indicates the size of growth spillovers from neighbours and its positive sign indicates that countries with similar growth levels cluster together (countries with high growth cluster together with high-growth neighbours, and vice versa). As is seen in Tables 3 and 4, ρ is positive and significant at the 1 per cent level across model specifications using all weight matrices. The Wald tests for the null hypothesis of ρ = 0 are overwhelmingly rejected and this finding gives convincing support to the proposition of positive growth spillovers across the countries under study. A significant economic globalisation variable and the spatially lagged growth variable therefore confirm the presence of indirect globalisation spillover effects (via neighbours’ growth).

Finally, the results in all estimations support the conditional convergence hypothesis where the coefficients of initial real GDP per capita are consistently negative and significant at the 1 per cent level. The rate of growth convergence ranging from 3.7 per cent to 4.0 per cent is apparently greater than the 0.8 per cent previously found in the OLS estimation. The finding of a higher rate of convergence when the growth model is spatially augmented is common in many spatial growth studies (Arbia et al. [2010], Ahmad & Hall [2017], Ho, Wang, and Yu [2013], to name a few), a further robust evidence against the omission of spatial dependence in growth regressions which would otherwise bias the estimates. The coefficients of the other growth determinants, namely, investment, population growth and inflation, are significant too with the expected signs. Education, however, is not.

5. Concluding Remarks

Studies investigating the globalisation–growth nexus using explicit spatial econometrics methodology and incorporating the concept of institutional proximity are of recent vintage. The present article thus seeks to contribute to the existing literature in this respect. By using a spatial panel fixed-effect estimation on a panel data set of 83 countries over a 30-year period, this study seeks to examine the effects of globalisation on the countries’ growth and spillovers. The results show that economic globalisation is a significant determinant of growth, and when it is spatially modelled, economic globalisation generates a positive spillover effect on neighbouring countries. Not only countries that are close in term of geographical settings, this study shows that the spillover effects of globalisation are also propagated across countries sharing similar legal origins and spoken languages. Additionally, the results are able to support the argument on significant complementary policies supporting the positive effect of globalisation, in this case supportive political institutions.

These results are expected to inform policymakers on appropriate globalisation and economic integration policies, particularly with respect to countries’ institutional settings and development. Since economic globalisation has been shown to be a significant growth determinant, coupled with a significant finding on the growth effects of political institutions, the results further illustrate the globalisation-institutional interplay in ensuring positive effects on globalisation. Furthermore, the evidence on the presence of globalisation spillover effects not only across countries located closer in the geographical sphere but also across countries sharing similar institutional characteristics, could pave the way for aspiring countries seeking to integrate with higher-income nations to focus on developing their institutional quality to levels similar to their prospective integration partners.

A potential extension to this study in future would be to consider various non-geographical weight matrices, not limited to institutions per se, and to combine two or more different matrices in an estimation for the purpose of appropriate comparison. Further, the latest estimation techniques in spatial analysis may be explored to investigate spatially dependent growth dynamics.

Footnotes

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

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

Appendix

Table A2

Description of Variables and Sources

Variable	Description	Sources
Real GDP per capita growth	Annual percentage change in gross domestic product per capita (constant 2005 US$)—average over five-year interval (in natural logarithmic form)	World Development Indicators (the World Bank, 2016)
Initial real GDP per capita	Initial value of GDP per capita (constant 2005 US$) at the start of the five-year interval (in natural logarithmic form)	World Development Indicators (the World Bank, 2016)
Investment	Gross fixed capital formation as a percentage of GDP
Population growth	Annual percentage change in population size
Inflation	Annual percentage change in GDP deflator. GDP deflator is the ratio of GDP in current local currency to GDP in constant local currency.
Human capital	Average years of total schooling for population age 15 and above as a percentage of population	Barro and Lee (2013)
Law and order	An assessment of the strength and impartiality of the legal system, and public observance of the law	ICRG data set (PRS Group, 2014)
Polity2	Measures key qualities in executive recruitment, constraints on executives, and political competition. It gives indication whether a regime is an institutionalised democracy or institutionalised autocracy or anocracies (mixed, or incoherent, authority regimes).	Polity IV data set (Marshall & Jaggers, 2014)
Economic globalisation	An index measuring economic integration comprises of actual trade, FDI, portfolio investment flows, as well as trade restrictions policies.	KOF index of globalisation (Dreher et al., 2008)
Political globalisation	An index measuring political integration comprises of data on embassies, membership in international organisations and participation in the UN Security Council commissions
Social globalisation	An index measuring social integration comprises of data on personal contact across countries, information flows, and cultural proximity

Notes

References

Acemoglu

, Johnson

, & Robinson

J. A.

(2001). The colonial origins of comparative development: An empirical investigation. The American Economic Review, 91(5), 1369–1401. doi: 10.1257/aer.91.5.1369

