Globalization: Between Optimism and Pessimism

Abstract

This article examines the globalization’s effect on the economy using a sample of thirty one developing countries over the period from 1981 to 2013. Globalization is captured by foreign direct investment, openness, and the KOF Globalization Index, measuring globalization along the economic, social, and political dimension. The Westerlund’s (Oxford Bulletin of Economics and Statistics 69:709-748, 2007) cointegration test suggests that the models are cointegrated. The long-run outputs from the autoregressive distributed lag model disclose the following results. First, foreign investment exerts a negative effect on industrialization captured by industry added value and industry employments, whereas trade openness enhances industrialization. Second, the effect of globalization on industrialization is not homogeneous across regions, African industries being those in which the effect is less important. Finally, market size and human capital skill have a negative and significant effect on industrialization, while innovation, infrastructures, and natural resource effects are positive and significant. The industrial effect of globalization is more important in Asia and Latin America than in Africa.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13132-023-01369-8.

Introduction

Since the end of the 1970s, part of the work related to international economics has been classified under two specific labels. One is the new international economics, and the other is the international industrial economics (Rainelli, 1996). These two variances, which cleave with the traditional international economics at least at the level of comparative advantages as explanatory factors of trade flows, come close to the latter on several levels. These theories believe in the virtues of openness, which result is a positive-sum game (Smith, 1776). However, they oppose the establishment of mechanisms likely to hinder the physical and financial flows between nations. At the convergence of all these doctrinal approaches is the disappearance of national borders between countries and the emergence of a universal geographic bloc, which globalization is one of the most eloquent consequences. Globalization, an operation to broaden the intervention space of economic agents, is associated with trade liberalization. It refers to the process of growing interdependence of national economies and the creation of an increasingly integrated world economic space (Thompson, 1999). Its praise, which is always insistent, eclipses its origins. However, it overshadows its distant history. Indeed, globalization is not a new fact. According to Cohen (2004), the current period is only the third act in history initiated in the sixteenth century by the discovery of America which marked the entry of countries into modernity, following monetary and financial integration driven by the gold standard. Since Michalet (1993), it is usual to distinguish three configurations of globalization. The first, international, characterizes globalization mainly through the exchange of goods and services between countries. The second, multinational, identifies it essentially with the mobility of the production of goods and services with the segmentation of the production process Lassurdie-Duchêne (1982). It is then carried by the flow of foreign direct investment (FDI). The last, global, determines globalization by the predominance of financial integration. These configurations, rather cumulative, can be for more than one reason, hope for the world economy. In the long run, international trade promotes economic growth and development (Krugman, 1994). It induces social and political progress and facilitates the integration of countries into the global value chain. Better more, it increases the competitive advantage of nations (Poter, 1990). Related to optimist point of view, the World Bank (WB) (2019) contends that, between 2009 and 2018, globalization has reduced extreme poverty by 35% around the world and has improved industrial productivity and capital movements across the globe. In 1977, there were 11,000 parent companies and 82,000 international subsidiaries in the world. In 2018, their number was estimated at 80,000 and 840,000, respectively. They then produced around $ 25 trillion, representing more than a quarter of global gross domestic product (GDP), while employing around 75 million people (WB, 2019). Also, the free movement of labor and capital has enabled innovations from developed countries (DCs) to be promoted around the world. Industries are one of the main beneficiaries. From this technology transfer, the developing countries (DCs) have benefited (some better than others) from the efforts in terms of research and development (R&D) undertaken in the DCs. This idea, which is now widely accepted today, however, constitutes one of the paradoxes of globalization. This paradox is justified by the difficulty of perceiving day-to-day the benefits of trade openness, even though the transaction costs often appear more visible due to globalization. This supports the pessimist point of view. On this subject, List’s (1827) predictions already highlighted, through his educational protectionism, the unenviable effects of openness to trade as brought by globalization. He then saw in exchange, an instrument of domination of the strongest (through technology, structural, and organizational advantages) over the weakest (because of their economic fragility). Thereby, globalization can be considered as a zero-sum game where some win what others lose. Some figures can be presented to explain List’s conclusion. Indeed, according to the WTO (2018), in 2007, European, American, Asian, and African exports were, respectively, 5806, 2355, 4141, and 436 billion dollars. In 2017, they were 6501, 2960, 6373, and 416 billion dollars. Between the two periods, the share of exports in world exports from these four blocks was 37.38%, 16.75%, 33.06%, and 2.99%, respectively. Moreover, from 1981 to 2016, 28.20%, 10.60%, 19.01%, and 1.48% of FDI inflows were from Europe, Latin America, Asia, and Africa (WB, 2017). These figures were 37.74%, 3.2%, 17.30%, and 0.41% for FDI outflows according to the same source. This reflects the domination of the Europe, Asia, and America on the one hand and presents the marginalization of Africa on the other. The retreat from the virtues of globalization is not current. El Mouhoub (2006) points out that, in the late 1990s, during the period of market euphoria, public debate was already about the “economic horror” of globalization. Also, in the early 2000s, the debate focused attention on relocations to low-wage countries, the destruction of skilled jobs being a corollary. Very close to us, at the beginning of 2020, world economy experienced the disastrous effects of the new coronavirus pandemic. Its mode of contamination (in contact with patients) on the one hand and its speed of propagation (accelerated by the international mobility of people) on the other have led to the closure of land, sea, and air borders in most countries. This state of affairs, highlighting another limitation of globalization, has imposed a restriction on commercial, financial, and human movements. Regarding the above observations, what is the economic effects of globalization? Specially, in developing countries, does globalization benefit for industrialization? The theoretical framework of this question is on the one hand based on classical trade theory developed by Smith (1776) and on another hand related to the factor’s endowment theory. In both cases, since natural resources and raw materials are abundance in developing countries, globalization is expected to enhance industrialization. This effect is not automatically according to empirical study of Markusen and Venables (1999). To answer this question, the effect of globalization can be studied with regard to the externalities it releases, in particular on the environment, economic growth, inequalities, migration, and beyond, industrialization. In this article, we focus on the industrialization’s effect of globalization. Two reasons justify this choice. On the one hand, industrial development constitutes a sustained growth opportunity (Brasseul & Lavrard-Meyer, 2016). Also, it reduces social inequalities (Molitoris & Dribe, 2016) and limits the migratory flows which deprive developing countries of their human resources. On the other hand, industrialization, through employment, improves the well-being of populations. This article, with regard to data and econometric modeling, in the context of certain developing countries, contributes to this debate. Moreover, related to the approach, the article on the one hand examines the sample as a whole and on the second hand studies three specific samples (Africa, Asia, and South America). Doing this, the article contributes to present the area where globalization has the most effect on industrialization. For policy makers, we present the advantage to adapt industrialization process in the context of globalization. The article is built around five parts; this ending introduction being the first. The next two will present, respectively, the received ideas of globalization and the methodological framework of the study, while the last two will present the main results, the conclusion, and some suggestions.