Acemoglu

, Johnson

, & Robinson

J. A.

(2002). Reversal of fortune: Geography and institutions in the making of modern world income distribution. The Quarterly Journal of Economics, 117(4), 1231–1294. Retrieved from http://www.jstor.org/stable/4132478

Acemoglu

, Johnson

, & Robinson

J. A.

(2005). Institutions as a fundamental cause of long-run growth. In Aghion

& Durlauf

S. N.

(Eds.), Handbook of economic growth (Vol. 1A; pp. 386–471). Elsevier. doi: 10.1016/S1574–0684(05)01006-3

Ahmad

, & Hall

S. G.

(2012a). Institutions-growth spatial dependence: An empirical test. Procedia—Social and Behavioral Sciences, 65, 925–930. doi: 10.1016/j.sbspro.2012.11.221

Ahmad

, & Hall

S. G.

(2012b). Institutions and growth: Testing the spatial effect using weight matrix based on the institutional distance concept. MPRA Paper, 6356. Retrieved from http://ideas.repec.org/p/pra/mprapa/42294.html

Ahmad

, & Hall

S. G.

(2017). Economic growth and convergence: Do institutional proximity and spillovers matter? Journal of Policy Modeling, 39(6), 1065–1085. doi: 10.1016/j.jpolmod.2017.07.001

Alesina

, Devleeschauwer

, Easterly

, Kurlat

, & Wacziarg

(2003). Fractionalization. Journal of Economic Growth, 8(2), 155–194. doi: 10.1023/A:1024471506938

Alesina

, & Perotti

(1994). The political economy of growth: A critical survey of the recent literature. The World Bank Economic Review, 8(3), 351–371. doi: 10.1093/wber/8.3.351

Anselin

, Bera

A. K.

, Florax

, & Yoon

M. J.

(1996). Simple diagnostic tests for spatial dependence. Regional Science and Urban Economics, 26(1), 77–104. Retrieved from http://www.sciencedirect.com/science/article/pii/0166046295021116

10.

Arbia

, Battisti

, & Di Vaio

(2010). Institutions and geography: Empirical test of spatial growth models for European regions. Economic Modelling, 27(1), 12–21. doi: 10.1016/j.econmod.2009.07.004

11.

Barro

R. J.

(1991). Economic growth in a cross-section of countries. The Quarterly Journal of Economics, 106(2), 407. Retrieved from https://doi.org/10.2307/2937943

12.

Barro

R. J.

, & Lee

J. W.

(2013). A new data set of educational attainment in the world, 1950–2010. Journal of Development Economics, 104, 184–198. doi: 10.1016/j.jdeveco. 2012.10.001

13.

Beck

, Gleditsch

K. S.

, & Beardsley

(2006). Space is more than geography: Using spatial econometrics in the study of political economy. International Studies Quarterly, 50(1), 27–44. doi: 10.1111/j.1468-2478.2006.00391.x

14.

Borensztein

, De Gregorio

, & Lee

J.-W.

(1998). How does foreign direct investment affect economic growth? Journal of International Economics, 45, 115–135. Retrieved from https://www.olemiss.edu/courses/inst310/BorenszteinDeGLee98.pdf

15.

Calderón

, & Poggioa

(2010). Trade and economic growth: Evidence on the role of complementarities for CAFTA-DR countries. In Getting the most out of free trade Agreements in Central America (pp. 83–122). Washington DC: The World Bank.

16.

Central Intelligence Agency (2013). The World Factbook 2013–2014. Washington, DC. Retrieved from https://www.cia.gov/library/publications/the-world-factbook/index.html

17.

Dollar

(1992). Outward-oriented developing economies really do grow more rapidly: Evidence from 95 LDCs, 1976–1985. Economic Development and Cultural Change, 40(3), 523–544.

18.

Dreher

(2006). Does globalization affect growth? Evidence from a new index of globalization. Applied Economics, 38(10), 1091–1110. doi: 10.1080/00036840500392078

19.

Dreher

, Gaston

, & Martens

W. J. M.

(2008). Measuring globalisation: Gauging its consequences. New York: Springer-Verlag. Retrieved from https://www.springer.com/gb/book/9780387740676

20.

Easterly

, Ritzen

, & Woolcock

(2006). Social cohesion, institutions, and growth. Economics and Politics, 18(2), 103–120.

21.

Edwards

(1998). Openness, productivity and growth: What do we really know? Economic Journal, 108(447), 383–398.

22.

Elhorst

J. P.

(2003). Specification and estimation of spatial panel data models. International Regional Science Review, 26(3), 244–268. doi: 10.1177/0160017603253791

23.

Elhorst

J. P.

(2009). Spatial panel data models. In Fischer

M. M.

& Getis

(Eds.), Handbook of applied spatial analysis (pp. 377–407). Springer. doi: 10.1007/978-3-642-03647-7_19

24.