Globalization: Some Received Ideas Regarding Industrialization

More generally, the effects of globalization can be appreciated in terms of economic growth, inequalities or, industrialization. Industrialization is a process by which the secondary sector increases its weight in the economy. It can be observed through the relative importance of the technical capital in the production activity of goods or services. Industrialization is also the increasing share of secondary sector in terms of employment (related to total employment) and GDP (Echaudemaison, 2003). From the above, the transmission channel of globalization on industrialization can be two sets analyzed (Gui-Dybi & Renard, 2015). Directly, FDI can cause industrialization by modifying the input matrix of the supply and use table. This table records intermediate consumption of different industries by product. Indirectly, FDI inflows, by technology transfers in the secondary sector in the host economy, can cause industrialization through the improvement of the productivity of local firms in this sector and other related sectors. Two models can expose the direct effect of FDI inflows on industrialization. One is developed by Rodriguez-Clara (1996) and another by Markusen and Venables (1999). The latter analyzes this impact in terms of the number of enterprises and can be used to analyze the impact on industrialization in terms of GDP or value added. The first can be used to analyze the impact on industrialization in terms of employment, specifically the ratio of employment generated in upstream industries. This can be done through the demand for specialized inputs to the labor force hired directly by the firm. The location of MNCs in the host country can generate two positives effects. They are competition and linkage effect. The competition effect is based on the fact that MNCs compete with domestic firms. It results from the production by MNCs of substitutable goods which can also be imported. The size of this effect increases with the size of the surplus of products present on the market, as compared to the initial supply of products without MNCs. It decreases with the productivity of the local firms (Gui-Dybi & Renard, 2015). Linkage effects are due to the connections with local suppliers. Specifically, if the intensity of usage of local inputs by MNCs is lower compared with that of local firms, the exit of local firms producing final goods will be followed by the closure of domestic firms producing intermediate goods. This will occur because the demand for the latter will decrease. On the contrary, if MNCs use more local inputs than local firms producing the final good, the number of firms producing intermediate goods will increase due to backward linkages. In the case of an increase in the demand for intermediate goods, Markusen and Venables (1999) predict that new domestic firms will be created to satisfy the demand of multinational companies, which will contribute to the reduction of the price of intermediate goods. The decrease in the price of intermediate goods would be beneficial to domestic firms producing final goods because their cost of production would decrease. Other domestic firms in the industry of final goods will be able to breakeven and make non-negative profits through forward linkages. The emergence of these new firms would then be beneficial to other local firms through other rounds of backward and forward linkages. Rodriguez-Clare (1996), meanwhile, analyzes the impact of globalization on the economy in terms of job creation, and its conclusions concur with those of Markusen and Venables (1999) on the necessity for the enterprise to use intensively local inputs toward the objective of creating more local jobs, thus increasing forward and backward linkages. Two key conditions to achieve this objective are analyzed by Rodriguez-Clare (1996). Firstly, the good produced by the MNCs should be highly complex because the production of the final good will require access to a variety of inputs. Secondly, there are high costs of communication between the host and home countries of FDI as they will increase the necessity of using local inputs. Basically, indirect impacts are explained by technology transfers. The technology transfers relate to the way in which an innovation is transmitted from upstream to downstream, from fundamental research to development through the stage of applied research. The technology transfers modalities depend on the position on the market, their size, and the nature of their production. The term of technology transfer was especially developed within the framework of north–south relations. It can increase the productivity, value added, and profit of an enterprise (Gui-Dybi and Renard, 2015). In terms of the analytical framework, an increase in the profit of one local firm will attract more local investors to the activity until the profit of each firm is equal to zero or be in equilibrium. Technological transfers can be very expensive (the acquisition or licensing of a technology). It can be through official development assistance or labor mobility (Fosfuri et al., 2001) and Glass and Saggi (2002). The number of firms and jobs in the manufacturing sector and the volume of manufactured outputs (final and intermediate goods) would increase depending on the magnitude and the strength of backward and forward linkages for upstream and downstream firms, respectively, while horizontal spillovers will depend on the fluidity of the labor market and the capacity to acquire technologies. In particular, on the one hand, upstream local firms, which supply intermediate goods to foreign investors and domestic firms, can have access to foreign technology from the MNCs through the training of its staff, the recruitment of former staff of multinationals, or a direct licensing/acquisition of technology (vertical spillovers). All these factors would contribute to the production of final goods that meet standards set by the headquarters of the MNCs. On the other hand, domestic firms in the multinational’s industry would be able to increase their productivity by purchasing improved inputs from upstream firms, hiring former staff of MNCs, addressing inefficiency issues, or strengthening their research and development activities to copy the multinational’s products or improve their own products by imitating multinationals (Görg & Greenaway, 2005). This situation would also contribute to the development of more competitive domestic firms operating in the industry of the multinationals (horizontal spillovers). Openness can foster productivity as well as the economic performance, and this can lead to more job openings in general. However, some countries and economic actors look suspicious to free international trade, and the issue comes from the statement that international trade can negatively affect employment via transformation of labor to capital related to technology spillover. Indeed, if the structure of production changes from a labor-intensive form to a more capital-intensive form, jobs will be destroyed. In this context, it is widely known that the most open industry to the trade shocks is the manufacturing sector; and therefore, transformation in the structure of production (related to international trade) can negatively affect the jobs in manufacturing in particular (Gozgor, 2016, 2017). Rodrik (1997) illustrated new aspects of the effect of international trade on the labor market. He argued that the globalization measure (trade openness) can significantly affect the labor–demand elasticities. More specifically, Rodrik (1997) indicated that a more elastic labor demand that arises from the international trade yields to a higher volatility in the both real wages and employment, whereas the effect on the productivity volatility will remain constant. In addition, he pointed out that more elastic labor demand leads to a decline in the bargaining power of workers. Furthermore, when one considers the return of labor compared to the returns in other factors of production (capital and land), this process creates an inefficient surplus to the returns of other factor owners. This process finally tends to increase the non-wage costs of the production and again suppress welfare of workers.

Data, Model Specification, and Method

This section presents briefly the methodology framework of the study. To better understand, we expose firstly data before specifying our model.

Data

This article adopts a panel data approach covering 31 developing countries1 over the period 1981–2013, to examine the effect of globalization on industrialization. Meanwhile, industrialization is our dependent variable. Two indicators are concerned: industry added value in percentage of GDP and employment in industry (percentage of total employment). As mentioned earlier, globalization can be captured through three components, namely, goods and service mobility (country’s degree of openness measured by average exports and imports in percentage of GDP), multinational corporation’s activity (measured by FDI inflows in percentage of GDP), and KOF Globalization Index (developed by Gygli et al., 2019). In order to facilitate the comparison among countries, we use data in US dollar terms. Countries and period are chosen related data availability.

Control Variables

Five control variables are concerned: natural resources, infrastructures, human capital, innovation, and market size. The positive linkage between natural resources and industrialization is established by Lee (1981). Related to infrastructures, intuition is coming from the 9th sustainable development objective which makes infrastructures, a significant determinant of industrialization. The next variables are the result of endogenous growth theory (human capital and innovation) developed by Lucas (1988) and Romer (1986). Table 1 gives more description on the whole variables and presents their different sources.

Table 1.

Variables: measurements and sources

	Variables (abbreviation)	Measures	Sources
	Industrialization (Indust 1)	Added value of industrial sector (% of GDP)	WDI2
	Industrialization (Indust 2)	Employment in industrial sector (% of total employment)
Independent variables	Globalization (Global 1)	Foreign direct investment inflow (% of GDP)
	Globalization (Global 2)	Average imports and exports (% of GDP)
	Globalization (Global 3)	Composite index around the economic, social, and political dimension	kof.ethz.ch/globalization
	Market size (Mark)	GDP per capita	WDI
	Natural resources (Natres)	Total natural resources rents (% GDP)
	Infrastructures (Infr)	Composite indicator and related to fixed broadband, Internet subscribers per 100 people, telephone fixed lines per 100 people, mobile cellular subscriptions per 100 people, electric power consumption (kWh per capita)	LEAD3
	Human capital skills (Husk)	Composite indicator and related to literacy rate, adult total (% of people ages 15 and above), enrolment in tertiary education per 100,000 inhabitants mean years of schooling of adults
	Innovation (Innov)	Composite indicator and related to: number of patent grants per 1 million people, number of publications in scientific and technical journals per 1 million people

²https://databank.banquemondiale.org/reports.aspx?source=2&series=NV.IND.TOTL.ZS&country = #

³http://lead.univ-tln.fr/bdd.php

Source: authors

Method

In this article, we postulate that globalization enhances industrialization. To test this hypothesis, we use panel cointegration techniques. It is well-known that most residual-based cointegration tests in panels require that the long-run parameters for the variables in their levels are equal to the short-run parameters for the variables in their differences (Jalil et al., 2016). The standard panel models such as pooled OLS, fixed effects, and random effects models have some serious shortcomings. For instance, pooled OLS is a highly restrictive model since it imposes common intercept and slope coefficients for all cross-sections and thus disregards individual heterogeneity (Jalil et al., 2016). The fixed effects model, on the other hand, assumes that the estimator has common slopes and variance but country-specific intercepts. The parameter estimates produced by the fixed effects model are biased when some regressors are endogenous and correlated with the error term. We use the autoregressive distributed lag (ARDL) model, through which both the short-run and long-run effects can be estimated simultaneously from a data set with large cross-section and time dimensions. In addition, the ARDL model, especially PMG and MG, provides consistent coefficients. Since our data set includes time period which is fairly long, it very likely that the macroeconomic variables will follow a unit root process. The choice of unit root test will strongly depend on cross-section dependence test outcome.