Elhorst

J. P.

(2010). Applied spatial econometrics: Raising the bar. Spatial Economic Analysis, 5(1), 9–28. doi: 10.1080/17421770903541772

25.

Florax

R. J.

, Folmer

, & Rey

S. J.

(2003). Specification searches in spatial econometrics: The relevance of Hendry’s methodology. Regional Science and Urban Economics, 33(5), 557–579. Retrieved from http://www.sciencedirect.com/science/article/pii/S0166046203000024

26.

Glaeser

E. L.

, & Shleifer

(2002). Legal origins. The Quarterly Journal of Economics, 117(4), 1193–1229. doi: 10.1162/003355302320935016

27.

Grossman

G. M.

, & Helpman

(2015). Globalization and growth. American Economic Review, 105(5), 100–104. doi: 10.1257/aer.p20151068

28.

Gurgul

, & Lach

Ł.

(2014). Globalization and economic growth: Evidence from two decades of transition in CEE. Economic Modelling, 36, 99–107. doi: 10.1016/j.econmod.2013.09.022

29.

C. Y.

, Wang

, & Yu

(2013). Growth spillover through trade: A spatial dynamic panel data approach. Economics Letters, 120, 450–453. doi: 10.1016/j.econlet.2013.05.027

30.

La Porta

, Lopez-de-Silanes

, & Shleifer

(2008). The economic consequences of legal origins. Journal of Economic Literature, 46(2), 285–332. doi: 10.1257/jel.46.2.285

31.

La Porta

, Lopez-de-Silanes

, Shleifer

, & Vishny

R. W.

(1998). Law and finance. The Journal of Political Economy, 106(6), 1113–1155.

32.

Le Gallo

, & Ertur

(2003). Exploratory spatial data analysis of the distribution of regional per capita GDP in Europe, 1980–1995. Papers in Regional Science, 201(2), 175–201. doi: 10.1111/j.1467–8276.2006.00866.x

33.

LeSage

J. P.

, & Pace

R. K.

(2009). Introduction to spatial econometrics. Boca Raton: Chapman and Hall/CRC Press.

34.

Marshall

M. G.

, & Jaggers

(2014). Polity IV project: Political regime characteristics and transitions 1800–2014. Retrieved from http://www.systemicpeace.org/polity/polity4.htm

35.

North

D. C.

(1990). Institutions, institutional change and economic performance. Cambridge: Cambridge University Press. doi: 10.1017/CBO9780511808678

36.

PRS Group (2014). International Country Risk Guide dataset. Retrieved from www.prsgroup.com

37.

Rao

B. B.

, & Vadlamannati

K. C.

(2011). Globalization and growth in the low-income African countries with the extreme bounds analysis. Economic Modelling, 28, 795–805. doi: 10.1016/j.econmod.2010.10.009

38.

Rodriguez

, & Rodrik

(2000). Trade policy and economic growth: A skeptic’s guide to the cross-national evidence. NBER Macroeconomics Annual, 15, 261–325.

39.

Rodrik

(1998). Who needs capital account convertibility? In Fischer

(Ed.), Should the IMF pursue capital account convertibility? Essays in international finance (pp. 55–65). Princeton, NJ: Department of Economics, Princeton University.

40.

Sachs

J. D.

, Warner

, Aslund

, & Fischer

(1995). Economic reform and the process of global integration. Brookings Papers in On Economic Activity, 1995(1), 1–118.

41.

Samimi

, & Jenatabadi

H. S.

(2014). Globalization and economic growth: Empirical evidence on the role of complementarities. PLoS ONE, 9(4), 1–7. doi: 10.1371/journal.pone.0087824

42.

Shehata

E. A. E.

, & Mickaiel

S. K. A.

(2013). SPREGFEXT: Stata module to compute spatial panel fixed effects regression: Lag and Durbin models. Statistical Software Components, S457524. Retrieved from https://ideas.repec.org/c/boc/bocode/s457524.html

43.

Stiglitz

J. E.

(2004). Globalization and growth in emerging markets. Journal of Policy Modeling, 26(4), 465–484. doi: 10.1016/j.jpolmod.2004.04.010

44.

Wikipedia (2016). List of official languages by country and territory. Retrieved from https://en.wikipedia.org/wiki/List_of_official_languages_by_country_and_territory

45.

World Bank, The (2016). World development indicators. Washington, DC.

Initial GDP per capita	–0.792 (0.150)***
Investment	0.174 (0.017)***
Population growth	–0.567 (0.112)***
Education	0.104 (0.306)
Inflation	–0.001 (0.001)
Economic globalisation	0.280 (0.106)***
Political globalisation	0.072 (0.070)
Social globalisation	0.055 (0.132)
Constant	2.991 (1.060)***