The ARDL model takes the following form in our case:

$$Indust={\mu }_i+\sum _{j=1}^p{\lambda }_{i,j}Indus{t}_{i,t-j}+\sum _{j=0}^q{\delta }_{i,j}{x}_{i,t-j}+{\epsilon }_{i,t}$$ 1

where Indust is the industrialization in i group at t time, μ_i is included for the fixed effects, and x_it is vector of explanatory variables as presented in variable’s table. Equation (1) can be written as first difference form:

$$\mathrm{\Delta }Indus{t}_{i,t}={\mu }_i+{\varphi }_{i}Indus{t}_{i,t-j}+{\phi }_i{x}_{i,t}+\sum _{j=1}^{p-1}{\lambda }_{i,j}^{\ast }\mathrm{\Delta }Indus{t}_{i,t-1}+\sum _{j=0}^{q-1}{\delta }_{i,j}^{\ast }\mathrm{\Delta }{x}_{i,t-j}+{\epsilon }_{i,t}$$ 2

where

$${\varphi }_i=-(1-\sum _{j=1}^p{\lambda }_{i,j}),{\phi }_i=-\sum _{j=0}^p{\delta }_{i,j},{\lambda }_{i,j}^{\ast }=-\sum _{m=j+1}^p{\lambda }_{i,m},{\delta }_{i,j}^{\ast }=-\sum _{m=j+1}^q{\delta }_{i,m}$$ 3

with j = 1, 2, …, p-1

Equation (2) can be written as follows when we group the variables in levels:

$$\mathrm{\Delta }Indus{t}_{i,t}={\mu }_i+{\varphi }_i[Indus{t}_{i,t-j}-{\theta }_i{x}_{i,t}]+\sum _{j=1}^{p-1}{\lambda }_{i,j}^{\ast }\mathrm{\Delta }Indus{t}_{i,t-1}+\sum _{j=0}^{q-1}{\delta }_{i,j}^{\ast }\mathrm{\Delta }{x}_{i,t-j}+{\epsilon }_{i,t}$$ 4

Where:

$${\theta }_i=-\frac{{\phi }_i}{{\varphi }_i}$$ 5

This shows the long-run relationship between the variables involved and ϕ_i is the speed of adjustment toward the long-run equilibrium. On the one hand, our sample is estimated as a whole. Also, the sample is split into three sub-samples (Africa, Asia, and Latin America). This approach is helpful to investigate if the effect of globalization on industrialization varies across these regions. Lastly, some robustness analyses are performed using a common correlated effects model. This approach is useful in the case of non-stationary, heterogeneity, cross-sectional dependence, and unobserved common factors that can explain industrialization but are not included in the model.

Results

First of all, we will present here preliminary results such as Pesaran cross-section dependence (Pesaran CD) test and Pesaran cross-sectional dependence unit root (Pesaran CDIPS) test. Then, will follow summary statistics and baseline estimations. Finally, we will present short-run estimates and discuss the long-run relationship between globalization and industrialization.

Preliminary Results

The Pesaran CD test is performed on the sample. The results, recorded in Table 2, reject the null hypothesis.

Table 2.

Pesaran cross-section dependence test

H0: no cross-section dependence. H1: cross-section dependence (reject H0 if P value significant)
	Statistics	P value	Observations
Pesaran CD test	4.167	0.0000	Cross-section dependence
Absolute value	0.248

Source: authors

These results lead us to conclude that variables behave with cross-section dependence. It then justifies the choice of the second generation of unit root test, that is, the Pesaran cross-section dependence CDIPS test. Though testing for the order of integration of variables is not important when applying the ARDL model as long as the variables of interest are I(0) and I(1), we carry this test just to make sure that no series exceeds I(1) order of integration. The results in Table 3 show that the variables have mixed orders of integration and none is beyond I(1). This justifies the use of the ARDL framework.

Table 3.

Unit root test (CIPS)

Variables	Lags						Decision
	K = 1		K = 2		K = 3
			Values
	Empirical	Critical	Empirical	Critical	Empirical	Critical
Whole sample
Indust 1	− 3.510	− 2.3	− 3.611	− 2.3	− 3.558	− 2.3	I(0)
Indust 2	− 2.864	− 2.3	− 2.985	− 2.3	− 2.040	− 2.3	I(0)
Global 1	− 2.440	− 2.3	− 2.516	− 2.3	− 2.491	− 2.3	I(0)
Global 2	− 2.664	− 2.23	− 2.753	− 2.23	− 2.802	− 2.23	I(0)
Global 3	− 2.550	− 2.3	− 2.435	− 2.3	− 2.491	− 2.3	I(0)
Mark	− 1.339	− 2.23	− 1.524	− 2.23	− 1.324	− 2.23	I(1)
Natres	− 2.365	− 2.3	− 3.180	− 2.3	− 3.387	− 2.3	I(0)
Infr	− 1.252	− 2.23	− 1.898	− 2.23	− 1.296	− 2.23	I(1)
Husk	− 1.231	− 1.23	− 1.231	− 2.23	− 1.231	− 2.23	I(1)
Innov	− 2.367	− 2.23	− 2.401	− 2.23	− 2.367	− 2.23	I(0)

Source: authors

The cointegration test to examine the long‐run relationship between the variables was performed using the panel cointegration test (Westerlund, 2007) reported in Table 4. It has the advantage of allowing heterogeneity of parameters. Westerlund’s test (2007) is actually made up of four tests: G_t, G_a, P_t, and P_a. The first two tests are called group means tests, and the alternative hypothesis is that at least one case has cointegrated variables. The latter two are called panel tests, and in this case, the alternative hypothesis is that the panel, considered as a whole, is cointegrated. The hypothesis of non-cointegration is rejected (Table 4) at a threshold of 1%, except for the G_a statistic, depending on whether the delay introduced varies between 1 and 3. In the light of these results, we conclude that our variables are integrated.

Table 4.

Westerlund cointegration test

Calculating Westerlund ECM panel cointegration tests
Results for H0: no cointegration
Lag	Test	Value	Z value	P value
1	Gt	2.670	− 6.043	0.000
	Ga	− 6.714	0.478	0.684
	Pt	− 12.282	− 3.497	0.000
	Pa	8.310	− 5.590	0.000
2	Gt	− 2.434	− 4.442	0.000
	Ga	− 6.464	0.758	0.776
	Pt	− 12.282	− 3.497	0.000
	Pa	− 8.310	− 5.590	0.000
3	Gt	− 2.345	− 3.842	0.000
	Ga	− 6.194	1.060	0.855
	Pt	− 11.241	− 2.449	0.007
	Pa	− 7.756	− 4.831	0.000

Source: authors

Summary Statistics

Summary statistics provided in Table 5 (panel A) present the market size as the most volatile variable of our model since its standard deviation is more than 6000. This figure shows the high fluctuation of macroeconomic performance of sample’s countries. Moreover, innovation, infrastructure, human capital skill, and globalization (FDI) are the most stable variables of the model with their respective standard deviation equal to 0.0700, 0.1382, 0.1633, and 1.1704. But, the mean values of innovation (0.018), infrastructures (0.43), and human capital skill (0.63) capture the attention. In fact, these figures are far lower than those of developed countries such as France, the UK, and the USA which average value is around 0.3, 0.6, and 0.8, respectively.

Table 5.

Summary statistics and correlation test

Source: authors

Predicted Scatter Graphs

Predicted relationship between variables are presented in Fig. 1. In fact, the named figure presents a positive tendency between industrialization and globalization (when measured by openness), and negative correlation between the two variables when FDI is a proxy of globalization. More variables will be added in the model to estimate long-run relationship.

Fig. 1.

Globalization and industrialization.

Source: authors

Relationship Between Industrialization and Globalization

As mentioned in Tables 6 and 7, the named relationship can be presented in short run as well as in long run. Even if we will emphasize on long-run relationship, it is important to make two observations on short-run estimates. Frist, ECT coefficients are all negative and significant. One can then conclude that there exists a short-run relationship between globalization and industrialization. Second, in short run, globalization has a negative effect on industrialization. The result emphasizes once on the pessimism perception some make on globalization effect.

Table 6.

Short-run estimation

	Industrialization 1		Industrialization 2
	Pool mean group		Pool mean group
	(1)	(2)	(3)	(4)
D.Global1	− 0.9373***		− 0.9876***
	(0.5584)		(0.7557)
D.Global2		− 0.2284***		− 0.01651*
		(0.6084)		(0.0001)
D.Mark	− 0.0003***	− 0.6593*	− 0.2020***	− 0.0003*
	(0.0001)	(4.7281)	(0.0054)	(0.0760)
D.Natres	− 1.7286	− 0.0025***	0.4567***	0.6124*
	(0.3780)	(0.0001)	(0.3057)	(0.6100)
D.Innov	4.0336*	− 1.7416	0.6423*	0.6593*
	(1.7847)	(0.3980)	(3.2781)	(4.7281)
D.Infr	− 1.4172***	0.336*	23.6797***	0.4336*
	(0.9167)	(0.0134)	(5.5678)	(0.1234)
D.Husk	2.2575***	− 1.7812***	− 1.5413	3.8961
	(1.9446)	(3.9167)	(6.7890)	(3.7254)s
ECT	− 0.2044***	− 5.2575***	− 0.3509***	− 0.0079***
	(0.1362)	(4.9446)	(0.1287)	(0.0077)

*** 1% signification level, ** 5% signification level, * 10% signification level

Source: authors

Table 7.

Long-run estimation

	Industrialization 1				Industrialization 2
	Pool mean group				Pool mean group
	(1)	(2)	(5)	(3)	(4)	(6)
Global1	− 0.2851***			− 0.2929***
	(0.0280)			(0.0757)
Global2		0.0712**			0.0700***
		(0.0318)			(0.0426)
Global3			− 0.0377*			− 0.0176*
			(0.1543)			(0.2206)
Mark	− 0.0008***	− 0.0005***	− 0.0040***	− 0.0022***	− 0.0024***	− 0.0061***
	(0.0001)	(0.0001)	(0.0055)	(0.0003)	(0.0003)	(0.0288)
Natres	0.3528***	0.6672***	0.8219***	0.6456***	0.7446***	− 0.0344
	(0.0471)	(0.0667)	(0.0607)	(0.1057)	(0.1407)	(0.1288)
Innov	0.4107***	0.3097***	11.6112***	0.5693*	0.8103**	9.5564***
	(0.7477)	(0.5966)	(4.131606)	(4.2781)	(0.4663)	(3.2581)
Infr	1.8320***	19.5150***	29.0757***	37.6797***	40.2431***	7.35204***
	(3.9167)	(4.8332)	(5.5784)	(5.6220)	(7.0060)	(1.5921)
Husk	− 20.7287***	− 36.2815***	− 0.1313	− 1.7867	− 1.2296	74.3520
	(4.9446)	(5.8395)	(8.0377)	(7.4151)	(7.8490)	(13.5921)
Observations	1023	1023	1023	1023	1023	1023
Countries	31	31	31	31	31	31

*** 1% signification level, ** 5% signification level, * 10% signification level

Source: authors

Regarding long-run relationship results recorded in Table 7, two main remarks can be made. On the one hand, except for human capital skill and innovation coefficients in some cases (panels 3, 4. and 5), all coefficients are significant. On the other, some have expected signs (globalization in panels 2 and 4, natural resources, innovation, and infrastructures), and others do not (globalization in panels 1, 3, 5, and 6, market size, and human capital skill in panels 1 and 2).

The coefficient of globalization (captured by FDI) in panel 1 is − 0.2851. It means that, everything being equal, one point increase in FDI reduces industrialization up to 0.2851 point. This remark holds in panel 3 where employment in the industry is a proxy of industrialization and the coefficient of globalization, equal to − 0.2929. The same observation is made in the case of globalization captured by KOF Index. KOF Globalization Index goes belong the classical economic dimension of globalization and incorporate two other aspects which are very important. They are political and social dimension. These results differ from those of Rodriguez-Clare (1996) and Markusen and Venables (1999) for whom FDI directly improves industrialization trough industrial employment and the production of goods and services with high added value. They also differ from those of Haskel et al. (2007) for whom globalization indirectly strengthens the link with technology transfer which, affects industrial production. Some elements can explain these negative signs. The first is related to the FDI advantages for recipient countries, notably technology transfer. In a company, though technology used to be complementary for high-skilled jobs, the substitution of low-skilled jobs by machines can then be common. Accordingly, one can observe the reduction of low-skilled jobs employed in industry and globally the decline of total employment in industrial sector. In most cases, with task automation, it becomes possible to produce more with fewer employees. The predominance of machines in a sector then leads to staff dismissals. Our result concords with the findings of OECD (2018) who reports that the UK and France destroyed manufacturing jobs between 2007 and 2015 due to globalization. Moreover, the fall in manufacturing employment was faster in the UK (− 34%) than in France (− 20%) between 2000 and 2010 due to the adoption of new technologies by multinational firms. Also, in the case of France, Harrigan et al. (2016) show for the period 1994–2007 that automated tasks involved by technology in French companies have strongly reduced low-skilled jobs. The second explanation can be the nature of connection (horizontal or vertical) between local and foreign firms. In the case of a horizontal connection, no advantage is possible (Aitken & Harrison, 1999). At least two reasons exist. On the one hand, this linkage creates a difference in demand specificities of each type of firm and therefore the offers of each company. It is the case when local firms produce low-quality goods for the domestic market, while multinationals invest in the production of high-quality goods for the foreign market. On the other, based on the work of Liang (2016), multinationals can develop mechanisms that prevent knowledge transfer between them and local firms. Contrary, positive spillovers are more likely to occur if the link between these companies is vertical. Regarding the above explanation, the negative effect of FDI on industrialization can then be justified by the fact that foreign and local firm’s supplies are closed. Thirdly, one can explain these coefficients by the sectorial distribution of FDI in developing countries. Foreign investments are useful for industrialization if a significant part of these investments is directed toward the industrial sector. However, we note with the OECD (2018) that foreign investments in developing countries are mostly oriented toward the primary sector. This highlights the marginalization of the industrial sector regarding FDI. When globalization is captured by openness, one can observe that it enhances industrialization regardless of the proxy (0.0712 and 0.0700 for models 2 and 4, respectively). These figures show the advantages of exports on industrialization. Indeed, exports involve for concerned industry more production and factor demand such as labor. Those industries will then improve their contribution to the global added value: it is called industrialization by export strategy. This model has been gradually developed by a few Asian countries and has enjoyed worldwide success since the end of the 1980s. Moreover, for Balassa (1978), export-oriented policies result in a better industrialization outcome than policies that favor import substitution. This is because export policies, which similarly stimulate sales in national and international markets, lead to an allocation of resources according to comparative advantage. These policies allow greater use of production capacity as well as economies of scale. One can then observe the increase in job opportunities provided by industries. This ambiguous effect of globalization is documented by Bourguignon (2012). Indeed, it shows on the one hand that globalization has been a vector of growth for emerging economies through technology transfer and industrialization. On the other, it has been a source of inequalities within countries and a lever for social tensions and injustices.

For all panels, the unexpected negative signs for market size and human capital skill appear. The results can be justified in two ways. On the one hand, income would not be large enough to enable households to acquire manufactured products. Therefore, weak local demand limits in this case opportunities for industrial expansion. But one can note that its coefficient is nearly zero for all the models. On the other, human capital in terms of quality matter or is not at the necessary level to impact positively industrialization. This result is different from that of Morgan et al. (2013). According to these authors, England plays a key role during the industrial revolution because of the quality of its workforce. In our case, the negative sign may be the fact of the withdrawal of developing economies from the technological knowledge frontier.

The positive and significant coefficients of natural resources, innovation, and infrastructures denote the beneficial effect of those variables on industrialization. The positive relationship between natural resources and industrialization is not new in the economic literature. The first formulation, under the label of “resource-based industrialization,” is to our knowledge the work of Roemer (1979). This relationship may reflect the direction taken by the countries regarding their industrialization strategy. If industrialization by export substitution requires the transformation of primary products into manufactured products, then it is obvious in this case that natural resources constitute a vector of industrialization since they are considered as raw materials for those industries. In relation to industrialization, the role of infrastructures can be analyzed in the light of its components. The location of new companies in a country constitutes an opportunity for industrialization, and this depends on many variables among which infrastructures. Indeed, when they are related to transport, infrastructures facilitate the movement of goods and people while bringing production closer to markets. When they are related to energy, they increase the productive capacities of local and foreign companies. When linked to telecommunications, they facilitate the movement of knowledge and finance for production. In either case, they appear as an important determinant of industrialization. The same comment can be done regarding the positive contribution of innovation to industrialization. In fact, innovation can be defined as the implementation by a company and, for the first time, of a new or significantly improved product (good or service), a new production process, or a new organizational method. This highlights the mean variances of innovation (production, organizational, and process). Each of them or the whole improve industrial productivity (industry value added) as well they stimulate industrial production and therefore increase the share of industrial employment.

Differential Effect

By dividing our sample into three sub-samples (Africa, Asia, and Latin America), regarding the effect of globalization on industrialization, two observations emerge (Table 8). On the one hand, trade openness always has a positive effect on both employment and industrial added value. However, the globalization effect is differentiated by region. One point increase in openness, all other things being equal raises industrial added value up to 0.024, 0.183, and 0.5096 point and industrial employment up to 0.048, 0.3777, and 0.2163 point, respectively, for Africa, Asia, and Latin America. From these figures, we can conclude that industries benefit unequally from globalization. These effects depend on the regions; Africa is the continent where the effect of trade openness is less important. This differentiated effect may be industrialization strategies that vary from one continent to another. This can also be the fact of trade integration. The intra-African trade represented, for example, in 2019 only 17.69% of total trade of the continent (African Development Bank, 2019). This is under those of ASEAN (30%) and MERCOSUR 20% (UNCTAD, 2019). Finally, since African exports are based on commodities and raw materials, this can limit its industrial potential. Maybe Africa lacks appropriate technology to transform itself the raw materials in manufactured goods. Indeed, industrialization is a process strongly supported by technology, which innovation is one of its main components. However, we can see that its coefficient remains positive and statistically significant for Asia and Latin America even though it is not for Africa. This withdrawal of Africa from the technological frontier can be an obstacle for his industrialization in a context of globalization. Another result seems interesting to us. It is the coefficients of natural resources. While globalization induces industrialization, regional estimates show in the case of Africa a negative effect of natural resources on the industrialization. Indeed, the coefficients of models (1), (2), (3), and (4) are all negative and significant (− 0.5599, − 1.5903, − 0.9700, and − 0.0534). This result follows that of Avom and Carmignani (2010). Several mechanisms can explain this, qualified by many authors as a “raw material curse.” The first is from Sachs and Warner (2001) and Gylfason (2001). They particularly emphasize the importance of crowding out effects associated with commodities’ dependence. In addition to natural capital which crowds out other forms of capital, particularly physical and human, two other explanations can synthesize these effects. One is Dutch disease. It indicates that the abundance of natural resources contributes to the appreciation of the real exchange rate and reduces the competitiveness of the manufacturing sector. Another is the domestic wealth effect or expenditure effect. It increases in domestic demand and prices. We then witness a loss in competitiveness of the domestic manufacturing sector. Ultimately, the export concentration around natural resources increases where industrialization matters. The second mechanism explaining the negative effect of natural resources on industrialization is identified by Leite and Weidmann (1999) and Ross (2001). They show that the abundance of basic commodities promotes corruption and bad governance which are directly two main factors of weak economic growth. The third mechanism is analyzed by Deaton and Miller (1995). According to them, commodities are characterized by price instability in international markets. Therefore, countries that depend strongly on commodity exports are more vulnerable to external shocks and can no longer support the industrialization process.

Table 8.

Long-run estimation (Africa–Asia–Latin America)

Africa
	Industrialization 1		Industrialization 2
	Pool mean group		Pool mean group
	(1)	(2)	(3)	(4)
Global1	− 0.9869***		− 0.8824***
	(0.1750)		(0.2661)
Global2		0.0249***		0.0480***
		(0.0025)		(0.0150)
Mark	− 0.0050***	− 0.0017***	− 0.0001***	− 0.0023***
	(0.0005)	(0.0003)	(0.0005)	(0.0048)
Natres	− 0.5599***	− 1.5903***	− 0.0970**	− 0.0534**
	(0.0863)	(0.1800)	(0.0150)	(0.0090)
Innov	0.1497	0.1469	0.1034	0.1045
	(4.6267)	(6.5568)	(5.7263)	(1.0845)
Infr	1.9349	1.0283	5.1613**	3.503**
	(6.1911)	(5.9796)	(1.1022)	(1.3403)
Husk	− 1.8294*	− 8.0637*	− 30.3716	− 12.8454
	(0.2001)	(3.0085)	(16.2402)	(15.4521)
Asia
	Industrialization 1		Industrialization 2
	Pool mean group		Pool mean group
	(1)	(2)	(3)	(4)
Global1	− 0.1177		− 1.0549
	(1.2902)		(1.2110)
Global2		0.1815**		0.3777**
		(0.02593)		(0.1223)
Mark	0.0025**	0.0003***	0.00024**	0.0004**
	(0.0012)	(0.00031)	(0.0010)	(0.0001)
Natres	0.4456*	0.0545*	1.4930**	1.0409**
	(0.1308)	(0.01686)	(0.5252)	(0.3571)
Innov	1.5411*	1.7476**	1.3475*	16.1308***
	(0.2305)	(0.5763)	(0.4313)	(8.6113)
Infr	2.2712*	5.8992	9.8881**	14.2929**
	(0.1762)	(1.4104)	(3.9778)	(4.25291)
Husk	42.3023*	16.7505**	3.1228**	7.0618***
	(20.8950)	(8.0144)	(0.3616)	(0.4283)
Latin America
	Industrialization 1		Industrialization 2
	Pool mean group		Pool mean group
	(1)	(2)	(3)	(4)
Global1	− 0.4597		− 0.3000
	(0.4796)		(0.5504)
Global2		0.5069**		0.2163***
		(0.0493)		(0.0168)
Mark	0.0003*	0.0008***	0.0010***	0.0010***
	(0.0001)	0.0001	(0.0001)	(0.0001)
Natres	0.6962***	0.8437***	0.1022**	0.0476**
	(0.0687)	0.0809	(0.0472)	(0.0014)
Innov	0.2749**	0.1160***	0.2738	0.9694**
	(0.0150)	0.0021	(0.0071)	(0.0236)
Infr	0.7464***	6.1398***	7.0556**	15.6897***
	(0.0225)	0.563	(2.3001)	(4.1318)
Husk	− 3.9228***	− 4.3527***	2.0311	7.5943
	(0.3015)	0.0660	(3.4533)	(6.8023)

*** 1% signification level, ** 5% signification level, * 10% signification level

Source: authors

When analyzing the market size, we observe that in Asia and Latin America, it has an opposite effect of that found in Africa. In fact, market size determines positively and significantly industrialization in the Asia and Latin America. Since market size is captured by GDP per capita, this situation can be linked to the purchasing power of households which enhances domestic demand and then creates opportunities for producing manufacturing goods. The same remark on infrastructures is feasible. In this case, Africa remains the continent with a low effect of this variable on industrialization. This makes relevant to Kumar (2001) results. For this author, infrastructures are one on the main determinant of country’s attractivity. When its quality is questionable, it become an obstacle for industrialization.

Sensibility Check

To assess the robustness of our estimates, we estimated our model by using common correlated effect estimate (Table 9). This has the advantage that it captures the unobserved common factors (such as trade agreements among countries, financial crises, and international policies related to openness) which is absent in the model can significantly affect our estimate. Globalization is measured by our three variables. Reading the results, two mains remarks can be made. On the one hand, globally, the signs of different coefficients remain the same. On the other hand, the coefficients of human capital skills become significantly positive (panels 3, 5, and 6). In these cases, human capital skills improve industrialization since it can be determinant for the attractiveness of the country.

Table 9.

Sensibility check with common correlated effect

	Industrialization 1				Industrialization 2
Common correlated effects
	(1)	(2)	(5)	(3)	(4)	(6)
Global1	− 0.3133*			− 0.2963***
	(0.1766)			(0.1867)
Global2		0.1243***			0.1173***
		(0.1806)			(0.1908)
Global3			0.0852***			0.1056***
			(0.0002)			(0.0401)
Mark	− 0.0005***	− 0.0004***	− 0.0002***	− 0.0005***	− 0.0039***	− 0.0005***
	(0.0001)	(0.0007)	(0.0001)	(0.0008)	(0.0007)	(0.0002)
Natres	0.3129***	0.3062***	0.4856***	0.2875***	0.2755***	0.7475***
	(0.0352)	(0.3058)	(0.0317)	(0.3543)	(0.3398)	(0.2107)
Innov	0.0700*	0.9057***	0.4356***	0.0867*	0.4439**	0.8258**
	(0.4503)	(0.1596)	(0.7325)	(0.9930)	(0.7129)	(0.3896)
Infr	1.9304***	1.8315***	1.8420***	0.9999**	0.9329***	0.2635***
	(0.030)	(0.0962)	(3.9167)	(0.0950)	(0.1876)	(0.0520)
Husk	− 22.534***	− 25.7538***	0.0265*	0.8245*	− 0.6393	1.0026*
	(1.6623)	(4.0675)	(0.0025)	(0.2117)	(0.0427)	(0.4580)
Observations	1023	1023	0.72	1023	1023	0.61
Countries	31	31	1023	31	31	1023

*** 1% signification level, ** 5% signification level, * 10% signification level

Source: authors

Conclusion

According to Maddison (1995), the world growth rate has been very rapid over time, world trade having grown much faster than world production. World population and product per capita have followed this trend. Also, we should be noting that this development occurred disproportionately between the regions. Meaning that some benefit more than the other. Accordingly, should we trust in globalization? This study (1981–2013) then concerns a panel of thirty one developing countries. Eight (8) are from Africa, then ten (10) from Asia, and thirteen (13) from Latin America. The main objective is to analyze the economic effect of globalization precisely and then its effect on industrialization. At the methodology approach, we firstly applied the Pesaran cross-sectional dependence test to choose appropriate methodology. In fact, the rejection of the null hypothesis of cross-sectional independence, which can be justified by the presence of UCFs in the error terms, leads us to adopt the Pesaran’s (2007) CIPS unit root test (which is a second generation of unit root test). The results reveal mixed integration; therefore, Westerlund’s (2007) cointegration test was used to determine the presence of a long‐run relationship between dependent and independent variables. Since some variables were stationary at level and other at first difference, ARDL estimates have been performed for this purpose. The baseline estimation performed through fixed effect estimates has shown that the effect of globalization depends whether it is captured by FDI or openness. In the first case, the effect is negatively significant on the industry value added and industry employment in percentage of total employment which are the two measurements of industrialization. In the second case, the effect is positively significant on industrialization. These results remain the same when the model is enriched with control variables. For this purpose, two suggestions can be made. First, it would be profitable for FDI attractiveness policies to be oriented toward the manufacturing sector. Such an orientation will have twin effects: (i) stimulate the transformation of basic products and (ii) increase industrial production, which will positively influence both value added and industrial employment. Second, import substitution policy can be implemented. In this case, countries can transform their raw materials and produce manufactured goods which previously were imported. Doing that, governments create investment opportunities for those goods. Then, the related domestic production will stimulate industrialization though foreign capital inflow oriented on manufacturing firms. Moreover, by dividing our sample into three sub-samples to test the globalization regional effect, results lead to the conclusion that globalization benefits unequally to the industries. African industries being those in which the effect is less important. Globally, on the overall sample, market size and human capital skill have a negative and significant effect on industrialization, while innovation, infrastructures, and natural resource effects are positive and significant. We then suggest to improve human capital through training since labor efficiency determines the industry competitiveness. Moreover, market size can play an important rule on industrialization if the per capita income defined as a purchasing power of money allow residents to acquire domestic manufactured goods.

Supplementary Information

Below is the link to the electronic supplementary material.

Data Availability

All data generated or analyzed during this study are included in this published article and its supplementary information files.

Code Availability

Not applicable.

Declarations

Conflict of Interest

The authors declare no competing interests.